WEBVTT Kind: captions Language: en 00:00:07.420 --> 00:00:11.440 align:start position:0% and<00:00:08.500> for<00:00:09.500> secure<00:00:09.889> decentralized<00:00:10.789> optimum 00:00:11.440 --> 00:00:11.450 align:start position:0% and for secure decentralized optimum 00:00:11.450 --> 00:00:19.360 align:start position:0% and for secure decentralized optimum optimization<00:00:16.480> great<00:00:17.480> so<00:00:18.230> if<00:00:18.350> you're<00:00:18.530> debating 00:00:19.360 --> 00:00:19.370 align:start position:0% optimization great so if you're debating 00:00:19.370 --> 00:00:20.980 align:start position:0% optimization great so if you're debating whether<00:00:19.730> to<00:00:19.850> be<00:00:20.000> in<00:00:20.090> this<00:00:20.240> room<00:00:20.450> or<00:00:20.690> up<00:00:20.780> in<00:00:20.930> the 00:00:20.980 --> 00:00:20.990 align:start position:0% whether to be in this room or up in the 00:00:20.990 --> 00:00:21.610 align:start position:0% whether to be in this room or up in the main<00:00:21.110> hall 00:00:21.610 --> 00:00:21.620 align:start position:0% main hall 00:00:21.620 --> 00:00:24.310 align:start position:0% main hall we<00:00:22.040> can<00:00:22.190> cut<00:00:22.340> to<00:00:22.370> the<00:00:22.520> chase<00:00:22.940> the<00:00:23.420> topic<00:00:24.020> of<00:00:24.170> my 00:00:24.310 --> 00:00:24.320 align:start position:0% we can cut to the chase the topic of my 00:00:24.320 --> 00:00:26.890 align:start position:0% we can cut to the chase the topic of my talk<00:00:25.100> will<00:00:25.370> be<00:00:25.490> discussing<00:00:26.270> how<00:00:26.540> to<00:00:26.600> solve 00:00:26.890 --> 00:00:26.900 align:start position:0% talk will be discussing how to solve 00:00:26.900 --> 00:00:29.740 align:start position:0% talk will be discussing how to solve hard<00:00:27.290> math<00:00:27.530> problems<00:00:28.040> off<00:00:28.250> chain<00:00:28.610> how<00:00:29.450> can<00:00:29.660> we 00:00:29.740 --> 00:00:29.750 align:start position:0% hard math problems off chain how can we 00:00:29.750 --> 00:00:32.409 align:start position:0% hard math problems off chain how can we actually<00:00:29.900> address<00:00:30.910> problems<00:00:31.910> in<00:00:32.059> machine 00:00:32.409 --> 00:00:32.419 align:start position:0% actually address problems in machine 00:00:32.419 --> 00:00:34.569 align:start position:0% actually address problems in machine learning<00:00:32.570> market<00:00:33.230> clearing<00:00:33.579> operation 00:00:34.569 --> 00:00:34.579 align:start position:0% learning market clearing operation 00:00:34.579 --> 00:00:36.880 align:start position:0% learning market clearing operation scheduling<00:00:35.060> how<00:00:35.750> can<00:00:35.870> we<00:00:35.930> use<00:00:36.140> smart 00:00:36.880 --> 00:00:36.890 align:start position:0% scheduling how can we use smart 00:00:36.890 --> 00:00:38.410 align:start position:0% scheduling how can we use smart contracts<00:00:37.460> to<00:00:37.550> be<00:00:37.640> able<00:00:37.730> to<00:00:37.880> bring<00:00:38.150> those 00:00:38.410 --> 00:00:38.420 align:start position:0% contracts to be able to bring those 00:00:38.420 --> 00:00:40.270 align:start position:0% contracts to be able to bring those problems<00:00:38.930> in<00:00:39.079> consensus<00:00:39.739> on<00:00:39.920> a<00:00:39.950> global 00:00:40.270 --> 00:00:40.280 align:start position:0% problems in consensus on a global 00:00:40.280 --> 00:00:42.550 align:start position:0% problems in consensus on a global optimum<00:00:40.790> and<00:00:40.970> then<00:00:41.840> how<00:00:42.140> can<00:00:42.200> we<00:00:42.350> guarantee 00:00:42.550 --> 00:00:42.560 align:start position:0% optimum and then how can we guarantee 00:00:42.560 --> 00:00:45.520 align:start position:0% optimum and then how can we guarantee security<00:00:43.489> feasibility<00:00:44.420> and<00:00:44.690> optimality<00:00:45.140> of 00:00:45.520 --> 00:00:45.530 align:start position:0% security feasibility and optimality of 00:00:45.530 --> 00:00:49.720 align:start position:0% security feasibility and optimality of this<00:00:45.829> type<00:00:46.100> of<00:00:46.130> problem<00:00:47.560> the<00:00:48.560> general<00:00:49.310> outline 00:00:49.720 --> 00:00:49.730 align:start position:0% this type of problem the general outline 00:00:49.730 --> 00:00:51.880 align:start position:0% this type of problem the general outline I'm<00:00:49.970> gonna<00:00:50.149> be<00:00:50.270> talking<00:00:50.630> about<00:00:50.890> optimization 00:00:51.880 --> 00:00:51.890 align:start position:0% I'm gonna be talking about optimization 00:00:51.890 --> 00:00:54.069 align:start position:0% I'm gonna be talking about optimization problems<00:00:52.370> and<00:00:52.610> I<00:00:52.670> don't<00:00:52.730> mean<00:00:53.180> compiler 00:00:54.069 --> 00:00:54.079 align:start position:0% problems and I don't mean compiler 00:00:54.079 --> 00:00:56.530 align:start position:0% problems and I don't mean compiler optimization<00:00:54.710> here<00:00:54.970> but<00:00:55.970> a<00:00:56.000> couple<00:00:56.329> of 00:00:56.530 --> 00:00:56.540 align:start position:0% optimization here but a couple of 00:00:56.540 --> 00:00:58.090 align:start position:0% optimization here but a couple of different<00:00:56.870> classes<00:00:57.050> of<00:00:57.440> problems<00:00:57.649> that<00:00:58.040> I 00:00:58.090 --> 00:00:58.100 align:start position:0% different classes of problems that I 00:00:58.100 --> 00:01:00.850 align:start position:0% different classes of problems that I work<00:00:58.520> on<00:00:58.700> I'm<00:00:59.000> a<00:00:59.059> PhD<00:00:59.660> student<00:00:59.899> at<00:01:00.230> UC<00:01:00.500> Berkeley 00:01:00.850 --> 00:01:00.860 align:start position:0% work on I'm a PhD student at UC Berkeley 00:01:00.860 --> 00:01:03.880 align:start position:0% work on I'm a PhD student at UC Berkeley and<00:01:01.070> then<00:01:01.940> talk<00:01:02.329> about<00:01:02.360> how<00:01:02.899> we<00:01:02.960> can<00:01:03.320> integrate 00:01:03.880 --> 00:01:03.890 align:start position:0% and then talk about how we can integrate 00:01:03.890 --> 00:01:06.280 align:start position:0% and then talk about how we can integrate smart<00:01:04.399> contracts<00:01:04.969> to<00:01:05.119> really<00:01:05.420> improve<00:01:05.840> the 00:01:06.280 --> 00:01:06.290 align:start position:0% smart contracts to really improve the 00:01:06.290 --> 00:01:08.080 align:start position:0% smart contracts to really improve the application<00:01:06.890> of<00:01:07.159> these<00:01:07.490> types<00:01:07.970> of 00:01:08.080 --> 00:01:08.090 align:start position:0% application of these types of 00:01:08.090 --> 00:01:09.880 align:start position:0% application of these types of optimization<00:01:08.720> tools<00:01:08.960> to<00:01:09.200> a<00:01:09.229> wider<00:01:09.500> set<00:01:09.860> of 00:01:09.880 --> 00:01:09.890 align:start position:0% optimization tools to a wider set of 00:01:09.890 --> 00:01:10.450 align:start position:0% optimization tools to a wider set of problems 00:01:10.450 --> 00:01:10.460 align:start position:0% problems 00:01:10.460 --> 00:01:14.289 align:start position:0% problems I<00:01:11.350> come<00:01:12.350> from<00:01:12.470> a<00:01:12.650> energy<00:01:13.280> background<00:01:13.880> energy 00:01:14.289 --> 00:01:14.299 align:start position:0% I come from a energy background energy 00:01:14.299 --> 00:01:16.060 align:start position:0% I come from a energy background energy markets<00:01:14.720> and<00:01:15.020> the<00:01:15.350> electricity<00:01:15.950> market 00:01:16.060 --> 00:01:16.070 align:start position:0% markets and the electricity market 00:01:16.070 --> 00:01:18.460 align:start position:0% markets and the electricity market specifically<00:01:16.700> so<00:01:17.270> I'm<00:01:17.690> gonna<00:01:17.960> be<00:01:18.080> discussing 00:01:18.460 --> 00:01:18.470 align:start position:0% specifically so I'm gonna be discussing 00:01:18.470 --> 00:01:20.800 align:start position:0% specifically so I'm gonna be discussing an<00:01:18.650> application<00:01:19.190> of<00:01:19.310> this<00:01:19.690> that<00:01:20.690> I've 00:01:20.800 --> 00:01:20.810 align:start position:0% an application of this that I've 00:01:20.810 --> 00:01:22.810 align:start position:0% an application of this that I've researched<00:01:21.140> in<00:01:21.560> electricity<00:01:22.520> market 00:01:22.810 --> 00:01:22.820 align:start position:0% researched in electricity market 00:01:22.820 --> 00:01:26.789 align:start position:0% researched in electricity market formulation<00:01:23.390> for<00:01:23.869> small<00:01:24.260> micro<00:01:24.619> grids<00:01:25.720> so 00:01:26.789 --> 00:01:26.799 align:start position:0% formulation for small micro grids so 00:01:26.799 --> 00:01:28.899 align:start position:0% formulation for small micro grids so optimization<00:01:27.799> just<00:01:28.250> so<00:01:28.520> that<00:01:28.640> we're<00:01:28.760> all<00:01:28.790> on 00:01:28.899 --> 00:01:28.909 align:start position:0% optimization just so that we're all on 00:01:28.909 --> 00:01:31.390 align:start position:0% optimization just so that we're all on the<00:01:29.180> same<00:01:29.330> page<00:01:29.600> who<00:01:30.380> here<00:01:30.680> has<00:01:30.979> heard<00:01:31.220> of 00:01:31.390 --> 00:01:31.400 align:start position:0% the same page who here has heard of 00:01:31.400 --> 00:01:34.780 align:start position:0% the same page who here has heard of convex<00:01:31.820> optimization<00:01:32.000> before<00:01:33.070> okay<00:01:34.070> a<00:01:34.340> few 00:01:34.780 --> 00:01:34.790 align:start position:0% convex optimization before okay a few 00:01:34.790 --> 00:01:36.399 align:start position:0% convex optimization before okay a few hands<00:01:35.000> that's<00:01:35.270> great 00:01:36.399 --> 00:01:36.409 align:start position:0% hands that's great 00:01:36.409 --> 00:01:39.670 align:start position:0% hands that's great so<00:01:37.100> in<00:01:37.610> general<00:01:38.210> optimization<00:01:38.600> problems<00:01:39.500> have 00:01:39.670 --> 00:01:39.680 align:start position:0% so in general optimization problems have 00:01:39.680 --> 00:01:41.520 align:start position:0% so in general optimization problems have the<00:01:39.799> goal<00:01:39.950> of<00:01:39.979> minimizing<00:01:40.939> or<00:01:41.479> maximizing 00:01:41.520 --> 00:01:41.530 align:start position:0% the goal of minimizing or maximizing 00:01:41.530 --> 00:01:44.830 align:start position:0% the goal of minimizing or maximizing some<00:01:42.530> function<00:01:43.010> the<00:01:43.280> dark<00:01:43.790> curved<00:01:44.299> line<00:01:44.570> here 00:01:44.830 --> 00:01:44.840 align:start position:0% some function the dark curved line here 00:01:44.840 --> 00:01:47.560 align:start position:0% some function the dark curved line here subject<00:01:45.710> to<00:01:45.830> some<00:01:46.009> constraints<00:01:46.210> so<00:01:47.210> we<00:01:47.299> can't 00:01:47.560 --> 00:01:47.570 align:start position:0% subject to some constraints so we can't 00:01:47.570 --> 00:01:50.830 align:start position:0% subject to some constraints so we can't be<00:01:47.750> in<00:01:47.780> this<00:01:47.990> dark<00:01:48.350> blue<00:01:48.680> area<00:01:49.040> in<00:01:49.880> general 00:01:50.830 --> 00:01:50.840 align:start position:0% be in this dark blue area in general 00:01:50.840 --> 00:01:53.020 align:start position:0% be in this dark blue area in general these<00:01:51.140> are<00:01:51.200> pretty<00:01:51.500> hard<00:01:51.710> we<00:01:52.130> might<00:01:52.340> end<00:01:52.520> up<00:01:52.850> in 00:01:53.020 --> 00:01:53.030 align:start position:0% these are pretty hard we might end up in 00:01:53.030 --> 00:01:55.570 align:start position:0% these are pretty hard we might end up in the<00:01:53.150> select<00:01:53.420> local<00:01:53.780> minima<00:01:54.350> we'd<00:01:54.979> like<00:01:55.220> to<00:01:55.430> end 00:01:55.570 --> 00:01:55.580 align:start position:0% the select local minima we'd like to end 00:01:55.580 --> 00:01:58.090 align:start position:0% the select local minima we'd like to end up<00:01:55.850> in<00:01:56.030> the<00:01:56.210> global<00:01:56.600> minima<00:01:57.100> convex 00:01:58.090 --> 00:01:58.100 align:start position:0% up in the global minima convex 00:01:58.100 --> 00:02:00.069 align:start position:0% up in the global minima convex optimization<00:01:58.250> problems<00:01:59.180> make<00:01:59.659> this<00:01:59.780> a<00:01:59.840> lot 00:02:00.069 --> 00:02:00.079 align:start position:0% optimization problems make this a lot 00:02:00.079 --> 00:02:03.039 align:start position:0% optimization problems make this a lot easier<00:02:00.409> we<00:02:01.070> can<00:02:01.100> prove<00:02:01.579> that<00:02:01.939> a<00:02:02.360> type<00:02:02.869> of 00:02:03.039 --> 00:02:03.049 align:start position:0% easier we can prove that a type of 00:02:03.049 --> 00:02:04.630 align:start position:0% easier we can prove that a type of problem<00:02:03.470> it<00:02:03.590> basically<00:02:03.950> looks<00:02:04.159> like<00:02:04.220> a<00:02:04.340> bowl 00:02:04.630 --> 00:02:04.640 align:start position:0% problem it basically looks like a bowl 00:02:04.640 --> 00:02:06.340 align:start position:0% problem it basically looks like a bowl as<00:02:04.939> long<00:02:05.150> as<00:02:05.270> we<00:02:05.360> follow<00:02:05.600> the<00:02:05.689> slope<00:02:06.049> of<00:02:06.200> this 00:02:06.340 --> 00:02:06.350 align:start position:0% as long as we follow the slope of this 00:02:06.350 --> 00:02:08.139 align:start position:0% as long as we follow the slope of this downwards<00:02:06.860> we'll<00:02:07.100> end<00:02:07.340> up<00:02:07.490> at<00:02:07.700> the<00:02:07.970> right 00:02:08.139 --> 00:02:08.149 align:start position:0% downwards we'll end up at the right 00:02:08.149 --> 00:02:11.410 align:start position:0% downwards we'll end up at the right global<00:02:08.720> Optima<00:02:09.140> so<00:02:09.890> there's<00:02:10.580> a<00:02:10.640> lot<00:02:10.970> of<00:02:11.000> tools 00:02:11.410 --> 00:02:11.420 align:start position:0% global Optima so there's a lot of tools 00:02:11.420 --> 00:02:13.420 align:start position:0% global Optima so there's a lot of tools that<00:02:11.599> we<00:02:11.780> can<00:02:11.810> bring<00:02:12.170> to<00:02:12.349> bear<00:02:12.380> when<00:02:13.069> we<00:02:13.250> know 00:02:13.420 --> 00:02:13.430 align:start position:0% that we can bring to bear when we know 00:02:13.430 --> 00:02:14.140 align:start position:0% that we can bring to bear when we know that<00:02:13.610> a<00:02:13.640> problem<00:02:13.970> is 00:02:14.140 --> 00:02:14.150 align:start position:0% that a problem is 00:02:14.150 --> 00:02:16.750 align:start position:0% that a problem is convex<00:02:15.069> one<00:02:16.069> of<00:02:16.159> the<00:02:16.280> more<00:02:16.430> interesting 00:02:16.750 --> 00:02:16.760 align:start position:0% convex one of the more interesting 00:02:16.760 --> 00:02:18.759 align:start position:0% convex one of the more interesting components<00:02:17.450> of<00:02:17.510> this<00:02:17.659> is<00:02:17.870> that<00:02:18.140> we<00:02:18.230> can<00:02:18.260> think 00:02:18.759 --> 00:02:18.769 align:start position:0% components of this is that we can think 00:02:18.769 --> 00:02:21.069 align:start position:0% components of this is that we can think of<00:02:18.920> it<00:02:19.069> in<00:02:19.220> a<00:02:19.370> decentralized<00:02:20.269> optimization 00:02:21.069 --> 00:02:21.079 align:start position:0% of it in a decentralized optimization 00:02:21.079 --> 00:02:23.710 align:start position:0% of it in a decentralized optimization form<00:02:21.610> conventionally<00:02:22.610> we<00:02:22.879> just<00:02:23.120> have<00:02:23.329> one 00:02:23.710 --> 00:02:23.720 align:start position:0% form conventionally we just have one 00:02:23.720 --> 00:02:25.990 align:start position:0% form conventionally we just have one System<00:02:24.379> Operator<00:02:24.560> who<00:02:25.129> has<00:02:25.370> all<00:02:25.640> of<00:02:25.700> the 00:02:25.990 --> 00:02:26.000 align:start position:0% System Operator who has all of the 00:02:26.000 --> 00:02:28.059 align:start position:0% System Operator who has all of the information<00:02:26.150> about<00:02:26.659> all<00:02:27.019> of<00:02:27.500> the<00:02:27.590> data<00:02:27.769> all 00:02:28.059 --> 00:02:28.069 align:start position:0% information about all of the data all 00:02:28.069 --> 00:02:30.699 align:start position:0% information about all of the data all the<00:02:28.280> constraints<00:02:28.849> in<00:02:29.060> the<00:02:29.180> system<00:02:29.540> so<00:02:30.170> even<00:02:30.620> if 00:02:30.699 --> 00:02:30.709 align:start position:0% the constraints in the system so even if 00:02:30.709 --> 00:02:32.770 align:start position:0% the constraints in the system so even if these<00:02:30.890> are<00:02:31.069> say<00:02:31.310> individual<00:02:31.939> households<00:02:32.629> who 00:02:32.770 --> 00:02:32.780 align:start position:0% these are say individual households who 00:02:32.780 --> 00:02:34.179 align:start position:0% these are say individual households who have<00:02:32.810> mobility<00:02:33.260> constraints<00:02:33.920> on<00:02:34.040> their 00:02:34.179 --> 00:02:34.189 align:start position:0% have mobility constraints on their 00:02:34.189 --> 00:02:36.509 align:start position:0% have mobility constraints on their electric<00:02:34.519> vehicles<00:02:35.019> individual 00:02:36.509 --> 00:02:36.519 align:start position:0% electric vehicles individual 00:02:36.519 --> 00:02:38.080 align:start position:0% electric vehicles individual participants<00:02:37.519> who<00:02:37.609> have<00:02:37.790> different 00:02:38.080 --> 00:02:38.090 align:start position:0% participants who have different 00:02:38.090 --> 00:02:39.520 align:start position:0% participants who have different investment<00:02:38.599> goals<00:02:38.810> for<00:02:39.200> portfolio 00:02:39.520 --> 00:02:39.530 align:start position:0% investment goals for portfolio 00:02:39.530 --> 00:02:43.420 align:start position:0% investment goals for portfolio optimization<00:02:40.629> individual<00:02:42.430> manufactories 00:02:43.420 --> 00:02:43.430 align:start position:0% optimization individual manufactories 00:02:43.430 --> 00:02:45.880 align:start position:0% optimization individual manufactories that<00:02:43.670> have<00:02:43.819> constraints<00:02:44.450> on<00:02:44.569> production<00:02:45.049> one 00:02:45.880 --> 00:02:45.890 align:start position:0% that have constraints on production one 00:02:45.890 --> 00:02:47.949 align:start position:0% that have constraints on production one System<00:02:46.370> Operator<00:02:46.489> would<00:02:47.000> have<00:02:47.480> all<00:02:47.810> the 00:02:47.949 --> 00:02:47.959 align:start position:0% System Operator would have all the 00:02:47.959 --> 00:02:49.270 align:start position:0% System Operator would have all the knowledge<00:02:48.139> about<00:02:48.409> all<00:02:48.769> those<00:02:48.950> constraints 00:02:49.270 --> 00:02:49.280 align:start position:0% knowledge about all those constraints 00:02:49.280 --> 00:02:51.610 align:start position:0% knowledge about all those constraints and<00:02:49.790> run<00:02:50.239> the<00:02:50.329> centralized<00:02:50.900> optimization 00:02:51.610 --> 00:02:51.620 align:start position:0% and run the centralized optimization 00:02:51.620 --> 00:02:54.009 align:start position:0% and run the centralized optimization problem<00:02:52.040> this<00:02:52.879> is<00:02:53.030> great<00:02:53.299> if<00:02:53.569> all<00:02:53.810> of<00:02:53.930> those 00:02:54.009 --> 00:02:54.019 align:start position:0% problem this is great if all of those 00:02:54.019 --> 00:02:56.039 align:start position:0% problem this is great if all of those entities<00:02:54.200> are<00:02:54.590> owned<00:02:54.859> by<00:02:55.129> the<00:02:55.190> same<00:02:55.459> person 00:02:56.039 --> 00:02:56.049 align:start position:0% entities are owned by the same person 00:02:56.049 --> 00:02:59.170 align:start position:0% entities are owned by the same person it's<00:02:57.049> hell<00:02:57.349> if<00:02:57.650> you<00:02:57.950> end<00:02:58.099> up<00:02:58.220> with<00:02:58.430> people<00:02:58.970> who 00:02:59.170 --> 00:02:59.180 align:start position:0% it's hell if you end up with people who 00:02:59.180 --> 00:03:01.089 align:start position:0% it's hell if you end up with people who don't<00:02:59.359> trust<00:02:59.689> each<00:02:59.840> other<00:03:00.049> and<00:03:00.260> to<00:03:00.769> reach<00:03:00.950> a 00:03:01.089 --> 00:03:01.099 align:start position:0% don't trust each other and to reach a 00:03:01.099 --> 00:03:03.129 align:start position:0% don't trust each other and to reach a global<00:03:01.459> optimum<00:03:01.849> you<00:03:02.090> need<00:03:02.299> some<00:03:02.629> sort<00:03:02.840> of<00:03:02.989> way 00:03:03.129 --> 00:03:03.139 align:start position:0% global optimum you need some sort of way 00:03:03.139 --> 00:03:05.500 align:start position:0% global optimum you need some sort of way of<00:03:03.609> decentralizing<00:03:04.609> this<00:03:04.730> problem<00:03:05.150> into<00:03:05.359> a 00:03:05.500 --> 00:03:05.510 align:start position:0% of decentralizing this problem into a 00:03:05.510 --> 00:03:08.589 align:start position:0% of decentralizing this problem into a trustless<00:03:06.250> cooperative<00:03:07.250> system<00:03:07.670> so 00:03:08.589 --> 00:03:08.599 align:start position:0% trustless cooperative system so 00:03:08.599 --> 00:03:11.349 align:start position:0% trustless cooperative system so decentralized<00:03:09.319> problems<00:03:09.829> break<00:03:10.549> that<00:03:10.819> into<00:03:11.120> a 00:03:11.349 --> 00:03:11.359 align:start position:0% decentralized problems break that into a 00:03:11.359 --> 00:03:13.629 align:start position:0% decentralized problems break that into a set<00:03:11.810> of<00:03:11.840> local<00:03:12.200> optimization<00:03:12.950> problems<00:03:13.400> these 00:03:13.629 --> 00:03:13.639 align:start position:0% set of local optimization problems these 00:03:13.639 --> 00:03:15.940 align:start position:0% set of local optimization problems these blue<00:03:13.819> dots<00:03:14.030> that<00:03:14.810> then<00:03:14.930> coordinate<00:03:15.260> back<00:03:15.709> and 00:03:15.940 --> 00:03:15.950 align:start position:0% blue dots that then coordinate back and 00:03:15.950 --> 00:03:17.440 align:start position:0% blue dots that then coordinate back and forth<00:03:16.159> with<00:03:16.340> an<00:03:16.430> aggregator<00:03:16.879> until<00:03:17.329> they 00:03:17.440 --> 00:03:17.450 align:start position:0% forth with an aggregator until they 00:03:17.450 --> 00:03:20.020 align:start position:0% forth with an aggregator until they reach<00:03:17.629> consensus<00:03:18.290> on<00:03:18.440> that<00:03:18.530> global<00:03:19.340> optimum 00:03:20.020 --> 00:03:20.030 align:start position:0% reach consensus on that global optimum 00:03:20.030 --> 00:03:21.339 align:start position:0% reach consensus on that global optimum this<00:03:20.540> is<00:03:20.660> really<00:03:20.780> important<00:03:21.169> we're<00:03:21.260> actually 00:03:21.339 --> 00:03:21.349 align:start position:0% this is really important we're actually 00:03:21.349 --> 00:03:22.990 align:start position:0% this is really important we're actually able<00:03:21.590> to<00:03:21.709> guarantee<00:03:21.949> that<00:03:22.250> we're<00:03:22.760> going<00:03:22.879> to 00:03:22.990 --> 00:03:23.000 align:start position:0% able to guarantee that we're going to 00:03:23.000 --> 00:03:25.659 align:start position:0% able to guarantee that we're going to hit<00:03:23.239> the<00:03:23.419> global<00:03:23.870> optimum<00:03:24.349> even<00:03:25.190> though<00:03:25.370> we 00:03:25.659 --> 00:03:25.669 align:start position:0% hit the global optimum even though we 00:03:25.669 --> 00:03:28.240 align:start position:0% hit the global optimum even though we start<00:03:25.970> up<00:03:26.090> with<00:03:26.120> these<00:03:26.389> local<00:03:26.720> problems<00:03:27.410> why 00:03:28.240 --> 00:03:28.250 align:start position:0% start up with these local problems why 00:03:28.250 --> 00:03:31.000 align:start position:0% start up with these local problems why would<00:03:28.459> we<00:03:28.579> bother<00:03:28.730> with<00:03:28.940> this<00:03:29.120> so<00:03:29.780> one<00:03:30.470> really 00:03:31.000 --> 00:03:31.010 align:start position:0% would we bother with this so one really 00:03:31.010 --> 00:03:32.770 align:start position:0% would we bother with this so one really beautiful<00:03:31.639> part<00:03:31.849> of<00:03:31.909> this<00:03:32.060> is<00:03:32.329> that<00:03:32.540> we<00:03:32.629> can 00:03:32.770 --> 00:03:32.780 align:start position:0% beautiful part of this is that we can 00:03:32.780 --> 00:03:35.979 align:start position:0% beautiful part of this is that we can actually<00:03:32.930> take<00:03:33.739> a<00:03:34.069> hard<00:03:34.489> big<00:03:35.239> problem<00:03:35.750> break 00:03:35.979 --> 00:03:35.989 align:start position:0% actually take a hard big problem break 00:03:35.989 --> 00:03:38.559 align:start position:0% actually take a hard big problem break it<00:03:36.109> into<00:03:36.349> a<00:03:36.400> set<00:03:37.400> of<00:03:37.519> analytically<00:03:38.150> solvable 00:03:38.559 --> 00:03:38.569 align:start position:0% it into a set of analytically solvable 00:03:38.569 --> 00:03:40.809 align:start position:0% it into a set of analytically solvable subproblems<00:03:39.379> so<00:03:39.859> you<00:03:39.949> can<00:03:40.040> get<00:03:40.159> a<00:03:40.190> thousand<00:03:40.669> X 00:03:40.809 --> 00:03:40.819 align:start position:0% subproblems so you can get a thousand X 00:03:40.819 --> 00:03:43.689 align:start position:0% subproblems so you can get a thousand X speed-up<00:03:41.389> in<00:03:41.680> computation<00:03:42.680> time<00:03:42.829> even<00:03:43.579> while 00:03:43.689 --> 00:03:43.699 align:start position:0% speed-up in computation time even while 00:03:43.699 --> 00:03:45.479 align:start position:0% speed-up in computation time even while running<00:03:43.729> it<00:03:44.060> on<00:03:44.150> the<00:03:44.180> same<00:03:44.510> computing<00:03:45.109> node 00:03:45.479 --> 00:03:45.489 align:start position:0% running it on the same computing node 00:03:45.489 --> 00:03:48.069 align:start position:0% running it on the same computing node another<00:03:46.489> part<00:03:46.760> is<00:03:46.879> just<00:03:47.090> breaking<00:03:47.479> a<00:03:47.720> hard 00:03:48.069 --> 00:03:48.079 align:start position:0% another part is just breaking a hard 00:03:48.079 --> 00:03:50.710 align:start position:0% another part is just breaking a hard problem<00:03:48.440> that<00:03:49.010> has<00:03:49.340> say<00:03:49.909> greater<00:03:50.629> than 00:03:50.710 --> 00:03:50.720 align:start position:0% problem that has say greater than 00:03:50.720 --> 00:03:53.439 align:start position:0% problem that has say greater than quadratic<00:03:51.139> compute<00:03:51.889> complexity<00:03:52.489> and<00:03:52.699> break 00:03:53.439 --> 00:03:53.449 align:start position:0% quadratic compute complexity and break 00:03:53.449 --> 00:03:55.839 align:start position:0% quadratic compute complexity and break it<00:03:53.569> into<00:03:53.720> a<00:03:53.840> set<00:03:54.049> of<00:03:54.169> small<00:03:54.620> problems<00:03:55.129> even 00:03:55.839 --> 00:03:55.849 align:start position:0% it into a set of small problems even 00:03:55.849 --> 00:03:57.250 align:start position:0% it into a set of small problems even though<00:03:55.940> you're<00:03:56.090> breaking<00:03:56.389> into<00:03:56.540> n<00:03:56.780> smaller 00:03:57.250 --> 00:03:57.260 align:start position:0% though you're breaking into n smaller 00:03:57.260 --> 00:03:59.800 align:start position:0% though you're breaking into n smaller problems<00:03:57.799> you've<00:03:58.370> practically<00:03:59.239> reduced<00:03:59.449> your 00:03:59.800 --> 00:03:59.810 align:start position:0% problems you've practically reduced your 00:03:59.810 --> 00:04:02.050 align:start position:0% problems you've practically reduced your compute<00:04:00.199> time<00:04:00.409> then<00:04:01.280> the<00:04:01.400> other<00:04:01.549> part<00:04:01.849> is<00:04:02.000> that 00:04:02.050 --> 00:04:02.060 align:start position:0% compute time then the other part is that 00:04:02.060 --> 00:04:03.819 align:start position:0% compute time then the other part is that if<00:04:02.389> there's<00:04:02.720> those<00:04:02.959> privacy<00:04:03.349> and<00:04:03.590> trust 00:04:03.819 --> 00:04:03.829 align:start position:0% if there's those privacy and trust 00:04:03.829 --> 00:04:05.469 align:start position:0% if there's those privacy and trust issues<00:04:04.159> you<00:04:04.489> don't<00:04:04.699> want<00:04:05.000> to<00:04:05.150> share<00:04:05.299> all<00:04:05.449> that 00:04:05.469 --> 00:04:05.479 align:start position:0% issues you don't want to share all that 00:04:05.479 --> 00:04:08.159 align:start position:0% issues you don't want to share all that information<00:04:06.139> with<00:04:06.769> the<00:04:06.889> global<00:04:07.190> aggregator 00:04:08.159 --> 00:04:08.169 align:start position:0% information with the global aggregator 00:04:08.169 --> 00:04:11.500 align:start position:0% information with the global aggregator so<00:04:09.169> the<00:04:09.560> aggregator<00:04:10.099> that<00:04:11.000> person<00:04:11.419> who 00:04:11.500 --> 00:04:11.510 align:start position:0% so the aggregator that person who 00:04:11.510 --> 00:04:13.750 align:start position:0% so the aggregator that person who coordinates<00:04:11.780> between<00:04:12.229> all<00:04:12.590> these<00:04:12.769> nodes<00:04:13.069> has 00:04:13.750 --> 00:04:13.760 align:start position:0% coordinates between all these nodes has 00:04:13.760 --> 00:04:16.060 align:start position:0% coordinates between all these nodes has a<00:04:13.790> really<00:04:14.120> easy<00:04:14.359> job<00:04:14.659> fortunately<00:04:15.379> these 00:04:16.060 --> 00:04:16.070 align:start position:0% a really easy job fortunately these 00:04:16.070 --> 00:04:19.779 align:start position:0% a really easy job fortunately these nodes<00:04:16.519> have<00:04:16.940> hard<00:04:17.449> compute<00:04:18.019> problems<00:04:18.789> subject 00:04:19.779 --> 00:04:19.789 align:start position:0% nodes have hard compute problems subject 00:04:19.789 --> 00:04:21.520 align:start position:0% nodes have hard compute problems subject to<00:04:19.940> local<00:04:20.239> constraints<00:04:20.930> about<00:04:21.049> which<00:04:21.349> they 00:04:21.520 --> 00:04:21.530 align:start position:0% to local constraints about which they 00:04:21.530 --> 00:04:23.649 align:start position:0% to local constraints about which they have<00:04:21.650> private<00:04:22.070> information<00:04:22.610> but<00:04:23.450> the 00:04:23.649 --> 00:04:23.659 align:start position:0% have private information but the 00:04:23.659 --> 00:04:25.690 align:start position:0% have private information but the aggregator<00:04:24.229> only<00:04:24.590> has<00:04:24.710> a<00:04:24.740> very<00:04:25.099> simple 00:04:25.690 --> 00:04:25.700 align:start position:0% aggregator only has a very simple 00:04:25.700 --> 00:04:26.610 align:start position:0% aggregator only has a very simple arithmetic 00:04:26.610 --> 00:04:26.620 align:start position:0% arithmetic 00:04:26.620 --> 00:04:27.990 align:start position:0% arithmetic update<00:04:26.980> problem<00:04:27.340> in<00:04:27.490> a<00:04:27.550> lot<00:04:27.760> of<00:04:27.880> these 00:04:27.990 --> 00:04:28.000 align:start position:0% update problem in a lot of these 00:04:28.000 --> 00:04:30.659 align:start position:0% update problem in a lot of these formulations<00:04:28.449> so<00:04:29.290> even<00:04:29.680> though<00:04:29.800> this<00:04:30.070> can<00:04:30.550> be 00:04:30.659 --> 00:04:30.669 align:start position:0% formulations so even though this can be 00:04:30.669 --> 00:04:32.070 align:start position:0% formulations so even though this can be something<00:04:30.910> that<00:04:31.240> takes<00:04:31.270> minutes<00:04:31.930> or<00:04:32.020> even 00:04:32.070 --> 00:04:32.080 align:start position:0% something that takes minutes or even 00:04:32.080 --> 00:04:35.310 align:start position:0% something that takes minutes or even hours<00:04:32.380> to<00:04:32.740> compute<00:04:33.130> off<00:04:33.310> chain<00:04:33.820> on<00:04:34.090> a<00:04:34.150> CPU<00:04:35.020> or<00:04:35.290> a 00:04:35.310 --> 00:04:35.320 align:start position:0% hours to compute off chain on a CPU or a 00:04:35.320 --> 00:04:38.070 align:start position:0% hours to compute off chain on a CPU or a server<00:04:35.860> cluster<00:04:36.310> this<00:04:37.270> is<00:04:37.479> just<00:04:37.660> a<00:04:37.780> really 00:04:38.070 --> 00:04:38.080 align:start position:0% server cluster this is just a really 00:04:38.080 --> 00:04:40.439 align:start position:0% server cluster this is just a really simple<00:04:38.250> update<00:04:39.250> problem<00:04:39.490> so<00:04:39.880> this<00:04:40.000> is<00:04:40.120> just 00:04:40.439 --> 00:04:40.449 align:start position:0% simple update problem so this is just 00:04:40.449 --> 00:04:42.290 align:start position:0% simple update problem so this is just adding<00:04:40.690> all<00:04:40.930> those<00:04:41.080> numbers<00:04:41.440> together<00:04:41.530> and 00:04:42.290 --> 00:04:42.300 align:start position:0% adding all those numbers together and 00:04:42.300 --> 00:04:44.129 align:start position:0% adding all those numbers together and broadcasting<00:04:43.300> the<00:04:43.389> update<00:04:43.660> to<00:04:43.780> the<00:04:43.870> rest<00:04:44.020> of 00:04:44.129 --> 00:04:44.139 align:start position:0% broadcasting the update to the rest of 00:04:44.139 --> 00:04:46.290 align:start position:0% broadcasting the update to the rest of the<00:04:44.199> network<00:04:44.350> it's<00:04:45.310> sort<00:04:45.520> of<00:04:45.580> like<00:04:45.729> a<00:04:45.760> market 00:04:46.290 --> 00:04:46.300 align:start position:0% the network it's sort of like a market 00:04:46.300 --> 00:04:49.219 align:start position:0% the network it's sort of like a market so<00:04:46.479> individuals<00:04:47.260> have<00:04:47.800> a<00:04:47.830> sense<00:04:48.280> of<00:04:48.430> their 00:04:49.219 --> 00:04:49.229 align:start position:0% so individuals have a sense of their 00:04:49.229 --> 00:04:51.659 align:start position:0% so individuals have a sense of their value<00:04:50.229> functions<00:04:50.710> for<00:04:51.010> a<00:04:51.040> particular<00:04:51.490> good 00:04:51.659 --> 00:04:51.669 align:start position:0% value functions for a particular good 00:04:51.669 --> 00:04:53.939 align:start position:0% value functions for a particular good they<00:04:52.360> can<00:04:52.570> propose<00:04:52.900> how<00:04:53.199> much<00:04:53.380> they'd<00:04:53.590> like<00:04:53.740> to 00:04:53.939 --> 00:04:53.949 align:start position:0% they can propose how much they'd like to 00:04:53.949 --> 00:04:56.420 align:start position:0% they can propose how much they'd like to sell<00:04:54.130> or<00:04:54.280> buy<00:04:54.310> the<00:04:55.180> market<00:04:55.510> operator 00:04:56.420 --> 00:04:56.430 align:start position:0% sell or buy the market operator 00:04:56.430 --> 00:04:58.740 align:start position:0% sell or buy the market operator aggregates<00:04:57.430> all<00:04:57.580> those<00:04:57.729> together<00:04:57.910> estimates 00:04:58.740 --> 00:04:58.750 align:start position:0% aggregates all those together estimates 00:04:58.750 --> 00:05:00.210 align:start position:0% aggregates all those together estimates what<00:04:59.020> the<00:04:59.139> market<00:04:59.410> clearing<00:04:59.590> price<00:04:59.889> would<00:05:00.100> be 00:05:00.210 --> 00:05:00.220 align:start position:0% what the market clearing price would be 00:05:00.220 --> 00:05:02.250 align:start position:0% what the market clearing price would be broadcast<00:05:00.639> that<00:05:01.060> price<00:05:01.330> out<00:05:01.540> to<00:05:01.690> everybody 00:05:02.250 --> 00:05:02.260 align:start position:0% broadcast that price out to everybody 00:05:02.260 --> 00:05:05.250 align:start position:0% broadcast that price out to everybody people<00:05:02.560> can<00:05:02.680> update<00:05:02.889> their<00:05:03.780> their<00:05:04.780> estimates 00:05:05.250 --> 00:05:05.260 align:start position:0% people can update their their estimates 00:05:05.260 --> 00:05:09.360 align:start position:0% people can update their their estimates of<00:05:05.380> their<00:05:05.500> bids<00:05:05.830> or<00:05:06.760> asks<00:05:07.180> for<00:05:07.450> the<00:05:07.600> object<00:05:08.370> and 00:05:09.360 --> 00:05:09.370 align:start position:0% of their bids or asks for the object and 00:05:09.370 --> 00:05:11.100 align:start position:0% of their bids or asks for the object and when<00:05:09.880> you<00:05:09.970> iterate<00:05:10.150> back<00:05:10.510> and<00:05:10.720> forth<00:05:10.930> with 00:05:11.100 --> 00:05:11.110 align:start position:0% when you iterate back and forth with 00:05:11.110 --> 00:05:12.870 align:start position:0% when you iterate back and forth with that<00:05:11.260> you're<00:05:11.470> able<00:05:11.650> to<00:05:11.680> reach<00:05:11.950> consensus<00:05:12.639> and 00:05:12.870 --> 00:05:12.880 align:start position:0% that you're able to reach consensus and 00:05:12.880 --> 00:05:14.790 align:start position:0% that you're able to reach consensus and for<00:05:13.419> these<00:05:13.570> comics<00:05:13.930> optimization<00:05:14.500> problems 00:05:14.790 --> 00:05:14.800 align:start position:0% for these comics optimization problems 00:05:14.800 --> 00:05:16.140 align:start position:0% for these comics optimization problems we<00:05:14.889> can<00:05:15.040> prove<00:05:15.250> that<00:05:15.310> that's<00:05:15.639> the<00:05:15.940> global 00:05:16.140 --> 00:05:16.150 align:start position:0% we can prove that that's the global 00:05:16.150 --> 00:05:17.450 align:start position:0% we can prove that that's the global optimum 00:05:17.450 --> 00:05:17.460 align:start position:0% optimum 00:05:17.460 --> 00:05:21.779 align:start position:0% optimum so<00:05:18.460> going<00:05:19.210> from<00:05:19.510> these<00:05:19.830> centralized<00:05:20.830> or<00:05:21.280> sorry 00:05:21.779 --> 00:05:21.789 align:start position:0% so going from these centralized or sorry 00:05:21.789 --> 00:05:24.060 align:start position:0% so going from these centralized or sorry decentralized<00:05:22.750> problems<00:05:23.260> to<00:05:23.710> a<00:05:23.740> fully 00:05:24.060 --> 00:05:24.070 align:start position:0% decentralized problems to a fully 00:05:24.070 --> 00:05:26.219 align:start position:0% decentralized problems to a fully decentralized<00:05:24.280> model<00:05:25.210> one<00:05:26.020> of<00:05:26.110> the 00:05:26.219 --> 00:05:26.229 align:start position:0% decentralized model one of the 00:05:26.229 --> 00:05:27.719 align:start position:0% decentralized model one of the challenges<00:05:26.710> with<00:05:26.860> this<00:05:27.100> model<00:05:27.669> of 00:05:27.719 --> 00:05:27.729 align:start position:0% challenges with this model of 00:05:27.729 --> 00:05:29.400 align:start position:0% challenges with this model of architecture<00:05:28.240> is<00:05:28.300> the<00:05:28.330> aggregator<00:05:28.930> has<00:05:29.199> to 00:05:29.400 --> 00:05:29.410 align:start position:0% architecture is the aggregator has to 00:05:29.410 --> 00:05:31.230 align:start position:0% architecture is the aggregator has to have<00:05:29.710> direct<00:05:30.039> lines<00:05:30.340> of<00:05:30.490> communication<00:05:31.090> with 00:05:31.230 --> 00:05:31.240 align:start position:0% have direct lines of communication with 00:05:31.240 --> 00:05:33.810 align:start position:0% have direct lines of communication with every<00:05:31.630> single<00:05:31.990> node<00:05:32.200> so<00:05:32.770> not<00:05:33.310> a<00:05:33.340> problem<00:05:33.700> when 00:05:33.810 --> 00:05:33.820 align:start position:0% every single node so not a problem when 00:05:33.820 --> 00:05:35.790 align:start position:0% every single node so not a problem when you<00:05:33.880> just<00:05:34.030> have<00:05:34.180> three<00:05:34.479> nodes<00:05:34.660> here<00:05:35.050> for<00:05:35.740> the 00:05:35.790 --> 00:05:35.800 align:start position:0% you just have three nodes here for the 00:05:35.800 --> 00:05:37.469 align:start position:0% you just have three nodes here for the types<00:05:36.010> of<00:05:36.160> electricity<00:05:36.850> market<00:05:37.150> problems 00:05:37.469 --> 00:05:37.479 align:start position:0% types of electricity market problems 00:05:37.479 --> 00:05:39.060 align:start position:0% types of electricity market problems that<00:05:37.630> I<00:05:37.660> consider<00:05:38.169> where<00:05:38.530> you're<00:05:38.650> trying<00:05:38.919> to 00:05:39.060 --> 00:05:39.070 align:start position:0% that I consider where you're trying to 00:05:39.070 --> 00:05:41.340 align:start position:0% that I consider where you're trying to control<00:05:39.370> say<00:05:39.760> 10<00:05:40.120> million<00:05:40.330> electric<00:05:40.840> vehicles 00:05:41.340 --> 00:05:41.350 align:start position:0% control say 10 million electric vehicles 00:05:41.350 --> 00:05:43.589 align:start position:0% control say 10 million electric vehicles you<00:05:41.889> don't<00:05:42.070> want<00:05:42.250> to<00:05:42.400> have<00:05:42.580> one<00:05:42.970> server<00:05:43.270> that's 00:05:43.589 --> 00:05:43.599 align:start position:0% you don't want to have one server that's 00:05:43.599 --> 00:05:45.060 align:start position:0% you don't want to have one server that's trying<00:05:43.810> to<00:05:43.900> talk<00:05:44.080> to<00:05:44.260> 10<00:05:44.470> million<00:05:44.650> electric 00:05:45.060 --> 00:05:45.070 align:start position:0% trying to talk to 10 million electric 00:05:45.070 --> 00:05:47.670 align:start position:0% trying to talk to 10 million electric vehicles<00:05:45.099> all<00:05:45.760> at<00:05:45.970> once<00:05:46.120> you<00:05:46.840> need<00:05:46.960> a<00:05:47.200> more 00:05:47.670 --> 00:05:47.680 align:start position:0% vehicles all at once you need a more 00:05:47.680 --> 00:05:49.650 align:start position:0% vehicles all at once you need a more decentralized<00:05:48.070> model<00:05:48.789> so<00:05:49.270> fully 00:05:49.650 --> 00:05:49.660 align:start position:0% decentralized model so fully 00:05:49.660 --> 00:05:51.900 align:start position:0% decentralized model so fully decentralized<00:05:50.039> optimization<00:05:51.039> has<00:05:51.250> become<00:05:51.669> a 00:05:51.900 --> 00:05:51.910 align:start position:0% decentralized optimization has become a 00:05:51.910 --> 00:05:54.810 align:start position:0% decentralized optimization has become a tool<00:05:52.539> that's<00:05:52.780> in<00:05:52.930> vogue<00:05:53.139> in<00:05:53.440> particular<00:05:53.979> for 00:05:54.810 --> 00:05:54.820 align:start position:0% tool that's in vogue in particular for 00:05:54.820 --> 00:05:56.400 align:start position:0% tool that's in vogue in particular for these<00:05:54.970> really<00:05:55.180> large<00:05:55.539> decentralized 00:05:56.400 --> 00:05:56.410 align:start position:0% these really large decentralized 00:05:56.410 --> 00:05:58.529 align:start position:0% these really large decentralized optimization<00:05:57.070> problems<00:05:57.460> one<00:05:58.360> of<00:05:58.479> the 00:05:58.529 --> 00:05:58.539 align:start position:0% optimization problems one of the 00:05:58.539 --> 00:06:00.000 align:start position:0% optimization problems one of the challenges<00:05:59.020> though<00:05:59.200> is<00:05:59.410> that<00:05:59.440> you're<00:05:59.710> only 00:06:00.000 --> 00:06:00.010 align:start position:0% challenges though is that you're only 00:06:00.010 --> 00:06:01.710 align:start position:0% challenges though is that you're only sharing<00:06:00.370> information<00:06:00.729> with<00:06:01.180> your<00:06:01.300> neighbors 00:06:01.710 --> 00:06:01.720 align:start position:0% sharing information with your neighbors 00:06:01.720 --> 00:06:03.960 align:start position:0% sharing information with your neighbors and<00:06:01.930> those<00:06:02.860> neighbors<00:06:03.280> might<00:06:03.490> be<00:06:03.639> lying<00:06:03.880> to 00:06:03.960 --> 00:06:03.970 align:start position:0% and those neighbors might be lying to 00:06:03.970 --> 00:06:05.670 align:start position:0% and those neighbors might be lying to you<00:06:04.180> if<00:06:04.210> they've<00:06:04.570> been<00:06:04.990> compromised 00:06:05.670 --> 00:06:05.680 align:start position:0% you if they've been compromised 00:06:05.680 --> 00:06:07.680 align:start position:0% you if they've been compromised so<00:06:06.070> while<00:06:06.550> you<00:06:06.700> might<00:06:06.849> be<00:06:07.120> communicating<00:06:07.300> with 00:06:07.680 --> 00:06:07.690 align:start position:0% so while you might be communicating with 00:06:07.690 --> 00:06:09.810 align:start position:0% so while you might be communicating with your<00:06:07.900> neighbor<00:06:08.080> that<00:06:08.380> neighbor<00:06:09.070> may<00:06:09.460> have<00:06:09.639> had 00:06:09.810 --> 00:06:09.820 align:start position:0% your neighbor that neighbor may have had 00:06:09.820 --> 00:06:11.580 align:start position:0% your neighbor that neighbor may have had his<00:06:10.030> electric<00:06:10.419> vehicle<00:06:10.780> hacked<00:06:11.020> by<00:06:11.200> somebody 00:06:11.580 --> 00:06:11.590 align:start position:0% his electric vehicle hacked by somebody 00:06:11.590 --> 00:06:14.010 align:start position:0% his electric vehicle hacked by somebody and<00:06:11.800> that<00:06:12.250> hacker<00:06:12.550> is<00:06:12.789> attempting<00:06:13.270> to<00:06:13.300> cause<00:06:13.990> a 00:06:14.010 --> 00:06:14.020 align:start position:0% and that hacker is attempting to cause a 00:06:14.020 --> 00:06:15.719 align:start position:0% and that hacker is attempting to cause a blackout<00:06:14.530> in<00:06:14.590> the<00:06:14.800> electricity<00:06:15.310> grid<00:06:15.490> for 00:06:15.719 --> 00:06:15.729 align:start position:0% blackout in the electricity grid for 00:06:15.729 --> 00:06:17.790 align:start position:0% blackout in the electricity grid for instance<00:06:16.030> so<00:06:16.690> we<00:06:16.870> have<00:06:16.990> a<00:06:17.020> question<00:06:17.440> of<00:06:17.560> how<00:06:17.740> do 00:06:17.790 --> 00:06:17.800 align:start position:0% instance so we have a question of how do 00:06:17.800 --> 00:06:19.680 align:start position:0% instance so we have a question of how do we<00:06:17.889> identify<00:06:18.000> the<00:06:19.000> nodes<00:06:19.210> that<00:06:19.450> have<00:06:19.510> been 00:06:19.680 --> 00:06:19.690 align:start position:0% we identify the nodes that have been 00:06:19.690 --> 00:06:21.900 align:start position:0% we identify the nodes that have been compromised<00:06:20.200> how<00:06:20.740> do<00:06:20.800> we<00:06:20.919> reach<00:06:21.430> stable 00:06:21.900 --> 00:06:21.910 align:start position:0% compromised how do we reach stable 00:06:21.910 --> 00:06:24.830 align:start position:0% compromised how do we reach stable operation<00:06:22.300> and<00:06:22.720> then<00:06:23.139> can<00:06:23.470> we<00:06:23.590> guarantee 00:06:24.830 --> 00:06:24.840 align:start position:0% operation and then can we guarantee 00:06:24.840 --> 00:06:28.580 align:start position:0% operation and then can we guarantee feasibility<00:06:25.840> or<00:06:25.870> optimality<00:06:26.800> of<00:06:27.160> the<00:06:27.370> system 00:06:28.580 --> 00:06:28.590 align:start position:0% feasibility or optimality of the system 00:06:28.590 --> 00:06:35.240 align:start position:0% feasibility or optimality of the system when<00:06:29.590> these<00:06:29.800> nodes<00:06:30.070> have<00:06:30.160> been<00:06:30.250> compromised 00:06:35.240 --> 00:06:35.250 align:start position:0% 00:06:35.250 --> 00:06:38.360 align:start position:0% so<00:06:35.550> we<00:06:35.940> have<00:06:36.600> this<00:06:36.810> like<00:06:37.370> bi-directional 00:06:38.360 --> 00:06:38.370 align:start position:0% so we have this like bi-directional 00:06:38.370 --> 00:06:40.480 align:start position:0% so we have this like bi-directional communication<00:06:38.940> between<00:06:39.180> each<00:06:39.630> neighbor<00:06:39.960> so 00:06:40.480 --> 00:06:40.490 align:start position:0% communication between each neighbor so 00:06:40.490 --> 00:06:43.010 align:start position:0% communication between each neighbor so we<00:06:41.490> just<00:06:41.820> see<00:06:42.030> messages<00:06:42.360> that<00:06:42.750> our<00:06:42.840> neighbor 00:06:43.010 --> 00:06:43.020 align:start position:0% we just see messages that our neighbor 00:06:43.020 --> 00:06:45.410 align:start position:0% we just see messages that our neighbor is<00:06:43.200> sending<00:06:43.500> us<00:06:43.650> in<00:06:44.250> this<00:06:44.460> case<00:06:44.670> the<00:06:45.030> Red<00:06:45.180> Nosed 00:06:45.410 --> 00:06:45.420 align:start position:0% is sending us in this case the Red Nosed 00:06:45.420 --> 00:06:46.670 align:start position:0% is sending us in this case the Red Nosed has<00:06:45.660> been<00:06:45.840> compromised 00:06:46.670 --> 00:06:46.680 align:start position:0% has been compromised 00:06:46.680 --> 00:06:48.260 align:start position:0% has been compromised so<00:06:46.800> rather<00:06:47.070> than<00:06:47.220> the<00:06:47.430> normal<00:06:47.730> operation<00:06:48.210> of 00:06:48.260 --> 00:06:48.270 align:start position:0% so rather than the normal operation of 00:06:48.270 --> 00:06:50.810 align:start position:0% so rather than the normal operation of the<00:06:48.330> system<00:06:48.780> the<00:06:49.440> red<00:06:49.620> node<00:06:49.830> could<00:06:50.190> distort 00:06:50.810 --> 00:06:50.820 align:start position:0% the system the red node could distort 00:06:50.820 --> 00:06:53.660 align:start position:0% the system the red node could distort its<00:06:51.540> private<00:06:51.990> optimization<00:06:52.590> problem<00:06:52.950> so<00:06:53.250> it 00:06:53.660 --> 00:06:53.670 align:start position:0% its private optimization problem so it 00:06:53.670 --> 00:06:55.880 align:start position:0% its private optimization problem so it solves<00:06:54.480> some<00:06:54.840> problem<00:06:55.320> that<00:06:55.500> doesn't 00:06:55.880 --> 00:06:55.890 align:start position:0% solves some problem that doesn't 00:06:55.890 --> 00:06:58.220 align:start position:0% solves some problem that doesn't actually<00:06:56.010> reflect<00:06:56.580> reality<00:06:56.760> in<00:06:57.570> a<00:06:57.720> machine 00:06:58.220 --> 00:06:58.230 align:start position:0% actually reflect reality in a machine 00:06:58.230 --> 00:06:59.570 align:start position:0% actually reflect reality in a machine learning<00:06:58.410> or<00:06:58.680> data<00:06:58.860> science<00:06:59.190> problem<00:06:59.490> this 00:06:59.570 --> 00:06:59.580 align:start position:0% learning or data science problem this 00:06:59.580 --> 00:07:01.610 align:start position:0% learning or data science problem this could<00:06:59.760> be<00:06:59.880> that<00:07:00.090> you're<00:07:00.330> throwing<00:07:01.110> in<00:07:01.290> junk 00:07:01.610 --> 00:07:01.620 align:start position:0% could be that you're throwing in junk 00:07:01.620 --> 00:07:04.820 align:start position:0% could be that you're throwing in junk data<00:07:01.950> and<00:07:02.420> the<00:07:03.420> network<00:07:03.960> doesn't<00:07:04.230> necessarily 00:07:04.820 --> 00:07:04.830 align:start position:0% data and the network doesn't necessarily 00:07:04.830 --> 00:07:07.190 align:start position:0% data and the network doesn't necessarily know<00:07:04.980> that<00:07:05.220> it<00:07:05.460> just<00:07:05.520> sees<00:07:05.940> some<00:07:06.510> message<00:07:06.900> from 00:07:07.190 --> 00:07:07.200 align:start position:0% know that it just sees some message from 00:07:07.200 --> 00:07:08.930 align:start position:0% know that it just sees some message from you<00:07:07.410> it<00:07:08.010> might<00:07:08.250> not<00:07:08.400> have<00:07:08.550> looked<00:07:08.640> been 00:07:08.930 --> 00:07:08.940 align:start position:0% you it might not have looked been 00:07:08.940 --> 00:07:10.700 align:start position:0% you it might not have looked been entirely<00:07:09.330> expecting<00:07:09.870> that<00:07:09.990> message<00:07:10.350> but<00:07:10.560> it 00:07:10.700 --> 00:07:10.710 align:start position:0% entirely expecting that message but it 00:07:10.710 --> 00:07:12.860 align:start position:0% entirely expecting that message but it can't<00:07:10.980> know<00:07:11.370> whether<00:07:11.760> that's<00:07:11.940> because<00:07:12.570> you're 00:07:12.860 --> 00:07:12.870 align:start position:0% can't know whether that's because you're 00:07:12.870 --> 00:07:15.200 align:start position:0% can't know whether that's because you're acting<00:07:13.380> strange<00:07:13.800> or<00:07:14.460> whether<00:07:14.760> that's<00:07:15.030> because 00:07:15.200 --> 00:07:15.210 align:start position:0% acting strange or whether that's because 00:07:15.210 --> 00:07:17.780 align:start position:0% acting strange or whether that's because you're<00:07:15.720> upstream<00:07:16.260> participants<00:07:17.010> have<00:07:17.460> sent 00:07:17.780 --> 00:07:17.790 align:start position:0% you're upstream participants have sent 00:07:17.790 --> 00:07:19.700 align:start position:0% you're upstream participants have sent you<00:07:17.940> bad<00:07:18.150> data<00:07:18.360> so<00:07:18.930> there's<00:07:19.050> difficulty<00:07:19.620> in 00:07:19.700 --> 00:07:19.710 align:start position:0% you bad data so there's difficulty in 00:07:19.710 --> 00:07:22.190 align:start position:0% you bad data so there's difficulty in identifying<00:07:19.800> where<00:07:20.760> the<00:07:20.790> flaw<00:07:21.600> in<00:07:21.660> the<00:07:21.840> system 00:07:22.190 --> 00:07:22.200 align:start position:0% identifying where the flaw in the system 00:07:22.200 --> 00:07:26.990 align:start position:0% identifying where the flaw in the system is<00:07:22.910> also<00:07:23.910> a<00:07:25.130> compromised<00:07:26.130> node<00:07:26.340> can<00:07:26.550> just<00:07:26.730> send 00:07:26.990 --> 00:07:27.000 align:start position:0% is also a compromised node can just send 00:07:27.000 --> 00:07:29.240 align:start position:0% is also a compromised node can just send a<00:07:27.270> bad<00:07:27.630> signal<00:07:28.080> to<00:07:28.350> its<00:07:28.500> neighbors<00:07:28.890> like 00:07:29.240 --> 00:07:29.250 align:start position:0% a bad signal to its neighbors like 00:07:29.250 --> 00:07:31.700 align:start position:0% a bad signal to its neighbors like unidirectionally<00:07:30.210> and<00:07:30.390> if<00:07:31.260> it<00:07:31.440> knows<00:07:31.680> that 00:07:31.700 --> 00:07:31.710 align:start position:0% unidirectionally and if it knows that 00:07:31.710 --> 00:07:34.070 align:start position:0% unidirectionally and if it knows that there's<00:07:32.280> certain<00:07:33.150> types<00:07:33.390> of<00:07:33.600> problems<00:07:33.930> that 00:07:34.070 --> 00:07:34.080 align:start position:0% there's certain types of problems that 00:07:34.080 --> 00:07:36.680 align:start position:0% there's certain types of problems that this<00:07:34.350> node<00:07:34.620> cannot<00:07:35.240> address<00:07:36.240> her<00:07:36.510> like 00:07:36.680 --> 00:07:36.690 align:start position:0% this node cannot address her like 00:07:36.690 --> 00:07:39.410 align:start position:0% this node cannot address her like certain<00:07:37.080> areas<00:07:37.380> where<00:07:37.590> this<00:07:38.400> node<00:07:38.670> would<00:07:39.300> be 00:07:39.410 --> 00:07:39.420 align:start position:0% certain areas where this node would be 00:07:39.420 --> 00:07:41.630 align:start position:0% certain areas where this node would be infeasible<00:07:39.750> it<00:07:40.500> can<00:07:40.740> send<00:07:41.010> a<00:07:41.100> signal<00:07:41.310> that's 00:07:41.630 --> 00:07:41.640 align:start position:0% infeasible it can send a signal that's 00:07:41.640 --> 00:07:43.700 align:start position:0% infeasible it can send a signal that's always<00:07:41.910> going<00:07:42.120> to<00:07:42.150> be<00:07:42.240> infeasible<00:07:42.570> this<00:07:43.470> node 00:07:43.700 --> 00:07:43.710 align:start position:0% always going to be infeasible this node 00:07:43.710 --> 00:07:45.770 align:start position:0% always going to be infeasible this node basically<00:07:44.220> freaks<00:07:44.670> out<00:07:44.820> is<00:07:45.090> never<00:07:45.390> able<00:07:45.540> to 00:07:45.770 --> 00:07:45.780 align:start position:0% basically freaks out is never able to 00:07:45.780 --> 00:07:47.390 align:start position:0% basically freaks out is never able to reach<00:07:45.930> consensus<00:07:46.560> it's<00:07:46.770> never<00:07:46.980> able<00:07:47.160> to<00:07:47.250> reach 00:07:47.390 --> 00:07:47.400 align:start position:0% reach consensus it's never able to reach 00:07:47.400 --> 00:07:49.610 align:start position:0% reach consensus it's never able to reach feasibility<00:07:48.060> and<00:07:48.240> the<00:07:48.750> system<00:07:49.140> fails<00:07:49.440> to 00:07:49.610 --> 00:07:49.620 align:start position:0% feasibility and the system fails to 00:07:49.620 --> 00:07:52.880 align:start position:0% feasibility and the system fails to converge<00:07:50.750> then<00:07:51.750> also<00:07:51.960> a<00:07:52.110> compromised<00:07:52.650> node 00:07:52.880 --> 00:07:52.890 align:start position:0% converge then also a compromised node 00:07:52.890 --> 00:07:54.860 align:start position:0% converge then also a compromised node could<00:07:53.100> just<00:07:53.280> inject<00:07:53.640> noise<00:07:53.970> into<00:07:54.360> the<00:07:54.419> system 00:07:54.860 --> 00:07:54.870 align:start position:0% could just inject noise into the system 00:07:54.870 --> 00:07:57.560 align:start position:0% could just inject noise into the system so<00:07:55.110> that<00:07:55.140> at<00:07:55.770> every<00:07:55.980> update<00:07:56.460> step<00:07:56.760> you<00:07:57.150> have<00:07:57.180> a 00:07:57.560 --> 00:07:57.570 align:start position:0% so that at every update step you have a 00:07:57.570 --> 00:07:59.960 align:start position:0% so that at every update step you have a noisy<00:07:58.080> signal<00:07:58.320> and<00:07:58.710> you're<00:07:59.310> failing<00:07:59.700> to 00:07:59.960 --> 00:07:59.970 align:start position:0% noisy signal and you're failing to 00:07:59.970 --> 00:08:03.320 align:start position:0% noisy signal and you're failing to converge<00:08:00.300> because<00:08:00.720> of<00:08:00.870> that<00:08:02.000> so<00:08:03.000> what<00:08:03.120> can<00:08:03.270> we 00:08:03.320 --> 00:08:03.330 align:start position:0% converge because of that so what can we 00:08:03.330 --> 00:08:05.780 align:start position:0% converge because of that so what can we do<00:08:03.540> to<00:08:03.750> prevent<00:08:04.050> these<00:08:04.320> modes<00:08:04.560> of<00:08:04.680> attack<00:08:04.950> what 00:08:05.780 --> 00:08:05.790 align:start position:0% do to prevent these modes of attack what 00:08:05.790 --> 00:08:09.170 align:start position:0% do to prevent these modes of attack what we'd<00:08:05.910> like<00:08:06.120> to<00:08:06.180> do<00:08:06.510> is<00:08:06.750> perhaps<00:08:07.140> bypass<00:08:08.180> any 00:08:09.170 --> 00:08:09.180 align:start position:0% we'd like to do is perhaps bypass any 00:08:09.180 --> 00:08:11.540 align:start position:0% we'd like to do is perhaps bypass any node<00:08:09.630> that<00:08:09.840> might<00:08:10.080> be<00:08:10.200> compromised<00:08:10.740> if<00:08:11.040> you're 00:08:11.540 --> 00:08:11.550 align:start position:0% node that might be compromised if you're 00:08:11.550 --> 00:08:14.000 align:start position:0% node that might be compromised if you're trying<00:08:11.880> to<00:08:11.970> make<00:08:12.930> your<00:08:13.080> system<00:08:13.410> redundant<00:08:13.919> or 00:08:14.000 --> 00:08:14.010 align:start position:0% trying to make your system redundant or 00:08:14.010 --> 00:08:16.100 align:start position:0% trying to make your system redundant or resilient<00:08:14.850> to<00:08:14.970> attacks<00:08:15.300> up<00:08:15.630> to<00:08:15.750> end<00:08:15.900> nodes 00:08:16.100 --> 00:08:16.110 align:start position:0% resilient to attacks up to end nodes 00:08:16.110 --> 00:08:18.110 align:start position:0% resilient to attacks up to end nodes this<00:08:16.710> ultimately<00:08:17.250> would<00:08:17.460> mean<00:08:17.610> that<00:08:17.850> you<00:08:17.940> need 00:08:18.110 --> 00:08:18.120 align:start position:0% this ultimately would mean that you need 00:08:18.120 --> 00:08:19.909 align:start position:0% this ultimately would mean that you need a<00:08:18.150> fully<00:08:18.540> connected<00:08:18.810> graph<00:08:19.169> with<00:08:19.590> secure 00:08:19.909 --> 00:08:19.919 align:start position:0% a fully connected graph with secure 00:08:19.919 --> 00:08:22.100 align:start position:0% a fully connected graph with secure communications<00:08:20.669> channels<00:08:21.060> the<00:08:21.840> other<00:08:21.960> way 00:08:22.100 --> 00:08:22.110 align:start position:0% communications channels the other way 00:08:22.110 --> 00:08:23.510 align:start position:0% communications channels the other way that<00:08:22.140> we<00:08:22.290> could<00:08:22.470> address<00:08:22.620> this<00:08:22.890> would<00:08:23.070> be<00:08:23.160> have 00:08:23.510 --> 00:08:23.520 align:start position:0% that we could address this would be have 00:08:23.520 --> 00:08:26.210 align:start position:0% that we could address this would be have some<00:08:23.660> other<00:08:24.660> centralized<00:08:25.320> database<00:08:25.980> that 00:08:26.210 --> 00:08:26.220 align:start position:0% some other centralized database that 00:08:26.220 --> 00:08:28.310 align:start position:0% some other centralized database that provides<00:08:26.550> a<00:08:26.940> global<00:08:27.360> information<00:08:27.630> layer<00:08:28.050> on 00:08:28.310 --> 00:08:28.320 align:start position:0% provides a global information layer on 00:08:28.320 --> 00:08:30.920 align:start position:0% provides a global information layer on all<00:08:28.500> the<00:08:28.680> state<00:08:28.890> variables<00:08:29.400> now<00:08:30.330> this<00:08:30.750> is 00:08:30.920 --> 00:08:30.930 align:start position:0% all the state variables now this is 00:08:30.930 --> 00:08:32.480 align:start position:0% all the state variables now this is starting<00:08:31.200> to<00:08:31.260> look<00:08:31.380> a<00:08:31.530> little<00:08:31.710> bit<00:08:31.980> more<00:08:32.219> like 00:08:32.480 --> 00:08:32.490 align:start position:0% starting to look a little bit more like 00:08:32.490 --> 00:08:35.150 align:start position:0% starting to look a little bit more like a<00:08:32.520> blockchain<00:08:32.820> so<00:08:33.660> taking<00:08:34.560> that<00:08:34.919> information 00:08:35.150 --> 00:08:35.160 align:start position:0% a blockchain so taking that information 00:08:35.160 --> 00:08:38.959 align:start position:0% a blockchain so taking that information from<00:08:36.120> a<00:08:36.960> peer-to-peer<00:08:37.760> communication<00:08:38.760> and 00:08:38.959 --> 00:08:38.969 align:start position:0% from a peer-to-peer communication and 00:08:38.969 --> 00:08:41.480 align:start position:0% from a peer-to-peer communication and having<00:08:39.360> that<00:08:39.570> passed<00:08:40.289> trust<00:08:40.800> lessly<00:08:41.039> between 00:08:41.480 --> 00:08:41.490 align:start position:0% having that passed trust lessly between 00:08:41.490 --> 00:08:43.730 align:start position:0% having that passed trust lessly between all<00:08:41.729> the<00:08:41.910> peers<00:08:42.180> on<00:08:42.330> the<00:08:42.360> network<00:08:42.690> gives<00:08:43.530> us 00:08:43.730 --> 00:08:43.740 align:start position:0% all the peers on the network gives us 00:08:43.740 --> 00:08:46.040 align:start position:0% all the peers on the network gives us global<00:08:44.280> insight<00:08:44.700> into<00:08:45.030> the 00:08:46.040 --> 00:08:46.050 align:start position:0% global insight into the 00:08:46.050 --> 00:08:47.540 align:start position:0% global insight into the state<00:08:46.320> of<00:08:46.560> all<00:08:46.830> the<00:08:46.860> variables<00:08:47.279> in<00:08:47.459> the 00:08:47.540 --> 00:08:47.550 align:start position:0% state of all the variables in the 00:08:47.550 --> 00:08:48.980 align:start position:0% state of all the variables in the network<00:08:47.820> this<00:08:48.149> allows<00:08:48.480> us<00:08:48.690> to<00:08:48.810> perform 00:08:48.980 --> 00:08:48.990 align:start position:0% network this allows us to perform 00:08:48.990 --> 00:08:51.530 align:start position:0% network this allows us to perform security<00:08:49.620> checks<00:08:50.100> on<00:08:50.459> the<00:08:51.060> signals<00:08:51.480> that 00:08:51.530 --> 00:08:51.540 align:start position:0% security checks on the signals that 00:08:51.540 --> 00:08:53.000 align:start position:0% security checks on the signals that we're<00:08:51.720> receiving<00:08:51.959> from<00:08:52.260> our<00:08:52.380> neighbors<00:08:52.740> and 00:08:53.000 --> 00:08:53.010 align:start position:0% we're receiving from our neighbors and 00:08:53.010 --> 00:08:55.220 align:start position:0% we're receiving from our neighbors and thus<00:08:53.580> identify<00:08:54.269> the<00:08:54.330> nodes<00:08:54.779> that<00:08:54.930> have<00:08:55.110> been 00:08:55.220 --> 00:08:55.230 align:start position:0% thus identify the nodes that have been 00:08:55.230 --> 00:08:57.320 align:start position:0% thus identify the nodes that have been compromised<00:08:55.740> this<00:08:56.550> is<00:08:56.700> really<00:08:56.910> powerful 00:08:57.320 --> 00:08:57.330 align:start position:0% compromised this is really powerful 00:08:57.330 --> 00:08:59.420 align:start position:0% compromised this is really powerful because<00:08:57.450> it<00:08:57.660> means<00:08:57.810> that<00:08:57.899> we<00:08:58.170> no<00:08:58.380> longer<00:08:58.430> have 00:08:59.420 --> 00:08:59.430 align:start position:0% because it means that we no longer have 00:08:59.430 --> 00:09:01.639 align:start position:0% because it means that we no longer have to<00:08:59.459> trust<00:08:59.910> either<00:09:00.180> an<00:09:00.540> aggregator<00:09:01.080> somebody 00:09:01.639 --> 00:09:01.649 align:start position:0% to trust either an aggregator somebody 00:09:01.649 --> 00:09:04.490 align:start position:0% to trust either an aggregator somebody who's<00:09:01.769> holding<00:09:02.100> the<00:09:02.160> database<00:09:02.579> or<00:09:03.450> just<00:09:04.260> trust 00:09:04.490 --> 00:09:04.500 align:start position:0% who's holding the database or just trust 00:09:04.500 --> 00:09:06.650 align:start position:0% who's holding the database or just trust our<00:09:04.620> neighbors<00:09:05.010> but<00:09:05.519> we<00:09:05.579> get<00:09:05.850> the<00:09:06.029> benefits<00:09:06.450> of 00:09:06.650 --> 00:09:06.660 align:start position:0% our neighbors but we get the benefits of 00:09:06.660 --> 00:09:10.970 align:start position:0% our neighbors but we get the benefits of both<00:09:06.750> of<00:09:07.170> those<00:09:08.930> so<00:09:09.930> one<00:09:10.560> of<00:09:10.589> the<00:09:10.800> things 00:09:10.970 --> 00:09:10.980 align:start position:0% both of those so one of the things 00:09:10.980 --> 00:09:12.290 align:start position:0% both of those so one of the things that's<00:09:11.190> really<00:09:11.250> interesting<00:09:11.430> about<00:09:11.880> this<00:09:12.060> is 00:09:12.290 --> 00:09:12.300 align:start position:0% that's really interesting about this is 00:09:12.300 --> 00:09:13.880 align:start position:0% that's really interesting about this is we<00:09:12.540> think<00:09:12.750> of<00:09:12.870> block<00:09:13.050> chains<00:09:13.350> as<00:09:13.589> these 00:09:13.880 --> 00:09:13.890 align:start position:0% we think of block chains as these 00:09:13.890 --> 00:09:16.670 align:start position:0% we think of block chains as these iterative<00:09:14.870> computations<00:09:15.870> that<00:09:16.170> reach 00:09:16.670 --> 00:09:16.680 align:start position:0% iterative computations that reach 00:09:16.680 --> 00:09:18.860 align:start position:0% iterative computations that reach consensus<00:09:17.339> also<00:09:18.240> decentralized 00:09:18.860 --> 00:09:18.870 align:start position:0% consensus also decentralized 00:09:18.870 --> 00:09:21.889 align:start position:0% consensus also decentralized optimization<00:09:19.649> also<00:09:20.310> is<00:09:20.550> this<00:09:21.120> same<00:09:21.390> model<00:09:21.779> of 00:09:21.889 --> 00:09:21.899 align:start position:0% optimization also is this same model of 00:09:21.899 --> 00:09:24.710 align:start position:0% optimization also is this same model of iterating<00:09:22.560> between<00:09:22.589> local<00:09:23.310> problems<00:09:23.790> with<00:09:24.570> an 00:09:24.710 --> 00:09:24.720 align:start position:0% iterating between local problems with an 00:09:24.720 --> 00:09:26.509 align:start position:0% iterating between local problems with an aggregator<00:09:25.170> update<00:09:25.529> step<00:09:25.800> or<00:09:26.040> at<00:09:26.160> the<00:09:26.250> rest<00:09:26.430> of 00:09:26.509 --> 00:09:26.519 align:start position:0% aggregator update step or at the rest of 00:09:26.519 --> 00:09:28.190 align:start position:0% aggregator update step or at the rest of the<00:09:26.579> network<00:09:26.730> and<00:09:27.089> reaching<00:09:27.810> eventually 00:09:28.190 --> 00:09:28.200 align:start position:0% the network and reaching eventually 00:09:28.200 --> 00:09:31.190 align:start position:0% the network and reaching eventually consensus<00:09:28.769> on<00:09:28.920> the<00:09:29.010> system<00:09:29.600> the<00:09:30.600> research 00:09:31.190 --> 00:09:31.200 align:start position:0% consensus on the system the research 00:09:31.200 --> 00:09:32.660 align:start position:0% consensus on the system the research that<00:09:31.380> I've<00:09:31.470> been<00:09:31.500> doing<00:09:31.620> is<00:09:32.010> highlighting<00:09:32.550> the 00:09:32.660 --> 00:09:32.670 align:start position:0% that I've been doing is highlighting the 00:09:32.670 --> 00:09:35.240 align:start position:0% that I've been doing is highlighting the similarity<00:09:33.240> to<00:09:33.360> be<00:09:33.450> able<00:09:33.540> to<00:09:33.630> create<00:09:33.899> a<00:09:34.589> sort 00:09:35.240 --> 00:09:35.250 align:start position:0% similarity to be able to create a sort 00:09:35.250 --> 00:09:37.220 align:start position:0% similarity to be able to create a sort of<00:09:35.279> combined<00:09:35.700> model<00:09:36.149> where<00:09:36.480> these<00:09:36.779> local 00:09:37.220 --> 00:09:37.230 align:start position:0% of combined model where these local 00:09:37.230 --> 00:09:39.470 align:start position:0% of combined model where these local update<00:09:37.529> steps<00:09:37.860> are<00:09:38.459> coordinating<00:09:39.240> with<00:09:39.450> a 00:09:39.470 --> 00:09:39.480 align:start position:0% update steps are coordinating with a 00:09:39.480 --> 00:09:41.840 align:start position:0% update steps are coordinating with a smart<00:09:39.899> contract<00:09:40.079> to<00:09:40.680> be<00:09:40.769> able<00:09:40.829> to<00:09:40.950> create<00:09:41.579> or 00:09:41.840 --> 00:09:41.850 align:start position:0% smart contract to be able to create or 00:09:41.850 --> 00:09:44.420 align:start position:0% smart contract to be able to create or to<00:09:42.149> process<00:09:42.720> a<00:09:42.750> secure<00:09:43.290> verifiable<00:09:44.100> update 00:09:44.420 --> 00:09:44.430 align:start position:0% to process a secure verifiable update 00:09:44.430 --> 00:09:47.449 align:start position:0% to process a secure verifiable update step<00:09:44.700> ultimately<00:09:45.690> reaching<00:09:46.200> a<00:09:46.230> solution<00:09:46.740> that 00:09:47.449 --> 00:09:47.459 align:start position:0% step ultimately reaching a solution that 00:09:47.459 --> 00:09:49.910 align:start position:0% step ultimately reaching a solution that is<00:09:47.820> auditable<00:09:48.450> you<00:09:48.930> can<00:09:49.079> see<00:09:49.260> how<00:09:49.440> you<00:09:49.500> got<00:09:49.769> to 00:09:49.910 --> 00:09:49.920 align:start position:0% is auditable you can see how you got to 00:09:49.920 --> 00:09:51.470 align:start position:0% is auditable you can see how you got to that<00:09:50.010> solution<00:09:50.310> you<00:09:50.640> can<00:09:51.089> prove<00:09:51.329> the 00:09:51.470 --> 00:09:51.480 align:start position:0% that solution you can prove the 00:09:51.480 --> 00:09:53.750 align:start position:0% that solution you can prove the optimality<00:09:51.630> of<00:09:52.170> that<00:09:52.320> and<00:09:52.529> know<00:09:53.070> whether<00:09:53.610> or 00:09:53.750 --> 00:09:53.760 align:start position:0% optimality of that and know whether or 00:09:53.760 --> 00:09:55.460 align:start position:0% optimality of that and know whether or not<00:09:53.790> there<00:09:54.029> were<00:09:54.120> attacks<00:09:54.420> along<00:09:54.930> the<00:09:55.110> way<00:09:55.230> and 00:09:55.460 --> 00:09:55.470 align:start position:0% not there were attacks along the way and 00:09:55.470 --> 00:09:57.380 align:start position:0% not there were attacks along the way and then<00:09:56.070> that<00:09:56.220> solution<00:09:56.430> can<00:09:56.820> be<00:09:56.850> stored<00:09:57.180> on<00:09:57.270> the 00:09:57.380 --> 00:09:57.390 align:start position:0% then that solution can be stored on the 00:09:57.390 --> 00:09:59.980 align:start position:0% then that solution can be stored on the blockchain<00:09:57.600> for<00:09:58.020> everybody<00:09:58.350> to<00:09:58.500> access 00:09:59.980 --> 00:09:59.990 align:start position:0% blockchain for everybody to access 00:09:59.990 --> 00:10:01.730 align:start position:0% blockchain for everybody to access thinking<00:10:00.990> about<00:10:01.140> a<00:10:01.170> couple<00:10:01.560> of<00:10:01.620> the<00:10:01.709> different 00:10:01.730 --> 00:10:01.740 align:start position:0% thinking about a couple of the different 00:10:01.740 --> 00:10:04.040 align:start position:0% thinking about a couple of the different ways<00:10:02.190> that<00:10:02.430> these<00:10:02.579> different<00:10:03.180> models<00:10:03.329> fail<00:10:03.720> so 00:10:04.040 --> 00:10:04.050 align:start position:0% ways that these different models fail so 00:10:04.050 --> 00:10:05.810 align:start position:0% ways that these different models fail so we<00:10:04.170> first<00:10:04.440> talked<00:10:04.680> about<00:10:04.820> aggregator 00:10:05.810 --> 00:10:05.820 align:start position:0% we first talked about aggregator 00:10:05.820 --> 00:10:07.990 align:start position:0% we first talked about aggregator coordinated<00:10:06.510> decentralized<00:10:07.200> optimization 00:10:07.990 --> 00:10:08.000 align:start position:0% coordinated decentralized optimization 00:10:08.000 --> 00:10:10.639 align:start position:0% coordinated decentralized optimization prone<00:10:09.000> to<00:10:09.240> monopolies<00:10:09.810> distortions<00:10:10.380> so 00:10:10.639 --> 00:10:10.649 align:start position:0% prone to monopolies distortions so 00:10:10.649 --> 00:10:12.319 align:start position:0% prone to monopolies distortions so anytime<00:10:10.920> that<00:10:11.130> there's<00:10:11.459> a<00:10:11.520> market<00:10:11.880> operator<00:10:12.240> a 00:10:12.319 --> 00:10:12.329 align:start position:0% anytime that there's a market operator a 00:10:12.329 --> 00:10:14.300 align:start position:0% anytime that there's a market operator a scheduler<00:10:12.930> who<00:10:13.140> has<00:10:13.320> their<00:10:13.529> own<00:10:13.560> incentives 00:10:14.300 --> 00:10:14.310 align:start position:0% scheduler who has their own incentives 00:10:14.310 --> 00:10:16.160 align:start position:0% scheduler who has their own incentives and<00:10:14.459> you<00:10:14.970> may<00:10:15.089> not<00:10:15.270> trust<00:10:15.480> their<00:10:15.690> incentives 00:10:16.160 --> 00:10:16.170 align:start position:0% and you may not trust their incentives 00:10:16.170 --> 00:10:17.900 align:start position:0% and you may not trust their incentives you<00:10:16.740> wouldn't<00:10:16.980> want<00:10:17.130> to<00:10:17.190> trust<00:10:17.579> some 00:10:17.900 --> 00:10:17.910 align:start position:0% you wouldn't want to trust some 00:10:17.910 --> 00:10:20.269 align:start position:0% you wouldn't want to trust some monopolies<00:10:18.420> aggregator<00:10:18.870> they're<00:10:19.860> prone<00:10:20.130> to 00:10:20.269 --> 00:10:20.279 align:start position:0% monopolies aggregator they're prone to 00:10:20.279 --> 00:10:22.490 align:start position:0% monopolies aggregator they're prone to communication<00:10:20.850> dropouts<00:10:21.270> there<00:10:21.540> if<00:10:22.170> there's 00:10:22.490 --> 00:10:22.500 align:start position:0% communication dropouts there if there's 00:10:22.500 --> 00:10:25.130 align:start position:0% communication dropouts there if there's a<00:10:22.709> an<00:10:23.100> outage<00:10:23.670> in<00:10:23.820> the<00:10:23.850> electricity<00:10:24.390> grid<00:10:24.570> that 00:10:25.130 --> 00:10:25.140 align:start position:0% a an outage in the electricity grid that 00:10:25.140 --> 00:10:27.139 align:start position:0% a an outage in the electricity grid that monopolies<00:10:25.980> or<00:10:26.250> that<00:10:26.550> centralized 00:10:27.139 --> 00:10:27.149 align:start position:0% monopolies or that centralized 00:10:27.149 --> 00:10:29.630 align:start position:0% monopolies or that centralized aggregator<00:10:27.660> wouldn't<00:10:28.470> be<00:10:28.620> online<00:10:28.920> wouldn't 00:10:29.630 --> 00:10:29.640 align:start position:0% aggregator wouldn't be online wouldn't 00:10:29.640 --> 00:10:30.829 align:start position:0% aggregator wouldn't be online wouldn't be<00:10:29.699> able<00:10:29.790> to<00:10:29.880> coordinate<00:10:30.149> between<00:10:30.540> these 00:10:30.829 --> 00:10:30.839 align:start position:0% be able to coordinate between these 00:10:30.839 --> 00:10:33.920 align:start position:0% be able to coordinate between these local<00:10:31.170> nodes<00:10:31.670> conversely<00:10:32.670> the<00:10:33.029> fully 00:10:33.920 --> 00:10:33.930 align:start position:0% local nodes conversely the fully 00:10:33.930 --> 00:10:36.139 align:start position:0% local nodes conversely the fully decentralized<00:10:34.079> models<00:10:34.829> addressed<00:10:35.250> those<00:10:35.550> but 00:10:36.139 --> 00:10:36.149 align:start position:0% decentralized models addressed those but 00:10:36.149 --> 00:10:38.720 align:start position:0% decentralized models addressed those but had<00:10:36.750> exposed<00:10:37.320> us<00:10:37.560> to<00:10:37.709> these<00:10:37.860> other<00:10:38.130> weaknesses 00:10:38.720 --> 00:10:38.730 align:start position:0% had exposed us to these other weaknesses 00:10:38.730 --> 00:10:41.360 align:start position:0% had exposed us to these other weaknesses and<00:10:39.329> by<00:10:40.050> using<00:10:40.260> a<00:10:40.440> blockchain<00:10:40.890> to<00:10:41.010> provide<00:10:41.339> a 00:10:41.360 --> 00:10:41.370 align:start position:0% and by using a blockchain to provide a 00:10:41.370 --> 00:10:43.370 align:start position:0% and by using a blockchain to provide a global<00:10:41.970> information<00:10:42.240> layer<00:10:42.690> between<00:10:43.170> all<00:10:43.350> of 00:10:43.370 --> 00:10:43.380 align:start position:0% global information layer between all of 00:10:43.380 --> 00:10:45.439 align:start position:0% global information layer between all of these<00:10:43.589> we're<00:10:44.220> able<00:10:44.310> to<00:10:44.459> have<00:10:44.610> the<00:10:44.790> benefits<00:10:45.329> of 00:10:45.439 --> 00:10:45.449 align:start position:0% these we're able to have the benefits of 00:10:45.449 --> 00:10:48.439 align:start position:0% these we're able to have the benefits of the<00:10:46.050> centralized<00:10:46.560> aggregator<00:10:47.100> problem<00:10:47.550> while 00:10:48.439 --> 00:10:48.449 align:start position:0% the centralized aggregator problem while 00:10:48.449 --> 00:10:50.630 align:start position:0% the centralized aggregator problem while also<00:10:48.690> avoiding<00:10:49.079> the<00:10:49.560> pitfalls<00:10:49.740> of<00:10:50.250> the<00:10:50.310> fully 00:10:50.630 --> 00:10:50.640 align:start position:0% also avoiding the pitfalls of the fully 00:10:50.640 --> 00:10:54.170 align:start position:0% also avoiding the pitfalls of the fully decentralized<00:10:50.850> problem<00:10:52.459> so<00:10:53.459> let's<00:10:53.640> take<00:10:54.089> an 00:10:54.170 --> 00:10:54.180 align:start position:0% decentralized problem so let's take an 00:10:54.180 --> 00:10:56.990 align:start position:0% decentralized problem so let's take an example<00:10:54.380> this<00:10:55.380> is<00:10:55.589> where<00:10:56.040> my<00:10:56.220> research<00:10:56.579> is<00:10:56.760> in 00:10:56.990 --> 00:10:57.000 align:start position:0% example this is where my research is in 00:10:57.000 --> 00:10:59.660 align:start position:0% example this is where my research is in electricity<00:10:57.390> markets<00:10:58.160> and 00:10:59.660 --> 00:10:59.670 align:start position:0% electricity markets and 00:10:59.670 --> 00:11:01.460 align:start position:0% electricity markets and this<00:11:00.030> may<00:11:00.180> be<00:11:00.240> a<00:11:00.330> feel<00:11:00.540> that's<00:11:00.750> unfamiliar<00:11:01.260> to 00:11:01.460 --> 00:11:01.470 align:start position:0% this may be a feel that's unfamiliar to 00:11:01.470 --> 00:11:04.070 align:start position:0% this may be a feel that's unfamiliar to a<00:11:01.500> lot<00:11:01.650> of<00:11:01.770> you<00:11:01.860> but<00:11:02.430> as<00:11:02.670> we<00:11:03.000> look<00:11:03.240> ahead<00:11:03.630> at 00:11:04.070 --> 00:11:04.080 align:start position:0% a lot of you but as we look ahead at 00:11:04.080 --> 00:11:06.380 align:start position:0% a lot of you but as we look ahead at larger<00:11:04.500> penetrations<00:11:05.100> of<00:11:05.310> electric<00:11:05.940> vehicles 00:11:06.380 --> 00:11:06.390 align:start position:0% larger penetrations of electric vehicles 00:11:06.390 --> 00:11:08.900 align:start position:0% larger penetrations of electric vehicles and<00:11:06.540> photovoltaic<00:11:06.930> solar<00:11:07.620> systems<00:11:08.160> we'll<00:11:08.760> end 00:11:08.900 --> 00:11:08.910 align:start position:0% and photovoltaic solar systems we'll end 00:11:08.910 --> 00:11:11.240 align:start position:0% and photovoltaic solar systems we'll end up<00:11:09.030> with<00:11:09.090> these<00:11:09.360> big<00:11:09.630> swings<00:11:10.260> and<00:11:10.650> power<00:11:10.860> flows 00:11:11.240 --> 00:11:11.250 align:start position:0% up with these big swings and power flows 00:11:11.250 --> 00:11:14.360 align:start position:0% up with these big swings and power flows on<00:11:11.580> the<00:11:11.880> local<00:11:12.240> distribution<00:11:13.020> network<00:11:13.620> so 00:11:14.360 --> 00:11:14.370 align:start position:0% on the local distribution network so 00:11:14.370 --> 00:11:17.120 align:start position:0% on the local distribution network so going<00:11:14.850> from<00:11:15.210> a<00:11:15.540> model<00:11:16.020> where<00:11:16.350> you<00:11:16.620> can<00:11:16.650> always 00:11:17.120 --> 00:11:17.130 align:start position:0% going from a model where you can always 00:11:17.130 --> 00:11:19.220 align:start position:0% going from a model where you can always plan<00:11:17.490> for<00:11:17.940> the<00:11:18.000> power<00:11:18.180> consumption<00:11:18.690> because 00:11:19.220 --> 00:11:19.230 align:start position:0% plan for the power consumption because 00:11:19.230 --> 00:11:21.290 align:start position:0% plan for the power consumption because you<00:11:19.350> know<00:11:19.560> that<00:11:20.280> people<00:11:20.340> just<00:11:20.820> have<00:11:20.970> small 00:11:21.290 --> 00:11:21.300 align:start position:0% you know that people just have small 00:11:21.300 --> 00:11:23.180 align:start position:0% you know that people just have small loads<00:11:21.780> like<00:11:22.050> toasters<00:11:22.560> and<00:11:22.590> electric<00:11:23.040> light 00:11:23.180 --> 00:11:23.190 align:start position:0% loads like toasters and electric light 00:11:23.190 --> 00:11:26.360 align:start position:0% loads like toasters and electric light bulbs<00:11:23.570> to<00:11:24.570> a<00:11:24.600> world<00:11:24.840> where<00:11:25.170> you<00:11:25.290> have<00:11:25.500> the<00:11:26.130> Sun 00:11:26.360 --> 00:11:26.370 align:start position:0% bulbs to a world where you have the Sun 00:11:26.370 --> 00:11:28.550 align:start position:0% bulbs to a world where you have the Sun might<00:11:26.580> go<00:11:27.330> behind<00:11:27.600> a<00:11:27.690> cloud<00:11:27.930> everybody's 00:11:28.550 --> 00:11:28.560 align:start position:0% might go behind a cloud everybody's 00:11:28.560 --> 00:11:31.610 align:start position:0% might go behind a cloud everybody's photovoltaic<00:11:29.430> generation<00:11:30.390> cuts<00:11:30.750> off<00:11:30.930> at<00:11:31.470> the 00:11:31.610 --> 00:11:31.620 align:start position:0% photovoltaic generation cuts off at the 00:11:31.620 --> 00:11:33.260 align:start position:0% photovoltaic generation cuts off at the same<00:11:31.800> time<00:11:32.040> people's<00:11:32.430> electric<00:11:32.760> vehicles<00:11:33.120> are 00:11:33.260 --> 00:11:33.270 align:start position:0% same time people's electric vehicles are 00:11:33.270 --> 00:11:34.910 align:start position:0% same time people's electric vehicles are plugging<00:11:33.630> in<00:11:33.750> you<00:11:34.380> have<00:11:34.470> these<00:11:34.620> massive 00:11:34.910 --> 00:11:34.920 align:start position:0% plugging in you have these massive 00:11:34.920 --> 00:11:37.550 align:start position:0% plugging in you have these massive imbalances<00:11:35.730> and<00:11:35.850> generation<00:11:36.560> utilities 00:11:37.550 --> 00:11:37.560 align:start position:0% imbalances and generation utilities 00:11:37.560 --> 00:11:40.370 align:start position:0% imbalances and generation utilities would<00:11:37.710> like<00:11:37.890> to<00:11:38.070> be<00:11:38.220> able<00:11:38.310> to<00:11:39.050> address<00:11:40.050> these 00:11:40.370 --> 00:11:40.380 align:start position:0% would like to be able to address these 00:11:40.380 --> 00:11:43.040 align:start position:0% would like to be able to address these imbalances<00:11:40.830> in<00:11:41.460> a<00:11:41.610> way<00:11:41.730> that<00:11:42.030> minimizes<00:11:42.510> the 00:11:43.040 --> 00:11:43.050 align:start position:0% imbalances in a way that minimizes the 00:11:43.050 --> 00:11:45.530 align:start position:0% imbalances in a way that minimizes the cost<00:11:43.080> of<00:11:43.530> electricity<00:11:43.920> generation<00:11:44.460> that<00:11:45.270> cost 00:11:45.530 --> 00:11:45.540 align:start position:0% cost of electricity generation that cost 00:11:45.540 --> 00:11:47.090 align:start position:0% cost of electricity generation that cost minimization<00:11:45.960> means<00:11:46.740> that<00:11:46.890> we're<00:11:47.010> dealing 00:11:47.090 --> 00:11:47.100 align:start position:0% minimization means that we're dealing 00:11:47.100 --> 00:11:50.120 align:start position:0% minimization means that we're dealing with<00:11:47.280> an<00:11:47.460> optimization<00:11:48.090> problem<00:11:48.120> here<00:11:49.130> the 00:11:50.120 --> 00:11:50.130 align:start position:0% with an optimization problem here the 00:11:50.130 --> 00:11:51.740 align:start position:0% with an optimization problem here the other<00:11:50.610> challenge<00:11:51.150> though<00:11:51.330> is<00:11:51.450> that<00:11:51.480> the 00:11:51.740 --> 00:11:51.750 align:start position:0% other challenge though is that the 00:11:51.750 --> 00:11:53.540 align:start position:0% other challenge though is that the electricity<00:11:52.380> network<00:11:52.680> is<00:11:52.830> constrained<00:11:53.340> we 00:11:53.540 --> 00:11:53.550 align:start position:0% electricity network is constrained we 00:11:53.550 --> 00:11:56.450 align:start position:0% electricity network is constrained we can't<00:11:53.850> just<00:11:54.150> have<00:11:54.630> some<00:11:54.960> generators<00:11:55.350> off<00:11:55.710> a 00:11:56.450 --> 00:11:56.460 align:start position:0% can't just have some generators off a 00:11:56.460 --> 00:11:57.830 align:start position:0% can't just have some generators off a hundred<00:11:56.910> miles<00:11:57.030> away<00:11:57.210> that<00:11:57.330> are<00:11:57.660> going<00:11:57.750> to 00:11:57.830 --> 00:11:57.840 align:start position:0% hundred miles away that are going to 00:11:57.840 --> 00:11:59.900 align:start position:0% hundred miles away that are going to make<00:11:58.140> up<00:11:58.320> for<00:11:58.560> this<00:11:58.680> we<00:11:59.310> need<00:11:59.430> to<00:11:59.580> be<00:11:59.670> able<00:11:59.760> to 00:11:59.900 --> 00:11:59.910 align:start position:0% make up for this we need to be able to 00:11:59.910 --> 00:12:01.970 align:start position:0% make up for this we need to be able to respect<00:12:00.780> the<00:12:00.960> local<00:12:01.350> constraints<00:12:01.710> in<00:12:01.890> the 00:12:01.970 --> 00:12:01.980 align:start position:0% respect the local constraints in the 00:12:01.980 --> 00:12:03.920 align:start position:0% respect the local constraints in the system<00:12:02.340> particularly<00:12:02.760> in<00:12:03.390> what<00:12:03.570> I'm<00:12:03.660> modeling 00:12:03.920 --> 00:12:03.930 align:start position:0% system particularly in what I'm modeling 00:12:03.930 --> 00:12:05.780 align:start position:0% system particularly in what I'm modeling the<00:12:04.470> distribution<00:12:05.100> network<00:12:05.130> within<00:12:05.700> a 00:12:05.780 --> 00:12:05.790 align:start position:0% the distribution network within a 00:12:05.790 --> 00:12:08.360 align:start position:0% the distribution network within a neighborhood<00:12:06.710> one<00:12:07.710> of<00:12:07.800> the<00:12:07.890> challenges 00:12:08.360 --> 00:12:08.370 align:start position:0% neighborhood one of the challenges 00:12:08.370 --> 00:12:10.670 align:start position:0% neighborhood one of the challenges though<00:12:08.580> is<00:12:08.850> that<00:12:09.240> the<00:12:09.540> utilities<00:12:10.050> have<00:12:10.080> market 00:12:10.670 --> 00:12:10.680 align:start position:0% though is that the utilities have market 00:12:10.680 --> 00:12:12.680 align:start position:0% though is that the utilities have market power<00:12:10.710> and<00:12:11.250> all<00:12:11.550> of<00:12:11.730> the<00:12:11.910> people<00:12:12.210> who<00:12:12.330> provide 00:12:12.680 --> 00:12:12.690 align:start position:0% power and all of the people who provide 00:12:12.690 --> 00:12:14.870 align:start position:0% power and all of the people who provide your<00:12:12.990> smart<00:12:13.260> devices<00:12:13.560> may<00:12:14.220> be<00:12:14.250> competitors 00:12:14.870 --> 00:12:14.880 align:start position:0% your smart devices may be competitors 00:12:14.880 --> 00:12:15.790 align:start position:0% your smart devices may be competitors with<00:12:15.060> each<00:12:15.180> other<00:12:15.360> in<00:12:15.510> the<00:12:15.570> marketplace 00:12:15.790 --> 00:12:15.800 align:start position:0% with each other in the marketplace 00:12:15.800 --> 00:12:18.880 align:start position:0% with each other in the marketplace there's<00:12:16.800> a<00:12:16.890> lack<00:12:17.070> of<00:12:17.100> trust<00:12:17.430> not<00:12:18.030> only<00:12:18.090> between 00:12:18.880 --> 00:12:18.890 align:start position:0% there's a lack of trust not only between 00:12:18.890 --> 00:12:21.770 align:start position:0% there's a lack of trust not only between perhaps<00:12:19.890> you<00:12:20.310> and<00:12:20.580> the<00:12:20.700> utility<00:12:21.210> who<00:12:21.420> wants<00:12:21.660> to 00:12:21.770 --> 00:12:21.780 align:start position:0% perhaps you and the utility who wants to 00:12:21.780 --> 00:12:23.960 align:start position:0% perhaps you and the utility who wants to sell<00:12:21.930> you<00:12:22.080> power<00:12:22.290> but<00:12:22.590> also<00:12:23.160> between<00:12:23.460> all<00:12:23.940> of 00:12:23.960 --> 00:12:23.970 align:start position:0% sell you power but also between all of 00:12:23.970 --> 00:12:27.410 align:start position:0% sell you power but also between all of the<00:12:24.210> participants<00:12:24.900> in<00:12:25.400> this<00:12:26.400> marketplace<00:12:26.850> so 00:12:27.410 --> 00:12:27.420 align:start position:0% the participants in this marketplace so 00:12:27.420 --> 00:12:29.330 align:start position:0% the participants in this marketplace so you<00:12:28.170> might<00:12:28.320> have<00:12:28.470> a<00:12:28.530> household<00:12:28.920> that<00:12:29.040> has<00:12:29.310> a 00:12:29.330 --> 00:12:29.340 align:start position:0% you might have a household that has a 00:12:29.340 --> 00:12:31.250 align:start position:0% you might have a household that has a Tesla<00:12:29.880> electric<00:12:30.210> car<00:12:30.390> a<00:12:30.420> Ford<00:12:30.690> electric<00:12:31.110> car 00:12:31.250 --> 00:12:31.260 align:start position:0% Tesla electric car a Ford electric car 00:12:31.260 --> 00:12:32.690 align:start position:0% Tesla electric car a Ford electric car both<00:12:31.590> of<00:12:31.770> those<00:12:31.890> are<00:12:32.130> competing<00:12:32.550> with<00:12:32.580> each 00:12:32.690 --> 00:12:32.700 align:start position:0% both of those are competing with each 00:12:32.700 --> 00:12:35.330 align:start position:0% both of those are competing with each other<00:12:32.940> to<00:12:33.240> sell<00:12:33.810> regulation<00:12:34.410> services<00:12:34.890> on<00:12:35.190> an 00:12:35.330 --> 00:12:35.340 align:start position:0% other to sell regulation services on an 00:12:35.340 --> 00:12:38.930 align:start position:0% other to sell regulation services on an energy<00:12:35.640> market<00:12:36.000> you<00:12:36.750> might<00:12:36.900> have<00:12:37.940> different 00:12:38.930 --> 00:12:38.940 align:start position:0% energy market you might have different 00:12:38.940 --> 00:12:41.030 align:start position:0% energy market you might have different smart<00:12:39.270> thermostats<00:12:39.780> who<00:12:40.050> are<00:12:40.140> competing<00:12:40.470> to 00:12:41.030 --> 00:12:41.040 align:start position:0% smart thermostats who are competing to 00:12:41.040 --> 00:12:44.270 align:start position:0% smart thermostats who are competing to be<00:12:41.160> able<00:12:41.250> to<00:12:42.080> provide<00:12:43.080> energy<00:12:43.560> services<00:12:43.980> on<00:12:44.250> a 00:12:44.270 --> 00:12:44.280 align:start position:0% be able to provide energy services on a 00:12:44.280 --> 00:12:45.890 align:start position:0% be able to provide energy services on a market<00:12:44.730> while<00:12:44.970> also<00:12:45.180> provide<00:12:45.720> these 00:12:45.890 --> 00:12:45.900 align:start position:0% market while also provide these 00:12:45.900 --> 00:12:48.500 align:start position:0% market while also provide these household<00:12:46.410> level<00:12:46.530> services<00:12:47.160> we'd<00:12:48.000> like<00:12:48.210> to<00:12:48.390> be 00:12:48.500 --> 00:12:48.510 align:start position:0% household level services we'd like to be 00:12:48.510 --> 00:12:51.260 align:start position:0% household level services we'd like to be able<00:12:48.630> to<00:12:48.780> reach<00:12:49.650> consensus<00:12:50.310> on<00:12:50.550> the<00:12:50.970> way<00:12:51.090> that 00:12:51.260 --> 00:12:51.270 align:start position:0% able to reach consensus on the way that 00:12:51.270 --> 00:12:54.740 align:start position:0% able to reach consensus on the way that most<00:12:51.540> equitably<00:12:52.200> provides<00:12:53.510> compensation<00:12:54.510> for 00:12:54.740 --> 00:12:54.750 align:start position:0% most equitably provides compensation for 00:12:54.750 --> 00:12:56.690 align:start position:0% most equitably provides compensation for energy<00:12:55.050> services<00:12:55.560> that<00:12:56.220> each<00:12:56.430> of<00:12:56.610> these 00:12:56.690 --> 00:12:56.700 align:start position:0% energy services that each of these 00:12:56.700 --> 00:13:01.130 align:start position:0% energy services that each of these devices<00:12:57.140> gives<00:12:59.090> so<00:13:00.090> taking<00:13:00.810> that<00:13:00.840> same 00:13:01.130 --> 00:13:01.140 align:start position:0% devices gives so taking that same 00:13:01.140 --> 00:13:03.650 align:start position:0% devices gives so taking that same paradigm<00:13:01.620> that<00:13:01.650> I<00:13:01.890> described<00:13:02.280> earlier<00:13:02.660> we're 00:13:03.650 --> 00:13:03.660 align:start position:0% paradigm that I described earlier we're 00:13:03.660 --> 00:13:06.170 align:start position:0% paradigm that I described earlier we're breaking<00:13:04.320> this<00:13:04.500> big<00:13:05.010> scheduling<00:13:05.760> problem 00:13:06.170 --> 00:13:06.180 align:start position:0% breaking this big scheduling problem 00:13:06.180 --> 00:13:08.810 align:start position:0% breaking this big scheduling problem with<00:13:06.360> a<00:13:06.390> bunch<00:13:06.660> of<00:13:06.810> trustless<00:13:07.820> participants 00:13:08.810 --> 00:13:08.820 align:start position:0% with a bunch of trustless participants 00:13:08.820 --> 00:13:11.930 align:start position:0% with a bunch of trustless participants into<00:13:09.810> a<00:13:10.080> set<00:13:10.440> of<00:13:10.560> local<00:13:10.830> problems<00:13:11.460> where<00:13:11.670> each 00:13:11.930 --> 00:13:11.940 align:start position:0% into a set of local problems where each 00:13:11.940 --> 00:13:13.290 align:start position:0% into a set of local problems where each household<00:13:12.540> or<00:13:12.690> each<00:13:12.810> node 00:13:13.290 --> 00:13:13.300 align:start position:0% household or each node 00:13:13.300 --> 00:13:16.380 align:start position:0% household or each node the<00:13:13.420> network<00:13:13.720> as<00:13:13.980> defined<00:13:14.980> by<00:13:15.160> the<00:13:15.940> actual 00:13:16.380 --> 00:13:16.390 align:start position:0% the network as defined by the actual 00:13:16.390 --> 00:13:17.960 align:start position:0% the network as defined by the actual physical<00:13:16.870> constraints<00:13:17.350> of<00:13:17.440> the<00:13:17.589> system 00:13:17.960 --> 00:13:17.970 align:start position:0% physical constraints of the system 00:13:17.970 --> 00:13:20.550 align:start position:0% physical constraints of the system resolves<00:13:18.970> a<00:13:19.029> private<00:13:19.540> optimization<00:13:20.170> problem 00:13:20.550 --> 00:13:20.560 align:start position:0% resolves a private optimization problem 00:13:20.560 --> 00:13:22.620 align:start position:0% resolves a private optimization problem this<00:13:20.740> is<00:13:20.910> harder<00:13:21.910> than<00:13:21.940> you<00:13:22.240> something<00:13:22.570> you 00:13:22.620 --> 00:13:22.630 align:start position:0% this is harder than you something you 00:13:22.630 --> 00:13:24.810 align:start position:0% this is harder than you something you would<00:13:22.750> want<00:13:22.990> to<00:13:23.019> run<00:13:23.260> on<00:13:23.500> the<00:13:23.680> blockchain<00:13:23.890> but 00:13:24.810 --> 00:13:24.820 align:start position:0% would want to run on the blockchain but 00:13:24.820 --> 00:13:26.880 align:start position:0% would want to run on the blockchain but a<00:13:24.850> lot<00:13:25.120> cheaper<00:13:25.390> than<00:13:25.600> solving<00:13:26.350> it<00:13:26.440> all<00:13:26.529> is<00:13:26.680> one 00:13:26.880 --> 00:13:26.890 align:start position:0% a lot cheaper than solving it all is one 00:13:26.890 --> 00:13:29.670 align:start position:0% a lot cheaper than solving it all is one centralized<00:13:27.370> problem<00:13:27.850> those<00:13:28.779> work<00:13:29.470> together 00:13:29.670 --> 00:13:29.680 align:start position:0% centralized problem those work together 00:13:29.680 --> 00:13:31.680 align:start position:0% centralized problem those work together through<00:13:30.399> this<00:13:30.550> a<00:13:30.610> smart<00:13:31.000> contract<00:13:31.240> that 00:13:31.680 --> 00:13:31.690 align:start position:0% through this a smart contract that 00:13:31.690 --> 00:13:34.350 align:start position:0% through this a smart contract that coordinates<00:13:32.230> the<00:13:32.500> update<00:13:32.890> step<00:13:33.160> to<00:13:33.970> be<00:13:34.089> able 00:13:34.350 --> 00:13:34.360 align:start position:0% coordinates the update step to be able 00:13:34.360 --> 00:13:37.050 align:start position:0% coordinates the update step to be able to<00:13:34.510> eventually<00:13:35.110> reach<00:13:35.560> consensus<00:13:36.310> on<00:13:36.459> the 00:13:37.050 --> 00:13:37.060 align:start position:0% to eventually reach consensus on the 00:13:37.060 --> 00:13:39.810 align:start position:0% to eventually reach consensus on the best<00:13:37.360> schedule<00:13:38.230> for<00:13:38.620> all<00:13:38.950> of<00:13:39.399> these<00:13:39.490> devices 00:13:39.810 --> 00:13:39.820 align:start position:0% best schedule for all of these devices 00:13:39.820 --> 00:13:42.660 align:start position:0% best schedule for all of these devices so<00:13:40.690> the<00:13:41.079> goal<00:13:41.260> is<00:13:41.470> to<00:13:41.649> minimize<00:13:41.800> the<00:13:42.370> power 00:13:42.660 --> 00:13:42.670 align:start position:0% so the goal is to minimize the power 00:13:42.670 --> 00:13:45.480 align:start position:0% so the goal is to minimize the power cost<00:13:43.029> for<00:13:43.300> everybody<00:13:43.920> provide<00:13:44.920> compensation 00:13:45.480 --> 00:13:45.490 align:start position:0% cost for everybody provide compensation 00:13:45.490 --> 00:13:47.069 align:start position:0% cost for everybody provide compensation for<00:13:46.000> all<00:13:46.029> of<00:13:46.300> the<00:13:46.420> devices<00:13:46.630> that<00:13:46.870> are 00:13:47.069 --> 00:13:47.079 align:start position:0% for all of the devices that are 00:13:47.079 --> 00:13:49.290 align:start position:0% for all of the devices that are participating<00:13:47.410> in<00:13:47.740> this<00:13:47.860> network<00:13:48.130> and<00:13:48.459> do<00:13:49.120> so 00:13:49.290 --> 00:13:49.300 align:start position:0% participating in this network and do so 00:13:49.300 --> 00:13:52.680 align:start position:0% participating in this network and do so automatically<00:13:50.730> so<00:13:51.730> we<00:13:51.850> have<00:13:52.029> a<00:13:52.060> model<00:13:52.329> of<00:13:52.600> our 00:13:52.680 --> 00:13:52.690 align:start position:0% automatically so we have a model of our 00:13:52.690 --> 00:13:54.600 align:start position:0% automatically so we have a model of our network<00:13:53.019> this<00:13:53.230> is<00:13:53.290> the<00:13:53.440> the<00:13:53.890> physical<00:13:54.070> network 00:13:54.600 --> 00:13:54.610 align:start position:0% network this is the the physical network 00:13:54.610 --> 00:13:56.940 align:start position:0% network this is the the physical network that<00:13:54.820> actually<00:13:55.079> provides<00:13:56.079> these<00:13:56.320> constraints 00:13:56.940 --> 00:13:56.950 align:start position:0% that actually provides these constraints 00:13:56.950 --> 00:14:00.240 align:start position:0% that actually provides these constraints that<00:13:57.790> gets<00:13:58.120> fed<00:13:58.360> into<00:13:58.390> these<00:13:59.290> smart<00:13:59.649> contracts 00:14:00.240 --> 00:14:00.250 align:start position:0% that gets fed into these smart contracts 00:14:00.250 --> 00:14:02.069 align:start position:0% that gets fed into these smart contracts so<00:14:00.490> we<00:14:00.610> have<00:14:00.730> the<00:14:00.880> optimization<00:14:01.480> problem<00:14:01.720> the 00:14:02.069 --> 00:14:02.079 align:start position:0% so we have the optimization problem the 00:14:02.079 --> 00:14:05.759 align:start position:0% so we have the optimization problem the smart<00:14:02.740> meter<00:14:02.950> then<00:14:03.850> those<00:14:04.269> the<00:14:05.140> optimization 00:14:05.759 --> 00:14:05.769 align:start position:0% smart meter then those the optimization 00:14:05.769 --> 00:14:07.259 align:start position:0% smart meter then those the optimization updates<00:14:06.070> get<00:14:06.310> communicated<00:14:06.910> with<00:14:07.000> smart 00:14:07.259 --> 00:14:07.269 align:start position:0% updates get communicated with smart 00:14:07.269 --> 00:14:09.509 align:start position:0% updates get communicated with smart contracts<00:14:07.750> on<00:14:07.839> the<00:14:07.870> blockchain<00:14:08.110> one<00:14:09.070> for 00:14:09.509 --> 00:14:09.519 align:start position:0% contracts on the blockchain one for 00:14:09.519 --> 00:14:11.040 align:start position:0% contracts on the blockchain one for actually<00:14:09.850> coordinating<00:14:10.420> the<00:14:10.510> update<00:14:10.810> step 00:14:11.040 --> 00:14:11.050 align:start position:0% actually coordinating the update step 00:14:11.050 --> 00:14:13.380 align:start position:0% actually coordinating the update step once<00:14:11.860> we<00:14:11.920> reach<00:14:12.160> consensus<00:14:12.820> that<00:14:13.120> can<00:14:13.300> be 00:14:13.380 --> 00:14:13.390 align:start position:0% once we reach consensus that can be 00:14:13.390 --> 00:14:15.300 align:start position:0% once we reach consensus that can be bounced<00:14:13.750> over<00:14:14.050> to<00:14:14.350> a<00:14:14.380> smart<00:14:14.709> contract<00:14:15.160> that 00:14:15.300 --> 00:14:15.310 align:start position:0% bounced over to a smart contract that 00:14:15.310 --> 00:14:18.240 align:start position:0% bounced over to a smart contract that does<00:14:15.880> billing<00:14:16.269> and<00:14:16.709> any<00:14:17.709> reconciliation 00:14:18.240 --> 00:14:18.250 align:start position:0% does billing and any reconciliation 00:14:18.250 --> 00:14:20.370 align:start position:0% does billing and any reconciliation between<00:14:18.760> the<00:14:19.089> expected<00:14:19.600> schedule<00:14:19.990> and<00:14:20.200> the 00:14:20.370 --> 00:14:20.380 align:start position:0% between the expected schedule and the 00:14:20.380 --> 00:14:23.310 align:start position:0% between the expected schedule and the actual<00:14:20.649> realized<00:14:21.040> schedule<00:14:21.520> and<00:14:22.000> so<00:14:22.660> with 00:14:23.310 --> 00:14:23.320 align:start position:0% actual realized schedule and so with 00:14:23.320 --> 00:14:26.730 align:start position:0% actual realized schedule and so with that<00:14:23.940> we're<00:14:24.940> able<00:14:25.060> to<00:14:25.209> eventually<00:14:25.870> identify 00:14:26.730 --> 00:14:26.740 align:start position:0% that we're able to eventually identify 00:14:26.740 --> 00:14:28.860 align:start position:0% that we're able to eventually identify when<00:14:26.980> we<00:14:27.100> pick<00:14:27.279> consensus<00:14:27.880> on<00:14:28.120> the<00:14:28.540> global 00:14:28.860 --> 00:14:28.870 align:start position:0% when we pick consensus on the global 00:14:28.870 --> 00:14:30.900 align:start position:0% when we pick consensus on the global optimum<00:14:29.320> we<00:14:29.770> can<00:14:29.890> guarantee<00:14:30.370> the<00:14:30.550> optimality 00:14:30.900 --> 00:14:30.910 align:start position:0% optimum we can guarantee the optimality 00:14:30.910 --> 00:14:32.730 align:start position:0% optimum we can guarantee the optimality of<00:14:31.120> the<00:14:31.240> system<00:14:31.600> along<00:14:32.230> the<00:14:32.290> way<00:14:32.529> we're 00:14:32.730 --> 00:14:32.740 align:start position:0% of the system along the way we're 00:14:32.740 --> 00:14:34.350 align:start position:0% of the system along the way we're performing<00:14:33.070> security<00:14:33.730> checks<00:14:34.000> to<00:14:34.209> make<00:14:34.240> sure 00:14:34.350 --> 00:14:34.360 align:start position:0% performing security checks to make sure 00:14:34.360 --> 00:14:35.940 align:start position:0% performing security checks to make sure that<00:14:34.660> none<00:14:34.870> of<00:14:34.899> these<00:14:35.170> nodes<00:14:35.440> have<00:14:35.740> been 00:14:35.940 --> 00:14:35.950 align:start position:0% that none of these nodes have been 00:14:35.950 --> 00:14:37.980 align:start position:0% that none of these nodes have been compromised<00:14:36.880> in<00:14:37.060> a<00:14:37.149> way<00:14:37.270> that<00:14:37.450> will<00:14:37.540> push<00:14:37.810> us 00:14:37.980 --> 00:14:37.990 align:start position:0% compromised in a way that will push us 00:14:37.990 --> 00:14:39.569 align:start position:0% compromised in a way that will push us into<00:14:38.200> an<00:14:38.350> in<00:14:38.470> feasible<00:14:38.620> region<00:14:39.160> caused<00:14:39.550> a 00:14:39.569 --> 00:14:39.579 align:start position:0% into an in feasible region caused a 00:14:39.579 --> 00:14:42.449 align:start position:0% into an in feasible region caused a blackout<00:14:39.820> and<00:14:40.660> ultimately<00:14:41.620> end<00:14:42.010> up<00:14:42.220> with 00:14:42.449 --> 00:14:42.459 align:start position:0% blackout and ultimately end up with 00:14:42.459 --> 00:14:45.150 align:start position:0% blackout and ultimately end up with basically<00:14:42.940> a<00:14:43.120> self<00:14:43.420> scheduling<00:14:43.829> utility<00:14:44.829> free 00:14:45.150 --> 00:14:45.160 align:start position:0% basically a self scheduling utility free 00:14:45.160 --> 00:14:49.980 align:start position:0% basically a self scheduling utility free electricity<00:14:45.760> system<00:14:47.940> so<00:14:48.940> just<00:14:49.630> to<00:14:49.750> sort<00:14:49.930> of 00:14:49.980 --> 00:14:49.990 align:start position:0% electricity system so just to sort of 00:14:49.990 --> 00:14:51.800 align:start position:0% electricity system so just to sort of show<00:14:50.050> some<00:14:50.560> pretty<00:14:50.740> graphs<00:14:50.980> because<00:14:51.370> we<00:14:51.490> have 00:14:51.800 --> 00:14:51.810 align:start position:0% show some pretty graphs because we have 00:14:51.810 --> 00:14:55.199 align:start position:0% show some pretty graphs because we have results<00:14:52.810> to<00:14:53.079> share<00:14:53.290> the<00:14:54.070> we<00:14:54.399> can<00:14:54.579> see<00:14:54.880> how 00:14:55.199 --> 00:14:55.209 align:start position:0% results to share the we can see how 00:14:55.209 --> 00:14:57.410 align:start position:0% results to share the we can see how we're<00:14:55.450> actually<00:14:55.570> shifting<00:14:56.199> around<00:14:56.440> the 00:14:57.410 --> 00:14:57.420 align:start position:0% we're actually shifting around the 00:14:57.420 --> 00:15:00.689 align:start position:0% we're actually shifting around the behind<00:14:58.420> the<00:14:58.480> meter<00:14:58.660> batteries<00:14:59.110> the<00:15:00.040> shapeable 00:15:00.689 --> 00:15:00.699 align:start position:0% behind the meter batteries the shapeable 00:15:00.699 --> 00:15:02.699 align:start position:0% behind the meter batteries the shapeable loads<00:15:01.029> in<00:15:01.329> the<00:15:01.449> purple<00:15:01.750> lines<00:15:01.930> deferrable 00:15:02.699 --> 00:15:02.709 align:start position:0% loads in the purple lines deferrable 00:15:02.709 --> 00:15:05.400 align:start position:0% loads in the purple lines deferrable loads<00:15:03.010> like<00:15:03.399> smart<00:15:03.850> washers<00:15:04.300> or<00:15:04.630> smart 00:15:05.400 --> 00:15:05.410 align:start position:0% loads like smart washers or smart 00:15:05.410 --> 00:15:07.769 align:start position:0% loads like smart washers or smart appliances<00:15:05.920> and<00:15:06.420> moving<00:15:07.420> all<00:15:07.570> of<00:15:07.600> those 00:15:07.769 --> 00:15:07.779 align:start position:0% appliances and moving all of those 00:15:07.779 --> 00:15:09.300 align:start position:0% appliances and moving all of those around<00:15:08.110> to<00:15:08.440> make<00:15:08.560> sure<00:15:08.740> that<00:15:08.949> we're<00:15:09.130> taking 00:15:09.300 --> 00:15:09.310 align:start position:0% around to make sure that we're taking 00:15:09.310 --> 00:15:12.000 align:start position:0% around to make sure that we're taking advantage<00:15:09.519> of<00:15:09.970> all<00:15:10.180> of<00:15:10.209> the<00:15:11.010> photovoltaic 00:15:12.000 --> 00:15:12.010 align:start position:0% advantage of all of the photovoltaic 00:15:12.010 --> 00:15:14.389 align:start position:0% advantage of all of the photovoltaic energy<00:15:12.430> that's<00:15:12.790> being<00:15:12.970> pumped<00:15:13.360> into<00:15:13.600> the<00:15:13.779> grid 00:15:14.389 --> 00:15:14.399 align:start position:0% energy that's being pumped into the grid 00:15:14.399 --> 00:15:16.350 align:start position:0% energy that's being pumped into the grid across<00:15:15.399> the<00:15:15.640> whole<00:15:15.790> node 00:15:16.350 --> 00:15:16.360 align:start position:0% across the whole node 00:15:16.360 --> 00:15:18.300 align:start position:0% across the whole node we're<00:15:16.480> actually<00:15:16.570> monitoring<00:15:17.350> the<00:15:17.890> physical 00:15:18.300 --> 00:15:18.310 align:start position:0% we're actually monitoring the physical 00:15:18.310 --> 00:15:20.670 align:start position:0% we're actually monitoring the physical constraints<00:15:18.820> of<00:15:18.940> the<00:15:19.089> system<00:15:19.510> at<00:15:19.779> every<00:15:20.350> point 00:15:20.670 --> 00:15:20.680 align:start position:0% constraints of the system at every point 00:15:20.680 --> 00:15:22.070 align:start position:0% constraints of the system at every point in<00:15:21.010> the 00:15:22.070 --> 00:15:22.080 align:start position:0% in the 00:15:22.080 --> 00:15:24.620 align:start position:0% in the the<00:15:22.680> distribution<00:15:23.250> network<00:15:23.279> we're<00:15:24.149> making 00:15:24.620 --> 00:15:24.630 align:start position:0% the distribution network we're making 00:15:24.630 --> 00:15:26.360 align:start position:0% the distribution network we're making sure<00:15:24.779> that<00:15:24.899> the<00:15:25.020> voltage<00:15:25.380> doesn't<00:15:25.860> go<00:15:26.070> out<00:15:26.220> of 00:15:26.360 --> 00:15:26.370 align:start position:0% sure that the voltage doesn't go out of 00:15:26.370 --> 00:15:28.190 align:start position:0% sure that the voltage doesn't go out of our<00:15:26.459> acceptable<00:15:27.000> bounds<00:15:27.330> this<00:15:27.899> is<00:15:28.080> the 00:15:28.190 --> 00:15:28.200 align:start position:0% our acceptable bounds this is the 00:15:28.200 --> 00:15:29.960 align:start position:0% our acceptable bounds this is the physical<00:15:28.740> constraint<00:15:29.220> that<00:15:29.459> defines<00:15:29.790> the 00:15:29.960 --> 00:15:29.970 align:start position:0% physical constraint that defines the 00:15:29.970 --> 00:15:33.440 align:start position:0% physical constraint that defines the system<00:15:30.390> and<00:15:30.600> we're<00:15:31.440> able<00:15:31.560> to<00:15:32.029> converge<00:15:33.029> on<00:15:33.390> a 00:15:33.440 --> 00:15:33.450 align:start position:0% system and we're able to converge on a 00:15:33.450 --> 00:15:36.079 align:start position:0% system and we're able to converge on a schedule<00:15:33.930> that<00:15:34.230> is<00:15:34.380> only<00:15:35.130> suboptimal<00:15:35.790> in 00:15:36.079 --> 00:15:36.089 align:start position:0% schedule that is only suboptimal in 00:15:36.089 --> 00:15:39.560 align:start position:0% schedule that is only suboptimal in price<00:15:36.360> by<00:15:36.980> 0.4<00:15:37.980> percent<00:15:38.130> so<00:15:38.490> really<00:15:39.089> close<00:15:39.330> to 00:15:39.560 --> 00:15:39.570 align:start position:0% price by 0.4 percent so really close to 00:15:39.570 --> 00:15:41.210 align:start position:0% price by 0.4 percent so really close to what<00:15:39.750> you<00:15:39.839> would<00:15:39.930> get<00:15:40.110> if<00:15:40.350> a<00:15:40.529> utility<00:15:40.860> were 00:15:41.210 --> 00:15:41.220 align:start position:0% what you would get if a utility were 00:15:41.220 --> 00:15:43.460 align:start position:0% what you would get if a utility were solving<00:15:41.880> all<00:15:42.060> of<00:15:42.360> this<00:15:42.510> privately 00:15:43.460 --> 00:15:43.470 align:start position:0% solving all of this privately 00:15:43.470 --> 00:15:45.800 align:start position:0% solving all of this privately and<00:15:43.649> then<00:15:44.459> we<00:15:44.700> can<00:15:44.940> also<00:15:45.060> guarantee<00:15:45.390> the 00:15:45.800 --> 00:15:45.810 align:start position:0% and then we can also guarantee the 00:15:45.810 --> 00:15:47.150 align:start position:0% and then we can also guarantee the constraints<00:15:46.290> are<00:15:46.410> satisfied<00:15:46.440> that<00:15:47.040> this 00:15:47.150 --> 00:15:47.160 align:start position:0% constraints are satisfied that this 00:15:47.160 --> 00:15:48.740 align:start position:0% constraints are satisfied that this isn't<00:15:47.490> going<00:15:47.640> to<00:15:47.730> blow<00:15:47.850> up<00:15:48.089> that<00:15:48.660> our 00:15:48.740 --> 00:15:48.750 align:start position:0% isn't going to blow up that our 00:15:48.750 --> 00:15:50.210 align:start position:0% isn't going to blow up that our electricity<00:15:49.260> system<00:15:49.620> is<00:15:49.740> still<00:15:49.920> going<00:15:50.100> to 00:15:50.210 --> 00:15:50.220 align:start position:0% electricity system is still going to 00:15:50.220 --> 00:15:52.670 align:start position:0% electricity system is still going to stay<00:15:50.399> stable<00:15:50.850> so<00:15:51.510> we've<00:15:51.899> solved<00:15:52.320> a<00:15:52.410> hard 00:15:52.670 --> 00:15:52.680 align:start position:0% stay stable so we've solved a hard 00:15:52.680 --> 00:15:54.410 align:start position:0% stay stable so we've solved a hard optimization<00:15:53.310> problem<00:15:53.760> with<00:15:53.970> physical 00:15:54.410 --> 00:15:54.420 align:start position:0% optimization problem with physical 00:15:54.420 --> 00:15:56.990 align:start position:0% optimization problem with physical constraints<00:15:55.050> with<00:15:55.800> an<00:15:56.399> algorithm<00:15:56.760> that 00:15:56.990 --> 00:15:57.000 align:start position:0% constraints with an algorithm that 00:15:57.000 --> 00:16:00.380 align:start position:0% constraints with an algorithm that respect<00:15:57.570> local<00:15:57.959> privacy<00:15:58.649> and<00:15:58.980> can<00:15:59.970> guarantee 00:16:00.380 --> 00:16:00.390 align:start position:0% respect local privacy and can guarantee 00:16:00.390 --> 00:16:02.360 align:start position:0% respect local privacy and can guarantee the<00:16:00.690> optimality<00:16:01.170> and<00:16:01.649> the<00:16:01.770> feasibility<00:16:02.160> of 00:16:02.360 --> 00:16:02.370 align:start position:0% the optimality and the feasibility of 00:16:02.370 --> 00:16:06.019 align:start position:0% the optimality and the feasibility of the<00:16:02.520> solution<00:16:04.850> challenges<00:16:05.850> and 00:16:06.019 --> 00:16:06.029 align:start position:0% the solution challenges and 00:16:06.029 --> 00:16:07.220 align:start position:0% the solution challenges and opportunities<00:16:06.720> looking<00:16:06.959> ahead 00:16:07.220 --> 00:16:07.230 align:start position:0% opportunities looking ahead 00:16:07.230 --> 00:16:10.070 align:start position:0% opportunities looking ahead so<00:16:07.500> these<00:16:08.070> mirror<00:16:08.430> the<00:16:08.730> the<00:16:08.760> three<00:16:09.600> big<00:16:09.839> areas 00:16:10.070 --> 00:16:10.080 align:start position:0% so these mirror the the three big areas 00:16:10.080 --> 00:16:11.870 align:start position:0% so these mirror the the three big areas the<00:16:10.440> Vitalik<00:16:10.860> was<00:16:11.190> talking<00:16:11.550> about<00:16:11.580> yesterday 00:16:11.870 --> 00:16:11.880 align:start position:0% the Vitalik was talking about yesterday 00:16:11.880 --> 00:16:15.350 align:start position:0% the Vitalik was talking about yesterday one<00:16:12.839> is<00:16:13.020> privacy<00:16:13.440> so<00:16:13.860> I'm<00:16:14.370> very<00:16:14.910> interested<00:16:15.120> in 00:16:15.350 --> 00:16:15.360 align:start position:0% one is privacy so I'm very interested in 00:16:15.360 --> 00:16:17.060 align:start position:0% one is privacy so I'm very interested in the<00:16:15.480> sessions<00:16:15.810> tomorrow<00:16:15.959> on<00:16:16.290> zero<00:16:16.890> knowledge 00:16:17.060 --> 00:16:17.070 align:start position:0% the sessions tomorrow on zero knowledge 00:16:17.070 --> 00:16:19.370 align:start position:0% the sessions tomorrow on zero knowledge proof<00:16:17.190> to<00:16:18.000> be<00:16:18.089> able<00:16:18.149> to<00:16:18.240> identify<00:16:18.750> how<00:16:19.140> can<00:16:19.320> we 00:16:19.370 --> 00:16:19.380 align:start position:0% proof to be able to identify how can we 00:16:19.380 --> 00:16:21.380 align:start position:0% proof to be able to identify how can we actually<00:16:19.560> make<00:16:20.459> sure<00:16:20.520> that<00:16:20.700> the<00:16:21.000> information 00:16:21.380 --> 00:16:21.390 align:start position:0% actually make sure that the information 00:16:21.390 --> 00:16:23.360 align:start position:0% actually make sure that the information we're<00:16:21.810> sending<00:16:22.050> back<00:16:22.200> and<00:16:22.529> forth<00:16:22.770> is<00:16:23.100> not 00:16:23.360 --> 00:16:23.370 align:start position:0% we're sending back and forth is not 00:16:23.370 --> 00:16:25.910 align:start position:0% we're sending back and forth is not sensitive<00:16:24.180> so<00:16:24.480> in<00:16:24.750> the<00:16:25.050> model<00:16:25.560> that<00:16:25.649> I<00:16:25.709> just 00:16:25.910 --> 00:16:25.920 align:start position:0% sensitive so in the model that I just 00:16:25.920 --> 00:16:28.340 align:start position:0% sensitive so in the model that I just described<00:16:26.430> the<00:16:27.120> actual<00:16:27.450> power<00:16:27.720> consumption 00:16:28.340 --> 00:16:28.350 align:start position:0% described the actual power consumption 00:16:28.350 --> 00:16:30.019 align:start position:0% described the actual power consumption for<00:16:28.560> each<00:16:28.649> household<00:16:29.160> ends<00:16:29.820> up<00:16:29.910> getting 00:16:30.019 --> 00:16:30.029 align:start position:0% for each household ends up getting 00:16:30.029 --> 00:16:32.630 align:start position:0% for each household ends up getting stored<00:16:30.510> on<00:16:30.660> the<00:16:30.810> blockchain<00:16:31.640> there's 00:16:32.630 --> 00:16:32.640 align:start position:0% stored on the blockchain there's 00:16:32.640 --> 00:16:35.510 align:start position:0% stored on the blockchain there's concerns<00:16:33.180> about<00:16:33.480> say<00:16:33.779> a<00:16:34.230> thief<00:16:34.860> being<00:16:35.250> able<00:16:35.370> to 00:16:35.510 --> 00:16:35.520 align:start position:0% concerns about say a thief being able to 00:16:35.520 --> 00:16:38.569 align:start position:0% concerns about say a thief being able to identify<00:16:35.779> when<00:16:36.779> somebody<00:16:37.320> is<00:16:37.529> away<00:16:38.190> from<00:16:38.250> home 00:16:38.569 --> 00:16:38.579 align:start position:0% identify when somebody is away from home 00:16:38.579 --> 00:16:40.100 align:start position:0% identify when somebody is away from home and<00:16:38.820> thus<00:16:39.000> could<00:16:39.209> break<00:16:39.360> into<00:16:39.540> the<00:16:39.660> house<00:16:39.810> if 00:16:40.100 --> 00:16:40.110 align:start position:0% and thus could break into the house if 00:16:40.110 --> 00:16:41.960 align:start position:0% and thus could break into the house if they<00:16:40.709> know<00:16:40.860> what<00:16:41.130> everybody's<00:16:41.490> expected 00:16:41.960 --> 00:16:41.970 align:start position:0% they know what everybody's expected 00:16:41.970 --> 00:16:44.420 align:start position:0% they know what everybody's expected power<00:16:42.149> consumption<00:16:42.690> is<00:16:42.870> the<00:16:43.079> other<00:16:43.800> part<00:16:44.130> is 00:16:44.420 --> 00:16:44.430 align:start position:0% power consumption is the other part is 00:16:44.430 --> 00:16:46.790 align:start position:0% power consumption is the other part is there's<00:16:45.360> a<00:16:45.420> potential<00:16:45.810> for<00:16:46.110> data<00:16:46.290> leakage 00:16:46.790 --> 00:16:46.800 align:start position:0% there's a potential for data leakage 00:16:46.800 --> 00:16:49.519 align:start position:0% there's a potential for data leakage through<00:16:47.430> the<00:16:47.730> iterations<00:16:48.270> back<00:16:48.540> and<00:16:48.750> forth<00:16:48.990> so 00:16:49.519 --> 00:16:49.529 align:start position:0% through the iterations back and forth so 00:16:49.529 --> 00:16:51.380 align:start position:0% through the iterations back and forth so that<00:16:49.560> even<00:16:49.890> if<00:16:50.010> the<00:16:50.250> information<00:16:50.550> that<00:16:51.300> you're 00:16:51.380 --> 00:16:51.390 align:start position:0% that even if the information that you're 00:16:51.390 --> 00:16:53.510 align:start position:0% that even if the information that you're sending<00:16:51.420> is<00:16:52.020> not<00:16:52.290> fundamentally<00:16:53.040> sensitive 00:16:53.510 --> 00:16:53.520 align:start position:0% sending is not fundamentally sensitive 00:16:53.520 --> 00:16:56.000 align:start position:0% sending is not fundamentally sensitive that<00:16:54.089> the<00:16:54.270> information<00:16:54.990> that<00:16:55.350> is<00:16:55.500> leaking<00:16:55.920> out 00:16:56.000 --> 00:16:56.010 align:start position:0% that the information that is leaking out 00:16:56.010 --> 00:16:57.860 align:start position:0% that the information that is leaking out of<00:16:56.190> the<00:16:56.310> system<00:16:56.670> by<00:16:57.029> how<00:16:57.329> much<00:16:57.570> you're 00:16:57.860 --> 00:16:57.870 align:start position:0% of the system by how much you're 00:16:57.870 --> 00:16:59.900 align:start position:0% of the system by how much you're changing<00:16:58.380> your<00:16:58.529> estimate<00:16:59.010> at<00:16:59.430> each<00:16:59.610> of<00:16:59.760> these 00:16:59.900 --> 00:16:59.910 align:start position:0% changing your estimate at each of these 00:16:59.910 --> 00:17:02.120 align:start position:0% changing your estimate at each of these steps<00:17:00.149> might<00:17:00.959> indicate<00:17:01.350> something<00:17:01.770> about 00:17:02.120 --> 00:17:02.130 align:start position:0% steps might indicate something about 00:17:02.130 --> 00:17:04.400 align:start position:0% steps might indicate something about what<00:17:02.640> your<00:17:02.940> private<00:17:03.270> optimization<00:17:04.050> problem 00:17:04.400 --> 00:17:04.410 align:start position:0% what your private optimization problem 00:17:04.410 --> 00:17:08.660 align:start position:0% what your private optimization problem is<00:17:05.360> so<00:17:06.360> the<00:17:06.600> security<00:17:07.140> we've<00:17:08.010> been<00:17:08.250> working<00:17:08.550> on 00:17:08.660 --> 00:17:08.670 align:start position:0% is so the security we've been working on 00:17:08.670 --> 00:17:10.970 align:start position:0% is so the security we've been working on like<00:17:09.179> provable<00:17:09.750> guarantees<00:17:10.199> of<00:17:10.620> the 00:17:10.970 --> 00:17:10.980 align:start position:0% like provable guarantees of the 00:17:10.980 --> 00:17:14.210 align:start position:0% like provable guarantees of the feasibility<00:17:11.459> with<00:17:12.449> untrusted<00:17:12.990> nodes<00:17:13.199> one<00:17:14.010> big 00:17:14.210 --> 00:17:14.220 align:start position:0% feasibility with untrusted nodes one big 00:17:14.220 --> 00:17:17.090 align:start position:0% feasibility with untrusted nodes one big comment<00:17:14.790> here<00:17:15.120> is<00:17:15.209> that<00:17:15.360> these<00:17:16.100> proofs 00:17:17.090 --> 00:17:17.100 align:start position:0% comment here is that these proofs 00:17:17.100 --> 00:17:19.040 align:start position:0% comment here is that these proofs generally<00:17:17.670> have<00:17:17.850> much<00:17:18.179> weaker<00:17:18.600> conditions 00:17:19.040 --> 00:17:19.050 align:start position:0% generally have much weaker conditions 00:17:19.050 --> 00:17:19.790 align:start position:0% generally have much weaker conditions than<00:17:19.410> the 00:17:19.790 --> 00:17:19.800 align:start position:0% than the 00:17:19.800 --> 00:17:23.960 align:start position:0% than the tea<00:17:20.420> would<00:17:21.420> give<00:17:21.630> so<00:17:21.930> in<00:17:22.500> these<00:17:22.970> constraint 00:17:23.960 --> 00:17:23.970 align:start position:0% tea would give so in these constraint 00:17:23.970 --> 00:17:26.120 align:start position:0% tea would give so in these constraint systems<00:17:24.390> your<00:17:25.200> ability<00:17:25.650> to<00:17:25.680> guarantee 00:17:26.120 --> 00:17:26.130 align:start position:0% systems your ability to guarantee 00:17:26.130 --> 00:17:28.190 align:start position:0% systems your ability to guarantee feasibility<00:17:26.760> is<00:17:26.880> going<00:17:27.120> to<00:17:27.240> be<00:17:27.480> dependent<00:17:28.050> on 00:17:28.190 --> 00:17:28.200 align:start position:0% feasibility is going to be dependent on 00:17:28.200 --> 00:17:31.220 align:start position:0% feasibility is going to be dependent on the<00:17:28.620> topology<00:17:29.010> of<00:17:29.520> how<00:17:29.840> faulty<00:17:30.840> nodes<00:17:31.080> are 00:17:31.220 --> 00:17:31.230 align:start position:0% the topology of how faulty nodes are 00:17:31.230 --> 00:17:34.160 align:start position:0% the topology of how faulty nodes are connected<00:17:31.350> and<00:17:32.000> in<00:17:33.000> general<00:17:33.420> though<00:17:33.630> because 00:17:34.160 --> 00:17:34.170 align:start position:0% connected and in general though because 00:17:34.170 --> 00:17:35.960 align:start position:0% connected and in general though because it's<00:17:34.500> going<00:17:34.620> to<00:17:34.710> be<00:17:34.800> dependent<00:17:35.160> on<00:17:35.250> that<00:17:35.430> we 00:17:35.960 --> 00:17:35.970 align:start position:0% it's going to be dependent on that we 00:17:35.970 --> 00:17:38.240 align:start position:0% it's going to be dependent on that we can't<00:17:36.270> guarantee<00:17:36.480> the<00:17:36.960> same<00:17:37.170> one<00:17:37.620> third<00:17:37.890> proof 00:17:38.240 --> 00:17:38.250 align:start position:0% can't guarantee the same one third proof 00:17:38.250 --> 00:17:40.370 align:start position:0% can't guarantee the same one third proof against<00:17:38.640> attack<00:17:38.910> that<00:17:39.600> bft<00:17:39.990> systems 00:17:40.370 --> 00:17:40.380 align:start position:0% against attack that bft systems 00:17:40.380 --> 00:17:43.610 align:start position:0% against attack that bft systems generally<00:17:40.650> have<00:17:41.300> scalability<00:17:42.300> so<00:17:42.620> currently 00:17:43.610 --> 00:17:43.620 align:start position:0% generally have scalability so currently 00:17:43.620 --> 00:17:45.830 align:start position:0% generally have scalability so currently looking<00:17:44.340> at<00:17:44.430> converging<00:17:45.120> in<00:17:45.210> about<00:17:45.390> 20 00:17:45.830 --> 00:17:45.840 align:start position:0% looking at converging in about 20 00:17:45.840 --> 00:17:47.600 align:start position:0% looking at converging in about 20 iterations<00:17:45.960> of<00:17:46.620> the<00:17:46.770> system<00:17:47.100> each<00:17:47.310> of<00:17:47.490> those 00:17:47.600 --> 00:17:47.610 align:start position:0% iterations of the system each of those 00:17:47.610 --> 00:17:50.540 align:start position:0% iterations of the system each of those iterations<00:17:47.820> requires<00:17:48.660> update<00:17:49.500> steps<00:17:49.830> or 00:17:50.540 --> 00:17:50.550 align:start position:0% iterations requires update steps or 00:17:50.550 --> 00:17:52.430 align:start position:0% iterations requires update steps or variable<00:17:51.240> updates<00:17:51.540> sent<00:17:51.810> to<00:17:51.930> the<00:17:52.050> aggregator 00:17:52.430 --> 00:17:52.440 align:start position:0% variable updates sent to the aggregator 00:17:52.440 --> 00:17:55.070 align:start position:0% variable updates sent to the aggregator node<00:17:52.710> the<00:17:53.280> aggregator<00:17:53.730> node<00:17:53.940> performing<00:17:54.720> that 00:17:55.070 --> 00:17:55.080 align:start position:0% node the aggregator node performing that 00:17:55.080 --> 00:17:56.720 align:start position:0% node the aggregator node performing that or<00:17:55.500> the<00:17:55.710> aggregator<00:17:56.010> smart<00:17:56.310> contract 00:17:56.720 --> 00:17:56.730 align:start position:0% or the aggregator smart contract 00:17:56.730 --> 00:17:58.220 align:start position:0% or the aggregator smart contract performing<00:17:57.210> that<00:17:57.300> update<00:17:57.660> and<00:17:57.810> then 00:17:58.220 --> 00:17:58.230 align:start position:0% performing that update and then 00:17:58.230 --> 00:18:01.520 align:start position:0% performing that update and then broadcasting<00:17:58.860> that<00:17:58.980> back<00:17:59.190> so<00:18:00.170> each<00:18:01.170> of<00:18:01.230> those 00:18:01.520 --> 00:18:01.530 align:start position:0% broadcasting that back so each of those 00:18:01.530 --> 00:18:03.850 align:start position:0% broadcasting that back so each of those is<00:18:01.800> at<00:18:02.040> minimum<00:18:02.220> two<00:18:02.670> to<00:18:02.790> three<00:18:03.000> secured<00:18:03.480> box 00:18:03.850 --> 00:18:03.860 align:start position:0% is at minimum two to three secured box 00:18:03.860 --> 00:18:06.940 align:start position:0% is at minimum two to three secured box so<00:18:04.860> in<00:18:05.160> net<00:18:05.460> this<00:18:05.820> could<00:18:06.000> be<00:18:06.090> a<00:18:06.120> relatively 00:18:06.940 --> 00:18:06.950 align:start position:0% so in net this could be a relatively 00:18:06.950 --> 00:18:10.310 align:start position:0% so in net this could be a relatively long<00:18:07.950> slow<00:18:08.520> process<00:18:08.750> it's<00:18:09.750> an<00:18:09.840> area<00:18:09.990> where 00:18:10.310 --> 00:18:10.320 align:start position:0% long slow process it's an area where 00:18:10.320 --> 00:18:12.880 align:start position:0% long slow process it's an area where again<00:18:10.650> like<00:18:11.100> looking<00:18:11.430> at<00:18:11.550> plasma<00:18:12.120> or<00:18:12.270> other 00:18:12.880 --> 00:18:12.890 align:start position:0% again like looking at plasma or other 00:18:12.890 --> 00:18:15.680 align:start position:0% again like looking at plasma or other ways<00:18:13.890> of<00:18:14.040> speeding<00:18:14.280> this<00:18:14.520> up<00:18:14.700> could<00:18:15.420> improve 00:18:15.680 --> 00:18:15.690 align:start position:0% ways of speeding this up could improve 00:18:15.690 --> 00:18:17.780 align:start position:0% ways of speeding this up could improve the<00:18:15.870> system<00:18:16.230> with<00:18:17.190> that<00:18:17.310> I<00:18:17.340> think<00:18:17.490> we<00:18:17.700> have 00:18:17.780 --> 00:18:17.790 align:start position:0% the system with that I think we have 00:18:17.790 --> 00:18:33.470 align:start position:0% the system with that I think we have some<00:18:17.940> extra<00:18:18.180> time<00:18:18.620> sure 00:18:33.470 --> 00:18:33.480 align:start position:0% 00:18:33.480 --> 00:18:37.190 align:start position:0% one<00:18:34.050> of<00:18:34.200> the<00:18:35.390> one<00:18:36.390> of<00:18:36.480> the<00:18:36.570> general<00:18:36.870> techniques 00:18:37.190 --> 00:18:37.200 align:start position:0% one of the one of the general techniques 00:18:37.200 --> 00:18:38.510 align:start position:0% one of the one of the general techniques for<00:18:37.320> dealing<00:18:37.500> with<00:18:37.560> blockchain<00:18:37.920> scalability 00:18:38.510 --> 00:18:38.520 align:start position:0% for dealing with blockchain scalability 00:18:38.520 --> 00:18:40.190 align:start position:0% for dealing with blockchain scalability is<00:18:38.640> only<00:18:39.150> going<00:18:39.570> to<00:18:39.630> the<00:18:39.720> blockchain<00:18:39.930> when 00:18:40.190 --> 00:18:40.200 align:start position:0% is only going to the blockchain when 00:18:40.200 --> 00:18:41.930 align:start position:0% is only going to the blockchain when necessary<00:18:40.350> are<00:18:40.860> there<00:18:41.070> equivalents<00:18:41.730> here 00:18:41.930 --> 00:18:41.940 align:start position:0% necessary are there equivalents here 00:18:41.940 --> 00:18:43.640 align:start position:0% necessary are there equivalents here where<00:18:42.090> you<00:18:42.180> could<00:18:42.390> sort<00:18:42.930> of<00:18:43.050> determine<00:18:43.530> that 00:18:43.640 --> 00:18:43.650 align:start position:0% where you could sort of determine that 00:18:43.650 --> 00:18:45.560 align:start position:0% where you could sort of determine that some<00:18:43.920> subset<00:18:44.610> of<00:18:44.640> the<00:18:44.760> network<00:18:45.060> is<00:18:45.240> behaving 00:18:45.560 --> 00:18:45.570 align:start position:0% some subset of the network is behaving 00:18:45.570 --> 00:18:47.330 align:start position:0% some subset of the network is behaving badly<00:18:45.810> and<00:18:46.320> only<00:18:46.620> go<00:18:46.710> to<00:18:46.770> the<00:18:46.890> blockchain<00:18:47.070> in 00:18:47.330 --> 00:18:47.340 align:start position:0% badly and only go to the blockchain in 00:18:47.340 --> 00:18:49.340 align:start position:0% badly and only go to the blockchain in that<00:18:47.460> case<00:18:47.640> yeah<00:18:48.360> so<00:18:48.630> one<00:18:48.810> area<00:18:48.960> that<00:18:49.140> I'm 00:18:49.340 --> 00:18:49.350 align:start position:0% that case yeah so one area that I'm 00:18:49.350 --> 00:18:50.780 align:start position:0% that case yeah so one area that I'm interested<00:18:49.860> in<00:18:49.890> looking<00:18:50.100> at<00:18:50.220> is<00:18:50.340> state 00:18:50.780 --> 00:18:50.790 align:start position:0% interested in looking at is state 00:18:50.790 --> 00:18:53.140 align:start position:0% interested in looking at is state channels<00:18:51.210> to<00:18:51.390> be<00:18:51.480> able<00:18:51.600> to<00:18:51.840> identify 00:18:53.140 --> 00:18:53.150 align:start position:0% channels to be able to identify 00:18:53.150 --> 00:18:55.970 align:start position:0% channels to be able to identify basically<00:18:54.150> what<00:18:54.360> are<00:18:54.480> the<00:18:54.600> the<00:18:55.380> portions<00:18:55.830> that 00:18:55.970 --> 00:18:55.980 align:start position:0% basically what are the the portions that 00:18:55.980 --> 00:18:59.270 align:start position:0% basically what are the the portions that we<00:18:56.220> need<00:18:56.370> to<00:18:56.520> update<00:18:56.700> and<00:18:57.500> what<00:18:58.500> which 00:18:59.270 --> 00:18:59.280 align:start position:0% we need to update and what which 00:18:59.280 --> 00:19:00.710 align:start position:0% we need to update and what which portions<00:18:59.700> of<00:18:59.760> the<00:18:59.880> updates<00:19:00.150> need<00:19:00.360> to<00:19:00.420> be<00:19:00.510> saved 00:19:00.710 --> 00:19:00.720 align:start position:0% portions of the updates need to be saved 00:19:00.720 --> 00:19:02.360 align:start position:0% portions of the updates need to be saved to<00:19:00.810> the<00:19:00.870> blockchain<00:19:01.350> which<00:19:01.590> portions<00:19:02.280> can 00:19:02.360 --> 00:19:02.370 align:start position:0% to the blockchain which portions can 00:19:02.370 --> 00:19:05.390 align:start position:0% to the blockchain which portions can just<00:19:02.820> be<00:19:02.910> done<00:19:03.090> through<00:19:03.450> state<00:19:03.810> channels<00:19:04.400> and 00:19:05.390 --> 00:19:05.400 align:start position:0% just be done through state channels and 00:19:05.400 --> 00:19:07.940 align:start position:0% just be done through state channels and not<00:19:05.760> have<00:19:06.060> all<00:19:06.210> of<00:19:06.360> that<00:19:06.390> constantly<00:19:07.200> back<00:19:07.920> and 00:19:07.940 --> 00:19:07.950 align:start position:0% not have all of that constantly back and 00:19:07.950 --> 00:19:11.870 align:start position:0% not have all of that constantly back and forth<00:19:08.280> on<00:19:08.370> the<00:19:08.490> main<00:19:08.610> blockchain<00:19:10.880> you 00:19:11.870 --> 00:19:11.880 align:start position:0% forth on the main blockchain you 00:19:11.880 --> 00:19:13.280 align:start position:0% forth on the main blockchain you mentioned<00:19:12.210> there<00:19:12.330> are<00:19:12.360> some<00:19:12.450> nodes<00:19:12.990> that 00:19:13.280 --> 00:19:13.290 align:start position:0% mentioned there are some nodes that 00:19:13.290 --> 00:19:15.620 align:start position:0% mentioned there are some nodes that might<00:19:13.560> be<00:19:13.770> untrustworthy<00:19:14.190> but<00:19:14.910> it<00:19:15.300> seems<00:19:15.540> that 00:19:15.620 --> 00:19:15.630 align:start position:0% might be untrustworthy but it seems that 00:19:15.630 --> 00:19:17.090 align:start position:0% might be untrustworthy but it seems that every<00:19:15.900> node<00:19:16.050> corresponds<00:19:16.500> to<00:19:16.590> a<00:19:16.680> household 00:19:17.090 --> 00:19:17.100 align:start position:0% every node corresponds to a household 00:19:17.100 --> 00:19:20.540 align:start position:0% every node corresponds to a household and<00:19:17.400> every<00:19:17.880> household<00:19:18.240> has<00:19:18.360> you<00:19:18.990> know<00:19:19.110> is<00:19:19.550> it 00:19:20.540 --> 00:19:20.550 align:start position:0% and every household has you know is it 00:19:20.550 --> 00:19:23.450 align:start position:0% and every household has you know is it is<00:19:20.790> in<00:19:21.150> their<00:19:21.390> own<00:19:21.420> interest<00:19:21.990> to<00:19:22.080> act<00:19:22.370> you<00:19:23.370> know 00:19:23.450 --> 00:19:23.460 align:start position:0% is in their own interest to act you know 00:19:23.460 --> 00:19:25.310 align:start position:0% is in their own interest to act you know the<00:19:23.580> trustworthy<00:19:24.120> manner<00:19:24.300> so<00:19:24.660> why<00:19:24.810> is<00:19:24.870> that<00:19:25.140> an 00:19:25.310 --> 00:19:25.320 align:start position:0% the trustworthy manner so why is that an 00:19:25.320 --> 00:19:28.100 align:start position:0% the trustworthy manner so why is that an issue<00:19:25.560> here<00:19:26.220> so<00:19:26.730> if<00:19:26.880> you<00:19:27.000> think<00:19:27.210> of<00:19:27.420> say<00:19:27.720> the 00:19:28.100 --> 00:19:28.110 align:start position:0% issue here so if you think of say the 00:19:28.110 --> 00:19:30.320 align:start position:0% issue here so if you think of say the hack<00:19:28.530> against<00:19:28.920> the<00:19:29.010> Ukrainian<00:19:29.400> electricity 00:19:30.320 --> 00:19:30.330 align:start position:0% hack against the Ukrainian electricity 00:19:30.330 --> 00:19:32.060 align:start position:0% hack against the Ukrainian electricity grid<00:19:30.540> where<00:19:30.870> an<00:19:31.110> outside<00:19:31.500> party<00:19:31.920> was 00:19:32.060 --> 00:19:32.070 align:start position:0% grid where an outside party was 00:19:32.070 --> 00:19:33.290 align:start position:0% grid where an outside party was interested<00:19:32.400> in<00:19:32.460> actually<00:19:32.580> just<00:19:32.880> bringing 00:19:33.290 --> 00:19:33.300 align:start position:0% interested in actually just bringing 00:19:33.300 --> 00:19:35.120 align:start position:0% interested in actually just bringing down<00:19:33.450> the<00:19:33.660> whole<00:19:33.810> grid<00:19:34.110> for<00:19:34.410> malicious<00:19:34.740> intent 00:19:35.120 --> 00:19:35.130 align:start position:0% down the whole grid for malicious intent 00:19:35.130 --> 00:19:37.160 align:start position:0% down the whole grid for malicious intent you<00:19:35.970> don't<00:19:36.150> care<00:19:36.330> about<00:19:36.420> the<00:19:36.720> private 00:19:37.160 --> 00:19:37.170 align:start position:0% you don't care about the private 00:19:37.170 --> 00:19:39.860 align:start position:0% you don't care about the private household<00:19:37.790> reasons<00:19:38.790> for<00:19:38.970> the<00:19:39.060> attack<00:19:39.150> there 00:19:39.860 --> 00:19:39.870 align:start position:0% household reasons for the attack there 00:19:39.870 --> 00:19:41.360 align:start position:0% household reasons for the attack there might<00:19:40.050> be<00:19:40.080> somebody<00:19:40.560> who<00:19:40.590> just<00:19:41.010> wants<00:19:41.340> to 00:19:41.360 --> 00:19:41.370 align:start position:0% might be somebody who just wants to 00:19:41.370 --> 00:19:50.000 align:start position:0% might be somebody who just wants to cause<00:19:41.850> a<00:19:41.880> blackout<00:19:46.940> so<00:19:47.940> the<00:19:48.300> this<00:19:48.660> is<00:19:48.870> a<00:19:49.140> really 00:19:50.000 --> 00:19:50.010 align:start position:0% cause a blackout so the this is a really 00:19:50.010 --> 00:19:51.350 align:start position:0% cause a blackout so the this is a really unique<00:19:50.130> challenge<00:19:50.610> of<00:19:50.700> this<00:19:50.940> type<00:19:51.150> of<00:19:51.180> work 00:19:51.350 --> 00:19:51.360 align:start position:0% unique challenge of this type of work 00:19:51.360 --> 00:19:55.040 align:start position:0% unique challenge of this type of work the<00:19:51.930> the<00:19:52.860> private<00:19:53.340> problem<00:19:54.270> of<00:19:54.480> the<00:19:54.630> household 00:19:55.040 --> 00:19:55.050 align:start position:0% the the private problem of the household 00:19:55.050 --> 00:19:56.930 align:start position:0% the the private problem of the household is<00:19:55.200> dependent<00:19:55.680> on<00:19:55.740> constraint<00:19:56.400> information 00:19:56.930 --> 00:19:56.940 align:start position:0% is dependent on constraint information 00:19:56.940 --> 00:19:59.540 align:start position:0% is dependent on constraint information fed<00:19:57.450> to<00:19:57.600> them<00:19:57.750> by<00:19:57.990> all<00:19:58.320> the<00:19:58.470> devices<00:19:58.710> so 00:19:59.540 --> 00:19:59.550 align:start position:0% fed to them by all the devices so 00:19:59.550 --> 00:20:01.780 align:start position:0% fed to them by all the devices so somebody<00:19:59.910> could<00:20:00.030> hack<00:20:00.300> an<00:20:00.480> electric<00:20:00.930> car<00:20:01.290> and 00:20:01.780 --> 00:20:01.790 align:start position:0% somebody could hack an electric car and 00:20:01.790 --> 00:20:05.450 align:start position:0% somebody could hack an electric car and pass<00:20:02.790> bad<00:20:03.150> information<00:20:03.540> to<00:20:04.080> the<00:20:04.290> even<00:20:05.190> if<00:20:05.310> it's 00:20:05.450 --> 00:20:05.460 align:start position:0% pass bad information to the even if it's 00:20:05.460 --> 00:20:07.280 align:start position:0% pass bad information to the even if it's a<00:20:05.550> secure<00:20:05.850> Enclave<00:20:06.030> that<00:20:06.600> does<00:20:06.840> the<00:20:07.020> actual 00:20:07.280 --> 00:20:07.290 align:start position:0% a secure Enclave that does the actual 00:20:07.290 --> 00:20:09.440 align:start position:0% a secure Enclave that does the actual computation<00:20:07.590> if<00:20:08.370> that's<00:20:08.640> a<00:20:08.760> cure<00:20:08.940> Enclave<00:20:09.300> is 00:20:09.440 --> 00:20:09.450 align:start position:0% computation if that's a cure Enclave is 00:20:09.450 --> 00:20:12.980 align:start position:0% computation if that's a cure Enclave is getting<00:20:09.750> garbage<00:20:09.990> then<00:20:10.920> it<00:20:11.460> needs<00:20:11.790> to<00:20:11.990> you 00:20:12.980 --> 00:20:12.990 align:start position:0% getting garbage then it needs to you 00:20:12.990 --> 00:20:14.810 align:start position:0% getting garbage then it needs to you effectively<00:20:13.800> can't<00:20:14.160> expect<00:20:14.550> that<00:20:14.700> the 00:20:14.810 --> 00:20:14.820 align:start position:0% effectively can't expect that the 00:20:14.820 --> 00:20:16.550 align:start position:0% effectively can't expect that the message<00:20:15.120> is<00:20:15.210> being<00:20:15.390> passed<00:20:15.630> the<00:20:15.810> network<00:20:16.140> or 00:20:16.550 --> 00:20:16.560 align:start position:0% message is being passed the network or 00:20:16.560 --> 00:20:19.700 align:start position:0% message is being passed the network or trustworthy<00:20:18.020> one<00:20:19.020> question<00:20:19.350> though<00:20:19.470> there 00:20:19.700 --> 00:20:19.710 align:start position:0% trustworthy one question though there 00:20:19.710 --> 00:20:21.200 align:start position:0% trustworthy one question though there may<00:20:19.800> be<00:20:19.830> one<00:20:20.040> last<00:20:20.220> question<00:20:20.430> any<00:20:21.060> more 00:20:21.200 --> 00:20:21.210 align:start position:0% may be one last question any more 00:20:21.210 --> 00:20:26.270 align:start position:0% may be one last question any more questions<00:20:23.210> what<00:20:24.210> about<00:20:24.300> offline<00:20:25.140> when<00:20:26.070> the 00:20:26.270 --> 00:20:26.280 align:start position:0% questions what about offline when the 00:20:26.280 --> 00:20:28.820 align:start position:0% questions what about offline when the devices<00:20:26.940> are<00:20:27.090> being<00:20:27.480> offline<00:20:27.870> so<00:20:28.470> this<00:20:28.680> is<00:20:28.800> a 00:20:28.820 --> 00:20:28.830 align:start position:0% devices are being offline so this is a 00:20:28.830 --> 00:20:31.640 align:start position:0% devices are being offline so this is a great<00:20:29.550> question<00:20:29.810> I'm<00:20:30.810> interested<00:20:30.900> in<00:20:31.410> looking 00:20:31.640 --> 00:20:31.650 align:start position:0% great question I'm interested in looking 00:20:31.650 --> 00:20:34.250 align:start position:0% great question I'm interested in looking into<00:20:31.830> models<00:20:32.610> of<00:20:32.730> sharding<00:20:33.120> and<00:20:33.330> being<00:20:34.170> able 00:20:34.250 --> 00:20:34.260 align:start position:0% into models of sharding and being able 00:20:34.260 --> 00:20:36.260 align:start position:0% into models of sharding and being able to<00:20:34.350> split<00:20:34.590> this<00:20:34.740> up<00:20:34.800> that<00:20:35.520> is<00:20:35.670> one<00:20:35.850> of<00:20:36.000> the<00:20:36.090> big 00:20:36.260 --> 00:20:36.270 align:start position:0% to split this up that is one of the big 00:20:36.270 --> 00:20:38.370 align:start position:0% to split this up that is one of the big benefits<00:20:36.480> of<00:20:37.050> the<00:20:37.230> fully<00:20:37.470> decentralized 00:20:38.370 --> 00:20:38.380 align:start position:0% benefits of the fully decentralized 00:20:38.380 --> 00:20:39.990 align:start position:0% benefits of the fully decentralized -<00:20:38.440> algorithm<00:20:38.890> says<00:20:39.130> that<00:20:39.370> you<00:20:39.610> don't<00:20:39.730> care 00:20:39.990 --> 00:20:40.000 align:start position:0% - algorithm says that you don't care 00:20:40.000 --> 00:20:41.640 align:start position:0% - algorithm says that you don't care where<00:20:40.360> there's<00:20:40.600> a<00:20:40.660> break<00:20:40.870> in<00:20:40.900> the<00:20:41.110> system 00:20:41.640 --> 00:20:41.650 align:start position:0% where there's a break in the system 00:20:41.650 --> 00:20:44.180 align:start position:0% where there's a break in the system those<00:20:42.100> can<00:20:42.309> still<00:20:42.520> reach<00:20:42.700> local<00:20:43.000> optimality 00:20:44.180 --> 00:20:44.190 align:start position:0% those can still reach local optimality 00:20:44.190 --> 00:20:47.100 align:start position:0% those can still reach local optimality when<00:20:45.190> you<00:20:45.400> have<00:20:45.760> a<00:20:45.789> blockchain<00:20:46.059> being<00:20:46.870> able<00:20:46.960> to 00:20:47.100 --> 00:20:47.110 align:start position:0% when you have a blockchain being able to 00:20:47.110 --> 00:20:50.340 align:start position:0% when you have a blockchain being able to bring<00:20:47.770> those<00:20:48.010> to<00:20:48.340> basically<00:20:49.330> forked<00:20:49.840> chains 00:20:50.340 --> 00:20:50.350 align:start position:0% bring those to basically forked chains 00:20:50.350 --> 00:20:53.370 align:start position:0% bring those to basically forked chains back<00:20:51.160> into<00:20:51.520> consensus<00:20:52.059> is<00:20:52.900> something<00:20:53.289> I 00:20:53.370 --> 00:20:53.380 align:start position:0% back into consensus is something I 00:20:53.380 --> 00:20:58.049 align:start position:0% back into consensus is something I haven't<00:20:53.559> yet<00:20:53.770> looked<00:20:53.950> at<00:20:54.070> but<00:20:54.400> want<00:20:54.580> to<00:20:57.059> thanks 00:20:58.049 --> 00:20:58.059 align:start position:0% haven't yet looked at but want to thanks 00:20:58.059 --> 00:21:01.540 align:start position:0% haven't yet looked at but want to thanks Eric 00:21:01.540 --> 00:21:01.550 align:start position:0% 00:21:01.550 --> 00:21:09.130 align:start position:0% [Music]