[00:01.690 --> 00:08.370]  This is my first time attending HOPE, but I've been a hopeful person for quite some time.
[00:09.050 --> 00:13.250]  Welcome to my talk, Computation Comes to Democracy's Aid.
[00:13.530 --> 00:31.390]  In this presentation, I'm going to share a recap of our workshop last night, a few introductory slides about how I connect to this topic, an overview of POLIS, which is open-source software for collective intelligence, some case studies of how that
[00:31.390 --> 00:36.890]  software is naturalizing around the world, and a bit of explanation of how the system is designed.
[00:37.270 --> 00:39.550]  And then I'll close with a bit of a call to action.
[00:40.970 --> 00:56.310]  So, last night, immediately following Mitch's inspiring presentation about the power of hacker communities, 13 of us gathered in the group decision-making and governance workshop to share our own experiences of doing things with people.
[00:57.170 --> 01:01.730]  We talked about how it started, how it went, and maybe how it's still going.
[01:02.850 --> 01:18.310]  People shared stories at all scales, from sitting around a restaurant table, inventing a voting system to choose what appetizers to order, to starting and operating what was for a time the largest platform cooperative in the country, a driver-owned rideshare company.
[01:19.730 --> 01:37.470]  We heard from self-identified benevolent dictators who helped their community collectively acquire a building, and about a community-run food distribution network which grew up during COVID that's now thinking about how to transmit culture over time
[01:37.470 --> 01:41.930]  as old contributors phase out and new people come in to be onboarded.
[01:42.290 --> 01:52.270]  We discussed tactics for getting smarter inside of dumb systems like legacy technology companies or even inside of an unhinged Facebook group.
[01:54.330 --> 02:09.350]  Our conversation last night confirmed what I already knew about technological communities, that the breadth and depth of experience in this audience here today is not only in the realm of technology, it's also in governance.
[02:09.350 --> 02:21.030]  And as much as many of us here hate the very word, let us recognize our extensive experimentation in holding power together to act collectively.
[02:23.450 --> 02:39.310]  Okay, so backing up for a minute, previously on lizbarry.net, in the early aughts, I painted a sign that said, talk to me, and my friend and I sat in lawn chairs on the sidewalks of New York City, getting strangers to talk to each other.
[02:39.750 --> 02:45.890]  Over four years, we listened to over 100,000 people without using any technology.
[02:48.770 --> 02:56.710]  We traveled around the United States bicycling and hitchhiking and threw a party for everyone we met to meet each other.
[02:57.490 --> 02:59.250]  That was in midtown Manhattan.
[02:59.750 --> 03:08.390]  No one was allowed on microphone, not even me, because we had decorated the park with questions to spur self-organization.
[03:10.390 --> 03:20.690]  And in this party, we turned that old cliche, hey baby, talk to me, into an actual talk to me baby born to two strangers who met each other at our party.
[03:22.190 --> 03:32.890]  So after this kind of human relationship experiment, I moved into human relationships with the environment, creating a method for mapping every tree on every sidewalk in New York City.
[03:33.350 --> 03:37.250]  Something that volunteers could do using a measuring wheel and a soft tape measure.
[03:37.950 --> 03:41.350]  And thousands of New Yorkers mapped over three quarters of a million trees.
[03:41.730 --> 03:48.090]  And that spatial inventory is now the Department of Parks' production database and it runs the adopted tree stewardship map.
[03:48.990 --> 04:01.110]  From there, I went on to co-found the Public Laboratory for Open Technology and Science in response to a government-imposed information blackout during the BP oil spill.
[04:01.640 --> 04:06.950]  No scientists, no journalists, were allowed in to document the extent of the oil spill.
[04:07.390 --> 04:09.990]  Only BP corporates were self-reporting.
[04:10.810 --> 04:14.810]  From Public Lab, I moved on to the Gathering for Open-Science Hardware.
[04:14.810 --> 04:16.710]  I think we have a few members here today.
[04:17.090 --> 04:24.710]  The GOSH community, they then elected me into their first community council to help them with governance and collective action.
[04:27.110 --> 04:37.950]  So, my involvement with computational democracy dates back to 2014 when I went to Taiwan to present about Public Lab's open science collaboration.
[04:37.950 --> 04:41.070]  I went to the Open-Source Developers Conference.
[04:41.570 --> 04:48.070]  But as soon as I landed in Taipei, my contact said, don't go to your hotel, come to parliament.
[04:48.350 --> 04:51.930]  I found myself in the midst of Taiwan's sunflower revolution.
[04:52.390 --> 05:04.970]  The people of Taiwan had turned out in mass protest and occupied their parliament in response to their own elected representatives, making a trade deal that favored mainland China more than Taiwan itself.
[05:04.970 --> 05:23.570]  So, in these pictures of Taiwan's sunflower movement, I see many, many people expressing themselves with facilitators holding a respectful space for public discourse, routing comments to people who had previously expressed divergent opinions to get
[05:23.570 --> 05:38.030]  their feedback, summarizing the position of the small group after they had discussed for a while, and sharing those notes across the small group to create a platform that spanned tens of thousands of people.
[05:38.750 --> 05:44.450]  And so, what was their take on that controversial issue that made people get in the streets?
[05:45.010 --> 05:59.050]  Well, after these hundreds of thousands of people deliberated and arrived at a different position than their elected representatives, they politely presented that decision to parliament and then tidied up their already tidy encampments and went home.
[06:01.670 --> 06:19.530]  In this moment, the civic hackers, the technologists, I should say, who supported the documentation and live streaming of all these public activities, they reconfigured into a civic hacktivist group known as GovZero, with a slogan of fork the
[06:19.530 --> 06:24.790]  government and a mission of upgrading their government's democratic processes and institutions.
[06:26.030 --> 06:42.290]  And my relationship with Taiwan's GovZero community led to writing about their successful patches to Taiwan's government in my 2016 piece entitled, V.Taiwan, Public Participation Methods on the Cyberpunk Frontier of Democracy, now republished and
[06:42.290 --> 06:43.750]  hosted by Taiwan's government.
[06:43.750 --> 06:46.890]  The core piece... and illustrated.
[06:47.470 --> 06:54.770]  The core piece of software they institutionalized was POLIS, which is the primary topic of my talk today.
[06:56.030 --> 07:16.150]  So, in 2012, inspired by what social media, specifically Twitter, could, but also couldn't do for the Green Revolution in Iran, the 2011 Occupy Movement or Arab Spring overall, two young people named Colin McGill and Chris Small, who lived on an
[07:16.150 --> 07:27.330]  island off the coast of the Pacific Northwest United States, applied their skills in math, data science, and international relations to conceptualize and build POLIS.
[07:28.070 --> 07:35.030]  The question they were pursuing was, how can large amounts of people say new and coherent things together?
[07:35.750 --> 07:45.050]  The mobilizations of that time were bringing hundreds of thousands of people out on the street, but many felt that most of that potential went unrealized.
[07:46.670 --> 07:56.150]  POLIS was designed to patch that lack of coherence by augmenting our collective ability to listen at scale, POLIS's tagline.
[07:57.270 --> 08:09.950]  In 2018, I joined the POLIS creators to co-found the Computational Democracy Project, a nonprofit foundation stewarding POLIS's open-source code and social methods.
[08:09.950 --> 08:29.490]  I've spent my last two years as head of partnerships, working with facilitators, social movements, civil society organizations, journalists, indigenous nations, governments, both young and old democracies, and peace builders to implement listening at scale.
[08:31.210 --> 08:33.730]  So here's POLIS in one slide.
[08:34.990 --> 08:38.870]  There's a card with a single statement that somebody else wrote.
[08:39.410 --> 08:41.790]  You can agree, disagree, or pass.
[08:43.030 --> 08:50.930]  The intermediate result is people clustered into opinion groups and a map of all the groups.
[08:51.630 --> 09:04.150]  The ultimate product is a statement... are statements that people who otherwise disagree, as evidenced by their position in different opinion groups, all yet agree on.
[09:08.420 --> 09:25.260]  POLIS has been connected to policy making, policy review, ballot initiatives, town halls, constructive journalism, citizen assemblies at a national scale, the platform developments for movements and parties.
[09:25.260 --> 09:31.680]  When you need to update the position that originally brought you together, how do you actually update it with thousands of people?
[09:31.680 --> 09:36.200]  And the development of civilian agendas for war-torn nations.
[09:38.240 --> 09:43.640]  So in POLIS, the content is written by people in the conversation.
[09:43.640 --> 09:48.640]  And the meaning is made by people in the conversation.
[09:50.340 --> 09:55.780]  POLIS uses math to help us listen to each other at scale.
[09:56.980 --> 10:12.380]  Because the POLIS system has a goal, which is to find these underlying points of consensus, it uses our statements and the way we vote on them as the materials to reach this goal.
[10:13.040 --> 10:23.220]  POLIS is a weak AI, in that it does not learn across conversation, and it's not reading or making sense out of the content itself.
[10:26.730 --> 10:27.470]  Okay.
[10:27.970 --> 10:30.830]  That was my first warm-up overview.
[10:31.330 --> 10:32.730]  And now I'm going to share some case studies.
[10:35.110 --> 10:36.870]  So, back to Taiwan.
[10:37.590 --> 10:51.130]  About a month after that occupation resolved, a government minister came out on a weeknight to invite the civic hackers to create a participatory process for national rulemaking.
[10:52.030 --> 10:58.310]  They were invited, these technologists were invited, to upgrade democratic process nationally.
[10:59.090 --> 11:03.430]  And one of the tools that GovZero brought in was POLIS, built in Seattle.
[11:03.770 --> 11:09.290]  And having been started in 2012, was ready to be used as movement technology in 2014.
[11:10.010 --> 11:13.730]  The process the Taiwanese created is called VTaiwan.
[11:14.590 --> 11:20.870]  There is a lot of documentation available online about VTaiwan, and in more than 20 books.
[11:21.830 --> 11:24.890]  A young Audrey Tang appears here in the green shirt.
[11:25.150 --> 11:27.370]  If you don't know her, look her up.
[11:27.690 --> 11:33.150]  Perhaps by watching the upcoming documentary on her, called Good Enough Ancestor.
[11:37.230 --> 11:41.130]  POLIS helps parliaments and citizen assemblies reach more people.
[11:41.550 --> 11:43.670]  POLIS helps scale assemblies.
[11:44.630 --> 11:55.490]  So, two years ago, the 84 members of Austria's National Climate Assembly used POLIS to consult thousands more on their draft policy recommendations.
[11:56.810 --> 12:01.250]  Not that many people can actually be seated in a citizens panel.
[12:01.250 --> 12:09.890]  And they serve as, you know, at the pleasure of the executive, which extends them that power.
[12:10.390 --> 12:21.390]  So, adding scale and much more popular involvement adds another kind of legitimacy and more weight than the assemblies have on their own.
[12:30.290 --> 12:40.350]  In Uruguay, POLIS helps strengthen people's hold on ballot initiatives against powerful electives and lobbies.
[12:41.450 --> 12:46.950]  POLIS adds deliberation to referendums, which is a yes or a no.
[12:47.930 --> 13:04.790]  So, in Uruguay, when a political party collapsed over 200 policy positions into a single yes-no ballot referendum, a social movement set up POLIS, and 16,000 people used it to deliberate each policy in extensive detail.
[13:05.330 --> 13:14.530]  The text shows additional open-source software used and modified by the team to track changes.
[13:15.210 --> 13:16.750]  What's the current law?
[13:16.930 --> 13:18.110]  What's being removed?
[13:18.250 --> 13:19.270]  And what's being added?
[13:19.270 --> 13:24.910]  And then invites POLIS to comment on each change in POLIS.
[13:25.510 --> 13:43.890]  And the results here confirm the general sense that by shoving a security and policing provision in among all the other social welfare questions, the political party did successfully play on fears to get people to vote against their own self-interest.
[13:45.230 --> 13:47.490]  Government support was reduced to the poor.
[13:48.770 --> 13:55.230]  And this information about, like, what was the wedge issue here is still being used by social movements.
[13:57.730 --> 14:10.750]  In Slovakia, we are seeing a non-profit partnering with a national newspaper to offer a space to stabilize national discourse against external misinformation attacks.
[14:11.610 --> 14:25.610]  So, on the eastern border of the Eurozone, waves of Russian misinformation and propaganda attacks are driving xenophobic reactions against Slovakia's will to continue as an EU member state.
[14:26.910 --> 14:34.590]  DEMDIS is a civil society organization started by young lawyers who were furious at what was passing as public discourse.
[14:35.170 --> 14:39.130]  They were furious at the shifts in public opinion due to external forces.
[14:39.130 --> 14:44.950]  And they were committed to remaining and helping their country remain an open society and a democracy.
[14:45.370 --> 14:51.850]  They partnered with the national newspaper to host national conversations using Polis.
[14:54.410 --> 15:01.010]  In the United Kingdom, the government created Policy Lab, which is ten years old already.
[15:01.310 --> 15:03.410]  Their use of Polis dates from 2020.
[15:03.410 --> 15:18.610]  Their work with Polis began after the National Security Secretariat, NSS, had a government-identified need to increase diverse thinking in innovation and policy making.
[15:20.250 --> 15:24.710]  And SPARC stands for National Security Policy and Research CoLab.
[15:24.710 --> 15:42.030]  And this slide has fish instead of security topics on it, because last year they tackled, pun intended, up to 90 ethnographic interviews, hundreds of Polis participants, and brought them all together through a pop-up research and co-design process
[15:42.030 --> 15:50.970]  where 1,400 people contributed to end-to-end policy making on an extremely complex environmental regulatory package.
[15:55.100 --> 16:08.020]  Working across Pakistan, Bhutan, and Timor-Leste, over 30,000 youth deliberated to set the youth agenda that UNDP could bring to COP26.
[16:09.280 --> 16:13.440]  Here we're seeing Polis enabling multinational deliberation.
[16:14.340 --> 16:32.940]  So country teams from UNDP consulted over 30,000 youth, reaching them through sending over 10 million text messages, and asked them what is their vision on economic development in the context of a changing environment as they're entering the workforce.
[16:34.000 --> 16:50.760]  This project also invested in modern electric rickshaw tuk-tuks that offered free rides to anyone who would talk to the youth facilitator in the back seat, and learn about this question that was being asked.
[16:51.020 --> 16:54.480]  The facilitators would draw out youth until they were...
[16:54.480 --> 16:58.100]  they understood what the opportunity was and began sharing their thoughts.
[16:58.320 --> 17:06.200]  And then they would explain how the youth could access Polis on their own devices and drop into a multinational conversation.
[17:06.200 --> 17:09.220]  And this was very popular.
[17:09.600 --> 17:18.040]  Even some members of parliament made cameo appearances as the facilitators you might meet while you jump in to get a free tuk-tuk ride.
[17:21.210 --> 17:26.170]  So today, two nation states... and I have another one highlighted.
[17:26.750 --> 17:35.050]  Two nation states are running Polis as open-source, public agenda setting, and policy crafting technology.
[17:35.050 --> 17:38.790]  In other words, as national democratic infrastructure.
[17:39.450 --> 17:42.970]  This is Taiwan, and now the United Kingdom.
[17:44.070 --> 17:50.810]  Coming right behind them is the Netherlands, which has passed a law allowing Polis to be used in rulemaking.
[17:51.850 --> 18:00.170]  Finland has not institutionalized this yet, but has already run a conversation that had one million votes in it.
[18:01.370 --> 18:08.330]  That's half of the largest known conversation to date, which was in Germany, with two million votes.
[18:09.790 --> 18:21.970]  The nation of Canada, by that I mean programmers who work for the state, have already contributed code for accessibility with screen readers, as well as translation support.
[18:22.710 --> 18:33.450]  And so here we're seeing, as we see in other open-source software packages, but here very explicitly in terms of international geopolitics.
[18:33.810 --> 18:44.930]  The Polis code base is a site of international and multinational cooperation to make democracy possible in these times of complexity.
[18:48.620 --> 18:51.260]  Okay, that was my case study section.
[18:51.640 --> 18:52.380]  Thank you.
[18:58.270 --> 19:03.550]  Now I'm going to go into a little bit of, like, what's going on inside of Polis.
[19:03.710 --> 19:06.730]  How is it possible that all these people can do this together?
[19:07.930 --> 19:12.930]  So, told in emojis, Polis is a utility that makes a map.
[19:14.250 --> 19:16.410]  Polis makes a map of the opinion landscape.
[19:16.610 --> 19:18.470]  It doesn't provide an answer.
[19:19.390 --> 19:24.570]  This map contains a lot of people, a lot of issues, a lot of opinions and ideas, and makes them coherent.
[19:25.810 --> 19:29.970]  Now, sometimes I switch the emojis, and instead of a general gear, I put a plane.
[19:30.770 --> 19:32.570]  And I really abuse the metaphor.
[19:32.850 --> 19:45.230]  So, this slide is addressed mostly to people who would like to use Polis, not the technical developers who would want to work on the code itself.
[19:45.430 --> 19:50.250]  So, this is, like, addressed to social practitioners, if you can follow me in that.
[19:50.250 --> 19:56.210]  So, using an airplane as a metaphor, learning to work with Polis is like going to flight school.
[19:57.130 --> 20:01.630]  Flying a plane requires a pilot who has put time into receiving training.
[20:02.330 --> 20:07.230]  And that training often means flying with a co-pilot instructor for a while.
[20:07.610 --> 20:10.810]  That's the kind of work I help people do all over the planet.
[20:12.570 --> 20:15.050]  When you're in a plane, you've got a view from above.
[20:15.050 --> 20:22.550]  That provides this overhead view of the landscape similar to the map you get of the opinion landscape in Polis.
[20:23.130 --> 20:31.950]  And professionally, as pilots stay in touch with their flight schools to receive ongoing professional development, renewal of certifications.
[20:32.610 --> 20:47.190]  The Computational Democracy Project provides Polis facilitators with updates on the latest developments, what's been made, where in the world that improves Polis, changes in regulation, and emerging best social practices.
[20:49.210 --> 20:56.490]  My last abuse of this metaphor is that, with a plane, one person can help many people move together to where they want to go.
[20:56.490 --> 21:05.450]  And so, this is analogous to how Polis can speed up and scale your conversations to coherently include thousands in deliberation.
[21:07.570 --> 21:19.290]  So, to begin with, someone clicks a link and gets into a Polis conversation, where the first thing shown is a statement that somebody else wrote.
[21:19.290 --> 21:23.250]  And you have the option to agree, disagree, or pass.
[21:23.590 --> 21:28.650]  Or, write your own for others to see and agree, disagree, or pass.
[21:31.450 --> 21:33.530]  Agree is marked as one.
[21:34.330 --> 21:36.010]  Disagree as negative one.
[21:36.230 --> 21:37.850]  And pass is zero.
[21:40.510 --> 21:43.830]  These responses accumulate into a voting record.
[21:43.830 --> 21:47.250]  Here is one participant's first 25 votes.
[21:47.590 --> 21:49.650]  Agree, disagree, disagree, disagree, pass.
[21:49.950 --> 21:50.390]  Agree.
[21:52.150 --> 22:00.410]  So, as a conversation convener, you can't control the order of the cards, because that's what the machine learning is doing.
[22:01.070 --> 22:04.350]  No two people see the same cards in the same order.
[22:05.110 --> 22:08.030]  And at the same time, the order of the cards is not random.
[22:09.010 --> 22:23.330]  Whatever specific statement one might encounter next is thanks to this comment routing algorithm, which efficiently makes the most of varying amounts of participation, learning from how people respond to each other's comments.
[22:25.050 --> 22:33.850]  For instance, we up-weight comments that seem divisive, so we can understand where groups separate from each other.
[22:34.910 --> 22:40.750]  We up-weight consensus as well to then find out what can cross the groups.
[22:42.270 --> 22:45.110]  If there's passes, those get down-weighted.
[22:45.210 --> 22:54.630]  If recent comments come in and we need to have more people respond to them so we know how to handle them, those get up-weighted but only to certain people, and so on and so forth.
[22:54.730 --> 22:56.830]  This is the logic in the comment routing algorithm.
[22:59.010 --> 23:03.890]  So, this builds up into an opinion matrix of what everyone thinks about what everyone else said.
[23:05.050 --> 23:09.610]  If you see that some cells are empty, that's because not all statements will be seen by everyone.
[23:10.730 --> 23:12.190]  Those cells are left blank.
[23:12.790 --> 23:20.050]  Statements that are performing exactly the same as other statements, only one will be used and reused.
[23:22.330 --> 23:35.290]  So, in this conversation in Germany, the largest conversation known so far that I referenced earlier, you can imagine a virtual stadium of 33,000 people interacting almost 2 million times.
[23:36.210 --> 23:42.870]  On average, each person encountered and responded to the statements of 58 other people.
[23:42.870 --> 23:44.530]  That's a lot of listening.
[23:45.030 --> 24:01.650]  Even at this conference where I've been very excited to connect with people, even the person who all those many years ago hosted my talktome.org website for free, thank you for being in the audience today, I still haven't talked to 58 people.
[24:04.850 --> 24:08.110]  And then, people who think alike are grouped into clusters.
[24:09.630 --> 24:12.230]  High dimensionality becomes digestible.
[24:13.110 --> 24:21.270]  It feels like the statements that win high levels of consensus feels like finding surprise breakthroughs in the midst of polarization.
[24:23.450 --> 24:25.230]  Here's another look at the clusters.
[24:27.110 --> 24:31.370]  You can look at the clusters to see which statements define them.
[24:32.070 --> 24:37.310]  Each group is likely to be defined by a different set of statements.
[24:37.770 --> 24:43.230]  There's not one statement that is the single axis of polarization.
[24:43.790 --> 24:45.290]  Things are more complex than that.
[24:45.850 --> 24:49.530]  So, in this way, sometimes I say groups are differently different from each other.
[24:52.760 --> 24:58.040]  Some of the graphs that the automatic report makes are the bullseye and bee swarm.
[24:58.900 --> 25:01.380]  They both order statements by consensus.
[25:02.000 --> 25:05.020]  So, these are great for a quick overview.
[25:05.400 --> 25:13.000]  So, this is kind of a first stop to see in the bullseye, the statements in the center are most highly consensed upon.
[25:13.260 --> 25:18.600]  And as they fall to the edges of the outer bullseye, they're more divisive.
[25:19.700 --> 25:31.280]  In the bee swarm, which is this one, you could just quickly scroll from the left to see the statements that are highly consensed upon and just quickly get a preview of those statements below.
[25:31.780 --> 25:35.300]  And they're more divisive as they go to the right.
[25:37.520 --> 25:43.440]  But it's really the group-informed consensus that is the value that POLIS provides.
[25:45.880 --> 25:55.940]  In this, we're seeing how if you sort by group-informed consensus, the bars where different groups across the top...
[25:55.940 --> 26:01.660]  See, there's an overall column, and then there's group A, group B, group C, and group D.
[26:02.360 --> 26:03.040]  Oh, I see.
[26:03.140 --> 26:04.000]  My GIF isn't...
[26:04.000 --> 26:04.840]  Okay.
[26:05.140 --> 26:06.440]  My GIF is not playing.
[26:06.880 --> 26:21.460]  So, this is shown after you sort by group-informed consensus, where it ranks statements that all the groups, which are otherwise different, have high levels of agreement on.
[26:21.760 --> 26:37.380]  Before you sort by group-informed consensus, the disagree bars are highly varied across the different groups, which in the automated report will be between two and five, just as a basic.
[26:37.380 --> 26:41.360]  But with additional data science, you can do further exploration.
[26:44.460 --> 27:01.500]  The areas of uncertainty provide insight into the gray areas where people haven't made up their minds, which, if you're convening this conversation, if you're helping people get collectively smarter together, this is an area where people are maybe
[27:01.500 --> 27:05.600]  letting you know that they don't know enough to make an informed opinion.
[27:05.600 --> 27:14.960]  This can add efficiency to the educational resources or informational support you might want to provide to your participants.
[27:16.720 --> 27:26.880]  I've heard that the field of conflict transformation refers to these type of areas as windows, like windows of opportunity.
[27:29.380 --> 27:44.240]  And metadata in POLIS emphasizes lived experience, like, I do not have sick leave at my job, or I take care of a dependent, or my family has lived here for three generations.
[27:45.340 --> 27:49.760]  It works less well if you try to force-fit a multiple-choice question.
[27:51.940 --> 28:00.880]  So just, you think about treating your demographics differently in POLIS, or putting a survey in front of it.
[28:04.650 --> 28:15.310]  So, the interface is available in 22 languages, although since I made these slides, a 23rd has been added, which is TAMO in India.
[28:15.310 --> 28:19.990]  And we support right to left.
[28:20.390 --> 28:21.050]  We just...
[28:21.050 --> 28:43.050]  I'm just back, I guess, five weeks now from Ethiopia, where I supported a conversation that was running in Arabic, where thousands of people showed up day after days for five days in a row, remotely participating in a convention to form a civilian
[28:43.050 --> 28:45.230]  agenda in a war-torn nation.
[28:48.640 --> 28:54.960]  This is the third-party plug-in for translation that the Government of Canada built into the statement cards themselves.
[28:55.260 --> 28:55.940]  Thank you, Canada.
[28:58.420 --> 28:59.060]  And...
[29:00.440 --> 29:02.900]  So, while the interface...
[29:02.900 --> 29:10.020]  The user experience is always going to be that your participants see one card at a time, but think about what's...
[29:10.020 --> 29:20.000]  If you're convening a conversation, you also need to be thinking about the social methods as much as understanding the technical methods that underlie it.
[29:20.240 --> 29:25.260]  You'll have a bit more on your mind than the people who are doing the agree, disagree, or pass.
[29:25.680 --> 29:30.420]  For instance, you might be thinking about outreach.
[29:30.800 --> 29:32.500]  How am I going to reach these people?
[29:32.620 --> 29:34.500]  How am I going to construct my public?
[29:35.560 --> 29:39.000]  How will I get it to my audience?
[29:40.600 --> 29:44.120]  Can I partner with a media outlet and have them host it?
[29:44.200 --> 29:46.260]  Or am I doing direct outreach in some way?
[29:46.380 --> 29:49.660]  Am I going through a multi-stakeholder approach?
[29:50.740 --> 29:52.660]  How's the question going to be framed?
[29:52.820 --> 29:54.020]  Who needs to buy into that?
[29:54.420 --> 29:56.760]  What are my moderation guidelines going to be?
[29:57.460 --> 30:00.240]  Do I want to phase this conversation?
[30:00.240 --> 30:08.260]  And, without a doubt, all those charts and graphs I just showed earlier will need different interpretations for different audiences.
[30:09.960 --> 30:21.560]  I want to mention here that because context is so important for effectively using POLIS at the scale I presented in those case studies, 10,000, 30,000 people.
[30:21.940 --> 30:39.920]  I want to say that it's actually easier for communities or movements to use it amongst themselves than it is for some kind of outside institution to go use it on or on top of a population that they themselves are not a member of.
[30:39.920 --> 30:48.820]  In this way, it's not like traditional polling, which can be used for extracting information.
[30:51.920 --> 31:09.360]  If you find yourself interested in how LLMs can be used to ease that human labor I just described, all those social areas of art I just talked about, perhaps you might want to check out this white paper on Archive by Chris Small, Colin McGill, some
[31:09.360 --> 31:13.140]  amazing engineers from Anthropic, and in very small part by myself.
[31:13.900 --> 31:25.780]  If you've heard of AI alignment assemblies, AI alignment assemblies in the past year, many of them have been supported by POLIS, which is an AI but is not an LLM.
[31:30.440 --> 31:45.480]  In 2022, POLIS co-creator Colin McGill worked behind the scenes with Twitter's machine learning engineers who, when they looked around to see who else was working on the problem of coherence in online interactions, they found POLIS.
[31:45.840 --> 31:52.900]  And the night before Elon took possession, Twitter engineers launched Birdwatch, now community notes.
[31:52.900 --> 32:04.720]  So if you've seen the message, readers out of context, they thought people might want to know, on Twitter, now X, you've seen a POLIS-like system of bridging algorithms.
[32:05.720 --> 32:15.640]  Bridging algorithms is the name that Harvard researchers have applied to this class of math, and they've named POLIS as foundational example.
[32:16.420 --> 32:20.080]  Chris Small is recognized as having written the first bridging algorithm.
[32:21.120 --> 32:23.920]  And in this white paper...
[32:23.920 --> 32:25.980]  Oh, I see I'm a little low.
[32:26.180 --> 32:29.720]  We wrote up our vision for deliberation at scale.
[32:29.980 --> 32:40.480]  Should POLIS be able to run on any and all topics in a social media environment, not just on a piece of potential misinformation?
[32:43.770 --> 32:54.490]  And that's called coherent mode for the world's public square, which was Elon's tagline for what he wanted to see.
[32:56.690 --> 33:00.790]  So I'm going to return here to my one-slide overview of POLIS.
[33:02.910 --> 33:07.810]  To recap, cards have a single statement written by somebody else.
[33:07.810 --> 33:11.830]  You have the opportunity to agree, disagree, or pass.
[33:12.670 --> 33:15.450]  The clusters are intermediate results.
[33:15.710 --> 33:20.110]  They show groups of people who think similarly but differently from each other.
[33:20.410 --> 33:25.830]  And the ultimate product of POLIS are the group-informed consensus statements.
[33:25.830 --> 33:35.670]  These are statements that diverse, statements that groups of people who usually disagree on, all yet agree.
[33:38.450 --> 33:47.930]  So told in a meme version, it would be like a user sees statements on cards.
[33:48.710 --> 33:54.270]  And then the cards are moved among the participants by machine learning.
[33:54.270 --> 34:02.530]  That builds up to a matrix decorated with rainbows that is high-dimensional opinion space.
[34:03.210 --> 34:16.170]  And then advanced statistics, like principal component analysis and K-means, are used to detect structure and project it down to 2D, shown on the far right.
[34:16.170 --> 34:17.230]  Oh, very small.
[34:19.110 --> 34:26.390]  And statistics are also used to determine the representative statements for each group.
[34:27.990 --> 34:36.410]  That gives the users something to see about their model of themselves.
[34:37.770 --> 34:45.150]  And finally, the group gets these group-informed consensus statements, which are fantastic for agenda setting.
[34:50.890 --> 34:58.810]  So, we see the potential to lift the lid on group self-determination through two parts that go hand-in-hand.
[34:59.230 --> 35:01.010]  Agenda setting by the public.
[35:01.010 --> 35:01.750]  The public.
[35:01.930 --> 35:03.470]  Made possible with.
[35:08.740 --> 35:10.720]  All the people can set the agenda.
[35:11.080 --> 35:12.580]  Polis has shown this.
[35:13.400 --> 35:23.780]  And not be dependent on politicians or blocked by those who would claim to be their leaders when supported by open-source technology for coherent deliberation at scale.
[35:25.160 --> 35:40.280]  Immediately, it must be said that any technology that would be placed between citizens, among citizens, or between citizens and their government, ought to be open-source to enable meaningful oversight and ensure that every investment makes the system
[35:40.280 --> 35:42.580]  better for everyone in perpetuity.
[35:44.380 --> 35:50.740]  So, I'm part of a community and movement, as you all are here, championing open-source technology.
[35:50.740 --> 35:53.940]  Me, specifically, I champion this for the public sector.
[35:54.240 --> 36:04.260]  So, if any of you find yourself in the position to make technology choices or support someone who's making technology choices, perhaps you share these ethical principles.
[36:04.560 --> 36:08.740]  So that vendors can't hold the government hostage, choose open-source.
[36:09.080 --> 36:12.960]  So that investment by one is investment for all, forever.
[36:13.480 --> 36:17.120]  Choose infectious general public license, open-source.
[36:19.520 --> 36:25.480]  To enable independent evaluation and security testing, publish your algorithms.
[36:26.780 --> 36:35.420]  For participant safety, access to your own data records, as well as the right to the full results, which is not what polling companies give you.
[36:35.620 --> 36:36.660]  Choose open-source.
[36:37.580 --> 36:58.320]  And my closing statement is an invitation to consider how Taiwan's open-source developers conference, reconfigured themselves as GovZero with a mission of making government fit for democracy in these trying times.
[36:59.700 --> 37:17.160]  If those of you listening to me may identify as anarchists, but in these times where large groups of people are falling for oversimplified messages by autocrats, consider that the tools you already work on could help large groups of people be
[37:17.160 --> 37:20.440]  different, stay different, and stay together.
[37:21.280 --> 37:22.640]  Democracy needs you.
[37:23.460 --> 37:26.180]  Thank you for listening, and I'd like to know what you think.
[37:40.490 --> 37:43.250]  If there's any questions, we could chat.
[37:44.530 --> 37:45.730]  Oh, yes, hi.
[37:45.730 --> 37:46.670]  Is this on?
[37:47.630 --> 37:48.170]  Should it be?
[37:48.270 --> 37:48.870]  Is this on?
[37:54.160 --> 37:56.360]  I can hear you and repeat your question.
[37:56.460 --> 37:56.680]  Yes.
[37:59.540 --> 38:01.800]  Yes, it is.
[38:02.400 --> 38:04.520]  Okay, Jolly McPhee, Internet Society.
[38:05.300 --> 38:12.020]  So, first I'd like to compliment you on the presentation and say it's the best, you know, explanation of polis I've ever seen.
[38:16.440 --> 38:20.960]  I look forward to publishing the video widely.
[38:22.280 --> 38:37.660]  And then, I'm sorry I had to nip out during the early part of your presentation, but I remember that there was an example of this, a demo of this, in here in New York City, when Audrey Tang came and the prime produce with Vita NYC.
[38:37.900 --> 38:42.560]  Yes, I was an organizer supporting the Taiwanese government to visit.
[38:42.560 --> 38:43.420]  Yes, 2018.
[38:44.540 --> 38:49.160]  And so, we did an actual demo of the whole thing on the L train shutdown.
[38:51.000 --> 38:54.500]  Yeah, so, you know, it has happened in New York.
[38:54.840 --> 39:00.220]  And I sent that out to Zoom, so it's not published on the web.
[39:00.420 --> 39:03.400]  But I have the video very deep, I can tell you.
[39:04.160 --> 39:10.840]  But afterwards, the Taiwan government did an open reception and Audrey Tang made a speech.
[39:10.840 --> 39:12.720]  And it was a very good speech.
[39:12.980 --> 39:24.200]  And if you want to see this speech, if you Google Tang, Vita NYC, or one word, and video, it's the third thing that comes up.
[39:25.940 --> 39:26.680]  Thank you.
[39:26.820 --> 39:26.940]  Thank you.
[39:26.940 --> 39:29.320]  And thank you for all your recording, all these years.
[39:31.380 --> 39:37.000]  My question is, how are you building Polaris so that it can avoid censorship?
[39:37.400 --> 39:51.460]  So, like, say if there was something in that thing that the Taiwan government really didn't want people thinking about or hearing, how would you, how are you building it so that, you know, it can't be controlled by the actor who's running the software?
[39:55.350 --> 39:56.470]  Good question.
[39:57.190 --> 40:05.250]  We're interested in working with people to strengthen the accountability of each conversation.
[40:05.250 --> 40:13.830]  Right now, the conversation owner will, as a retail user, be able to moderate out any statement that they want.
[40:13.990 --> 40:18.930]  Without some, yeah, we're looking to add a little more legitimacy there.
[40:33.300 --> 40:35.840]  Yeah, would you be able to repeat your question?
[40:35.880 --> 40:36.540]  Yes, sorry.
[40:36.840 --> 40:37.020]  Thanks.
[40:38.540 --> 40:42.900]  How are you building Polaris so that it can avoid censorship?
[40:43.300 --> 40:48.840]  So, like, say there was something that people were talking about in Taiwan that the government really didn't want them talking about it.
[40:48.840 --> 40:55.780]  How is, how are you building it so that that can't be censored, but at the same time you can censor spam and cyber attacks?
[40:57.340 --> 41:05.500]  So, I referenced that we're looking to strengthen visibility over the moderation practices that people take.
[41:05.780 --> 41:16.580]  But I'll also say that it's possible for, like, for a long time, it was actually the civic hackers social movement themselves that were hosting it for the government.
[41:16.580 --> 41:20.720]  And the government said, we are using this, but the government wasn't hosting it.
[41:21.120 --> 41:23.740]  And there's a lot of options.
[41:24.180 --> 41:36.200]  I hope that in the case study I tried to convey that there's a lot of variations, there's a lot of options and choice that you have with your overall process design.
[41:36.200 --> 41:44.440]  Polaris is really like a coherence widget that can be put in various institutions held differently.
[41:44.860 --> 41:49.220]  And also it can just be one part of your overall participatory process.
[41:50.080 --> 41:51.380]  Thank you for your question.
[41:51.980 --> 41:53.520]  Maybe you can help us strengthen it.
[41:54.640 --> 41:55.420]  Hi.
[41:55.720 --> 41:55.740]  Hi.
[41:55.740 --> 42:13.940]  I was curious that, like, one thing that in popular culture is, like, the idea of, like, horseshoe theory, where, like, the opinion space wraps around on itself and formed, like, topologically a circle or something.
[42:13.940 --> 42:31.380]  But, like, have there been any cases where, like, k-means clustering is insufficient and the data space more resembles something more topologically less linear or nonlinear?
[42:32.080 --> 42:41.600]  And that, like, any looking into, like, topological data analysis?
[42:42.260 --> 42:45.720]  So, for that, I'm going to refer you to Christopher Small.
[42:46.040 --> 42:46.420]  Mm-hmm.
[42:46.900 --> 42:47.160]  Yeah.
[42:47.440 --> 42:49.780]  The data science co-creator of Polis.
[42:50.200 --> 42:50.440]  Gotcha.
[42:50.780 --> 42:51.060]  Thank you.
[42:51.360 --> 42:51.960]  Is he here?
[42:52.980 --> 42:54.180]  No, but he's online.
[42:54.520 --> 42:54.820]  Okay.
[42:57.640 --> 42:59.300]  Thank you very much for your presentation.
[43:00.060 --> 43:04.960]  I'm an active member of a student union, back where I'm from, Montreal.
[43:06.340 --> 43:07.880]  Our quorum is 1%.
[43:07.880 --> 43:10.040]  We have a democracy problem.
[43:10.480 --> 43:21.740]  So, I'm wondering if you have case studies that you can refer to or talk about of the usage of the software in worker unions or student unions?
[43:25.000 --> 43:29.980]  I'd actually like to work with you to make our first case study in a student union.
[43:30.860 --> 43:39.960]  It's a very lightweight to enable asynchronous participation on people's own time, own terms, and their own devices.
[43:39.960 --> 43:47.360]  So, even if you'll never... like you could get 500 people into your polis who would never actually show up synchronously in person.
[43:48.020 --> 43:52.120]  And does your college even have a room big enough that the students could use for that?
[43:52.320 --> 44:04.320]  So, there's lots of fun ways to mash up remote asynchronous participation with in person, like before, during, and even as follow-ups.
[44:04.320 --> 44:08.120]  And yeah, so, yeah, that would be a really fun project.
[44:08.440 --> 44:08.760]  Cool.
[44:09.840 --> 44:11.380]  Thank you very much for your presentation.
[44:11.680 --> 44:12.660]  I've really enjoyed it.
[44:14.300 --> 44:20.820]  Was there ever a consideration for how to connect these clusters to each other, like within themselves?
[44:21.540 --> 44:30.800]  Something about the way the algorithm was depicted was reminding me a lot of the old OkCupid days, where you take these opinion or personality quizzes, and that's how they match you up with another person.
[44:31.640 --> 44:36.100]  Could this be done to create sub-communities within a given community?
[44:37.700 --> 44:42.720]  Oh, that really reminds me of, yeah, or talk to me, baby.
[44:45.380 --> 44:47.160]  With the OkCupid reference.
[44:48.400 --> 44:49.220]  You know what?
[44:49.360 --> 44:54.040]  The automated report is meant to be both...
[44:54.040 --> 45:06.960]  So, it's meant to thoroughly reveal the statements that have one group-informed consensus, and to give a simple basic insight into the groupings.
[45:07.860 --> 45:16.280]  If you want to look more into the groupings, you can do additional exploration on the full data set.
[45:17.440 --> 45:24.320]  But the automated report will not show you how groups bridge in the cluster diagram.
[45:24.560 --> 45:28.600]  It will only show you the statements that bridge.
[45:30.000 --> 45:30.720]  Yes.
[45:30.980 --> 45:31.420]  Thank you.
[45:31.940 --> 45:32.340]  Cool.
[45:32.740 --> 45:34.360]  You kind of answered the question.
[45:34.580 --> 45:43.480]  Are the case study results made public, and are they minable by the public to get their own opinions about the information that's being gathered?
[45:43.480 --> 45:44.880]  Ah, yes.
[45:45.160 --> 45:59.860]  So, there is... we do maintain a GitHub repository of open data, which has the original example of when Taiwan regulated...
[45:59.860 --> 46:11.140]  Taiwan used a national multi-stakeholder deliberation process, both that happened in person and on Polis, to regulate Uber.
[46:11.140 --> 46:21.080]  I didn't mention the Uber case specifically, but that's what the slide was illustrating, and that's really the case that broke...
[46:21.080 --> 46:30.260]  I wrote about that in 2016, and it really broke out Polis into the... into those who are paying attention to civic tech into significance.
[46:32.320 --> 46:45.600]  The data, the full data from that case, and some others, like the national assemblies that Austria had, and a few others, are hosted as a GitHub repo, where you can do further exploration.
[46:46.060 --> 46:46.720]  Thank you.
[46:47.020 --> 46:47.740]  Yeah, thanks.
[46:49.080 --> 46:50.560]  I had a two-part question.
[46:50.720 --> 46:54.540]  The first one was about how people access the technology.
[46:54.800 --> 46:59.280]  Is it like non-smartphone text messages, a cell phone, tablet?
[46:59.760 --> 47:03.780]  How are they getting access to the Polis system?
[47:04.260 --> 47:05.500]  Like your participants?
[47:05.740 --> 47:06.380]  How would they get it?
[47:06.460 --> 47:06.960]  Yeah, yeah, yeah, yeah.
[47:07.480 --> 47:08.780]  You need to get them a link.
[47:09.560 --> 47:11.780]  A link meaning an Internet-connected device?
[47:12.320 --> 47:12.920]  Yes.
[47:13.060 --> 47:13.520]  Right.
[47:13.740 --> 47:28.880]  Yes, and it would be really exciting to extend the platform so that people can just be texting into the conversation without having to go to, like, where it's hosted in a website and type their answers there.
[47:28.880 --> 47:39.020]  Yeah, so if you're excited about, yeah, making Polis more multi-platform, love to have contributions.
[47:39.440 --> 47:54.040]  And then, it's about who possibly would be excluded, part of, like, a disability, also people who are in carceral institutions like nursing homes, prisons, psychiatric, you know, institutions, people who are in, or wouldn't have the ability to go to
[47:54.040 --> 48:01.800]  any other, you know, place that would have a cell phone, or maybe in, like, countryside areas, you know, or how do you get those people part of your process?
[48:02.640 --> 48:04.360]  Thank you so much for asking that.
[48:04.540 --> 48:06.820]  There's a couple of things to share in that regard.
[48:08.540 --> 48:14.400]  First of all, that actually it's not true that everywhere in the world there's a digital divide.
[48:14.400 --> 48:23.700]  For instance, in Taiwan, they were able to run with this because they're so well-provisioned in bandwidth on their island.
[48:25.420 --> 48:30.340]  And so it doesn't... it's no surprise that that was the first huge case study.
[48:33.920 --> 48:38.240]  Secondly, Polis can be brought into low-bandwidth areas.
[48:39.100 --> 48:44.180]  My understanding is that UNDP has set up tents along dirt roads.
[48:45.140 --> 48:49.240]  There have been kiosks set up in malls in Singapore.
[48:50.220 --> 49:04.200]  The city of Amsterdam found city workers pushing around a coffee cart offering free coffee to anyone who would stand there and, like, click the... and, like, use the Polis on their tablets.
[49:04.200 --> 49:07.000]  We refer to that as liquid democracy.
[49:09.340 --> 49:18.180]  And, yeah, those tuk-tuks, again, were, like, if people didn't have their own phones, those facilitators had tablets and people could just use them then.
[49:18.340 --> 49:21.240]  And we call that democracy in motion.
[49:24.400 --> 49:34.440]  I get to the other part about people who were in, like, nursing homes, jails, you know, facilities where they weren't necessarily, you know, they were very controlled through places, I guess?
[49:36.180 --> 49:42.040]  Well, organizers can always bring devices in.
[49:43.240 --> 49:58.020]  And if there's no... if there's no overt way for that to happen, then I would say you could follow the same pathways that people use to get their communications out of restricted environments currently.
[50:02.470 --> 50:02.990]  Hi.
[50:03.530 --> 50:04.830]  We're in a city here.
[50:05.110 --> 50:13.150]  I'm wondering if Polis is being used successfully in outside cities, metropolitan areas, in rural or country areas.
[50:13.610 --> 50:20.510]  And I'm also wondering if there is a consideration of identity for the so-called civil attacks.
[50:20.990 --> 50:21.410]  Thank you.
[50:21.570 --> 50:21.910]  Okay.
[50:22.090 --> 50:24.590]  We will stay up because I'm not going to remember your second question.
[50:25.030 --> 50:27.290]  So, it's true.
[50:28.290 --> 50:29.250]  We're in a city.
[50:29.390 --> 50:30.810]  Polis has been used here.
[50:30.830 --> 50:46.790]  by the Civic Engagement Commission to actually improve their participatory budgeting process by going upstream and have people debate on Polis what the menu items should be that are available for participatory budgeting.
[50:47.210 --> 50:49.870]  But you're asking about a rural area.
[50:50.650 --> 50:55.630]  I think the main... I think you'd be in an outreach campaign like any others.
[50:55.630 --> 51:03.790]  But you also... there's nothing stopping you from buying ads on social media and distributing it that way.
[51:03.870 --> 51:16.170]  We just had a conversation run in Greece with a thousand people that was dealing with tourism overrunning their... and short-term rentals overrunning their... their housing availability for actual locals.
[51:16.970 --> 51:24.070]  And I... yeah, I think there's no... no upper bound on the creativity you can bring to outreach.
[51:25.110 --> 51:27.450]  Could you repeat your second question again?
[51:28.110 --> 51:28.790]  Yes.
[51:28.790 --> 51:32.930]  I'm wondering about how identity works.
[51:33.190 --> 51:38.170]  Like, if anyone on the Internet can... can be... act as a person in this.
[51:38.170 --> 51:45.330]  Or if there's... or if there's... or if there's some kind of process of... of ensuring uniqueness of accounts.
[51:45.630 --> 51:46.490]  That sort of thing.
[51:47.210 --> 51:47.770]  Yes.
[51:47.950 --> 51:53.870]  Sampling strategy is crucial in high-stakes use of this technology.
[51:53.870 --> 51:55.730]  I mean, sampling strategy.
[51:55.930 --> 51:58.070]  Like, who is coming into your conversation?
[51:59.650 --> 52:05.830]  So... you can... you know, that... you don't have to do any identity management.
[52:05.990 --> 52:13.870]  You can let anyone who accesses the link participate in the poll and rely on the algorithm's own down weighting of...
[52:13.870 --> 52:24.130]  of... when statements aren't... like... when they... when it seems like there is like random activity, you can see that...
[52:24.130 --> 52:26.010]  the system will down weight it.
[52:26.190 --> 52:29.670]  And there's other ways to deal with coordinated influence attacks.
[52:30.410 --> 52:35.190]  But there are also ways to... to actually check for identity on the way in.
[52:35.590 --> 52:38.730]  You can... nothing stopping you from having people register...
[52:38.730 --> 52:40.750]  before they come into the POLIS.
[52:41.250 --> 52:43.030]  Or authenticate in some way.
[52:43.390 --> 52:46.670]  All of that is... that's what... where I talk about POLIS being a utility.
[52:47.390 --> 52:48.870]  You can... it's just one link.
[52:49.330 --> 52:57.450]  You know... Colin speaks of designing POLIS... with the philosophy of UNIX... as pipes.
[52:58.330 --> 53:06.070]  And... if you go and log into POL.IS, which is the home instance hosted by the Computational Democracy Project...
[53:06.070 --> 53:09.310]  you'll see it's pretty minimal... inside there.
[53:09.950 --> 53:12.450]  What you don't... what you won't see...
[53:12.450 --> 53:19.050]  is... I think... what these excellent audience questions are drawing out from me...
[53:19.050 --> 53:22.090]  in terms of... how are we gonna do outreach in rural areas?
[53:22.150 --> 53:23.430]  How are we gonna manage identity?
[53:23.790 --> 53:25.990]  You know... all... how are we gonna distribute this link?
[53:26.050 --> 53:29.470]  You know... all of those things... are open... to be designed.
[53:29.470 --> 53:38.210]  So... very often... a design process... begins with... mapping your... resources... and skills.
[53:38.770 --> 53:42.370]  Do you have connections to groups... who have large membership reaches?
[53:43.570 --> 53:48.130]  Um... you know... do you have... data scientists on staff... who are gonna be very good...
[53:48.130 --> 53:49.490]  at looking further into the results?
[53:49.850 --> 53:52.510]  Depending on what resources you have...
[53:52.510 --> 53:58.450]  for... running a conversation... your design will be unique... and fit to your context.
[54:00.050 --> 54:00.690]  Hi!
[54:01.130 --> 54:02.690]  Perfect... one more question...
[54:02.690 --> 54:07.530]  Okay... I... I... I believe you... kind of answered... a lot of my question... but... um... so...
[54:07.530 --> 54:15.530]  the purpose of polis... as I understand it... is not so much... it... it... the... the... identity of... the participants...
[54:15.530 --> 54:25.270]  is not so critical... because it's not a voting... concept... it's more of a... uh... an assembling of ideas...
[54:25.270 --> 54:28.970]  and figuring out what ideas people agree on... am I correct?
[54:29.830 --> 54:31.410]  I think so...
[54:31.410 --> 54:55.250]  In other words... it's not... uh... it... it doesn't have the same... um... um... um...
[54:55.250 --> 54:56.910]  is... is... is gonna be true...
[54:56.910 --> 55:00.590]  Oh... like... is this a consultation... or is there binding power?
[55:00.870 --> 55:02.850]  There you go...
[55:03.870 --> 55:08.290]  We... we... expect... to see... polis... connected to binding power...
[55:09.010 --> 55:13.890]  with the kind of robust... process design... and identity management... that people have been asking...
[55:13.890 --> 55:18.150]  questions about... yeah... we expect to see it... and... um... we're out of time... so I'll just say...
[55:18.150 --> 55:24.410]  I really do love to hear from people... 20 years ago... I did sit next to a sandwich board... that said...
[55:24.410 --> 55:31.590]  talk to me... and I have been... designing the culture... and doing community management... for open science...
[55:31.590 --> 55:36.190]  communities... numbering in the tens of thousands... and now I help that same number of people deliberate...
[55:36.770 --> 55:41.130]  so I don't mind a few more emails... I'd love to hear from you... thank you...