WEBVTT 1 00:00:00.000 --> 00:00:11.100 I'm just going to say that because this is the kind of conference that it is with many 2 00:00:11.100 --> 00:00:16.340 streams of information and in some cases multiple sources of truth, you may have seen that the 3 00:00:16.340 --> 00:00:18.220 title of my talk is Landslide. 4 00:00:18.220 --> 00:00:21.420 You may also have seen the title of my talk is Hold Fast. 5 00:00:21.420 --> 00:00:25.660 Both things are true at the same time because I gave Ted the information at different times. 6 00:00:25.660 --> 00:00:35.180 So we're doing Landslide Hold Fast. 7 00:00:35.180 --> 00:00:37.180 One more glitch. 8 00:00:37.180 --> 00:00:38.980 So hi, I'm Erin. 9 00:00:38.980 --> 00:00:44.440 I alternate between trying to figure out what the social Internet is doing to us and trying 10 00:00:44.440 --> 00:00:48.900 to use it to get better information to more people in more places. 11 00:00:48.900 --> 00:00:54.660 So right now I'm doing network thinking and projects at my mini studio, wreckage salvage 12 00:00:54.660 --> 00:01:00.740 and also co-running a networked volunteer mutual aid information project at Unbreaking, 13 00:01:00.740 --> 00:01:05.220 also a project of the RAAF Foundation. 14 00:01:05.220 --> 00:01:08.860 We work to nail down knowledge about what our government is doing to us in the United 15 00:01:08.860 --> 00:01:10.740 States. 16 00:01:10.740 --> 00:01:14.300 Before that I co-founded the COVID tracking project at the Atlantic. 17 00:01:14.300 --> 00:01:22.020 This was before the Atlantic. 18 00:01:22.020 --> 00:01:26.260 That was to help people understand what was happening in 2020 through public data. 19 00:01:26.260 --> 00:01:31.460 Before that I was at a journalism tech community that came out of the Moz Foundation. 20 00:01:31.460 --> 00:01:35.220 Before that it was all over the early web. 21 00:01:35.220 --> 00:01:36.380 So that's the nature of my work. 22 00:01:36.380 --> 00:01:39.740 I use the tools and I try to improve the tools and then I repeat. 23 00:01:39.740 --> 00:01:46.420 What I learned from yesterday's amazing pre-conference experience is that many of you are in much 24 00:01:46.420 --> 00:01:48.340 the same position, which I love. 25 00:01:48.340 --> 00:01:52.780 It's like a whole room of corner cases. 26 00:01:52.780 --> 00:01:57.700 I want to know what the networks are doing to us and what resources that people like 27 00:01:57.700 --> 00:02:04.680 us have to draw from to shape our networks in ways that ground and reorient our communities 28 00:02:04.680 --> 00:02:13.540 in useful knowledge and to maybe someday live together in peace and mutual respect. 29 00:02:13.540 --> 00:02:17.500 Right before the end of last year I posted another extremely long blog post, which is 30 00:02:17.500 --> 00:02:22.220 my jam, called Landslided Ghost Story, in which I tried to get closer to the core of 31 00:02:22.220 --> 00:02:27.340 our network's role in disorienting and disintegrating us. 32 00:02:27.340 --> 00:02:31.220 I'll talk about that one a little bit here because the new work I'm bringing builds on 33 00:02:31.220 --> 00:02:32.780 it. 34 00:02:32.780 --> 00:02:37.700 This talk, I think, and I hope is kind of a turning point for my work and what I can 35 00:02:37.700 --> 00:02:41.460 bring to you because I think I finally have a good enough handle on at least parts of 36 00:02:41.460 --> 00:02:46.460 this to be able to offer some paths forward that aren't either unattainable or just like 37 00:02:46.460 --> 00:02:49.940 disproportionately paltry. 38 00:02:49.940 --> 00:02:51.460 I want to make a side note. 39 00:02:51.460 --> 00:02:56.540 I take it as a given that one major underlying problem that prevents us from dealing with 40 00:02:56.540 --> 00:03:00.340 many other problems is centralization and lock-in, which decentralized networks are 41 00:03:00.340 --> 00:03:01.340 here to deal with. 42 00:03:01.340 --> 00:03:06.180 I just want to say I think that Jay and the rest of Blue Sky's founding team have done 43 00:03:06.180 --> 00:03:11.660 us all a tremendous service by, yes, making Blue Sky the app, but especially in using 44 00:03:11.660 --> 00:03:17.040 all of that available momentum to get this protocol ecosystem off the ground and stable 45 00:03:17.040 --> 00:03:21.180 enough to support the kind of independent development that I saw happening yesterday 46 00:03:21.180 --> 00:03:23.020 and that we'll be talking about here. 47 00:03:23.020 --> 00:03:25.380 I assume basically universal concordance about this. 48 00:03:25.380 --> 00:03:29.020 I'm not going to spend a lot of time talking about it, but I am so grateful for the work 49 00:03:29.020 --> 00:03:30.260 that the team has accomplished. 50 00:03:30.260 --> 00:03:33.180 I just want to say that. 51 00:03:33.180 --> 00:03:36.420 There's so much good stuff happening around the protocol, and I'm super psyched to see 52 00:03:36.420 --> 00:03:39.440 the next batch of it today. 53 00:03:39.440 --> 00:03:40.440 Yesterday was incredible. 54 00:03:40.440 --> 00:03:45.980 The work I have been doing is for you, and I brought a lot of things, so let's get to 55 00:03:45.980 --> 00:03:47.580 it. 56 00:03:47.580 --> 00:03:56.340 62 years ago yesterday, on Good Friday in 1964, a tectonic plate up under Alaska slid 57 00:03:56.340 --> 00:04:01.900 under another tectonic plate and produced an extraordinarily strong and sustained earthquake 58 00:04:01.900 --> 00:04:07.100 that we now know to be the second highest magnitude quake in recorded history. 59 00:04:07.100 --> 00:04:11.900 It actually reshaped the field of seismology, but more immediately it caused parts of the 60 00:04:11.900 --> 00:04:17.900 Alaskan coast to drop by nearly 40 feet, rise by nearly 40 feet, dropped others down below 61 00:04:17.900 --> 00:04:22.940 sea level, triggered enormous tsunamis, and according to geologists, it was so big that 62 00:04:22.940 --> 00:04:26.540 it rang the planet like a bell. 63 00:04:26.540 --> 00:04:32.460 Alaska is not densely populated, but this is what the earthquake looked like in, this 64 00:04:32.460 --> 00:04:33.800 is an anchorage. 65 00:04:33.800 --> 00:04:38.200 In cities and towns, you see the kind of damage you'd expect at tore up streets and knockdown 66 00:04:38.200 --> 00:04:43.420 buildings, and in Alaska, 139 people died. 67 00:04:43.420 --> 00:04:51.500 It also did something exceptionally weird and horrific in the port town of Valdez. 68 00:04:51.980 --> 00:04:58.980 Valdez is the northernmost year round ice free deep water port in North America and 69 00:04:58.980 --> 00:05:04.300 the southern end of the Trans-Alaska oil pipeline. 70 00:05:04.300 --> 00:05:09.260 My partner grew up in Valdez, so I heard about the earthquake and how weird it had been from 71 00:05:09.260 --> 00:05:14.780 him, and my father-in-law worked at the Valdez Museum and had produced a video from some 72 00:05:14.780 --> 00:05:18.820 of this really wild, grainy, contemporary footage of the quake, so when we went up to 73 00:05:18.820 --> 00:05:20.820 visit I got to see that. 74 00:05:21.140 --> 00:05:27.580 Much later, my family moved to the Cascadia subduction zone, and I became an earthquake 75 00:05:27.580 --> 00:05:30.660 nerd, as one does. 76 00:05:30.660 --> 00:05:36.300 I sort of just poked at this story and the records out of Valdez as part of my insomnia 77 00:05:36.300 --> 00:05:40.820 rounds of upsetting things to look at in the middle of the night, but it kind of stayed 78 00:05:40.820 --> 00:05:46.140 in the background until the end of last year when I discovered that the US Geological Survey 79 00:05:46.140 --> 00:05:51.700 had produced a new cut of that weird film footage from the Valdez Harbor. 80 00:05:51.700 --> 00:05:59.380 I'm not going to show it to you, but it is in the landslide post on my site. 81 00:05:59.380 --> 00:06:04.180 The footage only existed because a supply ship, the SS China, had come into the harbor 82 00:06:04.180 --> 00:06:09.500 right before the quake, and someone on the ship happened to have an 8mm camera and happened 83 00:06:09.500 --> 00:06:13.500 to be filming the crew just like dorking around on deck and throwing oranges and candy to 84 00:06:13.500 --> 00:06:17.500 the kids who'd come down to the docks and meet the ship. 85 00:06:17.500 --> 00:06:22.140 But I'd only ever seen the footage in a form that turned out to be highly manipulated, 86 00:06:22.140 --> 00:06:28.620 and the reconstructed cut from USGS revealed that the shots we always thought were of the 87 00:06:28.620 --> 00:06:36.980 water of the local tsunami produced by the quake were actually of the land, which had 88 00:06:36.980 --> 00:06:44.020 just turned to liquid and flowed out from under everything it was supporting. 89 00:06:44.020 --> 00:06:50.660 That's what used to be the harbor, and it all just fell in the sea, and the footage 90 00:06:50.660 --> 00:06:54.340 actually shows it just pouring out from under these buildings. 91 00:06:54.340 --> 00:07:00.060 That re-sequence footage remains the single most upsetting piece of film I have ever seen 92 00:07:00.060 --> 00:07:05.340 because the destabilization it depicts is happening at the level of basic physics, which 93 00:07:05.340 --> 00:07:08.540 is gross and weird. 94 00:07:08.540 --> 00:07:12.580 So here's what happened in Valdez, based on everything we know now. 95 00:07:12.580 --> 00:07:17.700 The town was built on glacial moraine, which is this disorganized debris piled up by a 96 00:07:17.700 --> 00:07:23.420 glacier, and down near the shoreline, that jumble of unconsolidated rock and sand went 97 00:07:23.420 --> 00:07:27.260 down 600 feet before you hit bedrock. 98 00:07:27.260 --> 00:07:31.580 And building on the moraine was not an obviously bad idea because it was still solid ground. 99 00:07:31.580 --> 00:07:35.340 You could step on it or drive on it, and it's fine because the contact forces between 100 00:07:35.340 --> 00:07:38.220 the particles would distribute the load. 101 00:07:38.220 --> 00:07:41.180 But Valdez is also extremely wet. 102 00:07:41.180 --> 00:07:46.220 In the winter, it's no so much people have to shovel their roofs to keep them from collapsing. 103 00:07:46.220 --> 00:07:50.620 And the water table down there by the shoreline is just a few feet from the surface of the 104 00:07:50.620 --> 00:07:51.620 town. 105 00:07:51.620 --> 00:07:57.740 So in late March of 1964, between that just like barely subterranean water and the melt 106 00:07:57.740 --> 00:08:05.660 from these exceptionally heavy snows, that loose jumbled substrate was flooded, was completely 107 00:08:05.660 --> 00:08:07.900 saturated. 108 00:08:07.900 --> 00:08:13.220 And it turns out that when a series of extraordinarily heavy and repeated shocks like this particular 109 00:08:13.220 --> 00:08:18.620 earthquake hit a loose substrate that's also completely saturated with water, the compression 110 00:08:18.620 --> 00:08:26.060 hits the water and pushes all those particles apart, and the solids just turn into liquid 111 00:08:26.060 --> 00:08:29.980 almost instantly and flow away and everything falls in. 112 00:08:29.980 --> 00:08:36.580 And in Valdez, the whole harbor flowed away and took out everything and everyone on it. 113 00:08:36.580 --> 00:08:43.020 So this is the three-part model of how this particularly horrific thing works. 114 00:08:43.020 --> 00:08:48.420 We've seen this elsewhere now, but we never had, especially we never had footage of anything 115 00:08:48.420 --> 00:08:49.420 like this. 116 00:08:49.420 --> 00:08:55.700 There were like some very loose records of what's called soil liquefaction. 117 00:08:55.700 --> 00:08:59.700 I don't recommend that you go look for videos of soil liquefaction. 118 00:08:59.700 --> 00:09:03.780 I was down this rabbit hole in December, the same time I was trying to take stock of almost 119 00:09:03.780 --> 00:09:08.060 a year of work at Unbreaking where we're just kind of down slogging through the mud of our 120 00:09:08.060 --> 00:09:09.060 information landscape. 121 00:09:09.060 --> 00:09:14.300 And I was trying to understand like the networks piece and also the underlying like journalism 122 00:09:14.300 --> 00:09:20.260 shaped stuff that's supposed to produce distributable knowledge and how our collective understanding 123 00:09:20.260 --> 00:09:24.260 was faring against repeated and violent shocks. 124 00:09:24.260 --> 00:09:28.980 So I wrote a lot of words about the information substrate and the saturation and flooding 125 00:09:28.980 --> 00:09:33.660 of network-based overload and how those two things together produce a colossal vulnerability 126 00:09:33.660 --> 00:09:34.660 to shocks. 127 00:09:34.660 --> 00:09:38.580 And I'm going to lay those out super quickly here just to get them on the table so we can 128 00:09:38.580 --> 00:09:41.900 move on to the good stuff. 129 00:09:41.900 --> 00:09:47.780 So starting with the substrate, in the US where I'm from and in many other parts of 130 00:09:47.780 --> 00:09:52.740 the world, our underlying foundational knowledge systems have taken a lot of damage. 131 00:09:52.740 --> 00:09:58.500 Like the business model underlying our media continues down this decades-long death spiral. 132 00:09:58.500 --> 00:10:03.500 And media consolidation which preceded the damage wrought by the internet by many decades 133 00:10:03.500 --> 00:10:09.380 continues to degrade and squeeze and hollow out a lot of our most, well, formerly most 134 00:10:09.380 --> 00:10:13.620 trustworthy sources of current information and basic analysis. 135 00:10:13.620 --> 00:10:15.700 But it's not just journalism. 136 00:10:15.700 --> 00:10:20.220 Extractive practices at every part of the knowledge production chain has dealt tremendous 137 00:10:20.220 --> 00:10:23.420 damage to our whole ecosystem. 138 00:10:23.420 --> 00:10:26.900 I have on my hard drive enough for like a dozen talks about what's happened to basic 139 00:10:26.900 --> 00:10:32.700 research and peer review and academic presses and general presses and even like web search, 140 00:10:32.700 --> 00:10:37.740 the niches that used to support independent writers and artists and some of our most recent 141 00:10:37.740 --> 00:10:42.660 knowledge consolidating institutions like Stack Overflow and Wikipedia are beginning 142 00:10:42.660 --> 00:10:44.540 to suffer. 143 00:10:44.540 --> 00:10:50.100 And all of this hollowing out has been accompanied by an upwelling of things that we wouldn't 144 00:10:50.100 --> 00:10:56.820 traditionally call news that range from genuinely vital eyewitness videos through outright and 145 00:10:56.820 --> 00:11:02.260 well-funded propaganda farms to purely fact-free slop. 146 00:11:02.260 --> 00:11:09.540 And all of this is enmeshed with the news organizations primarily devoted to discrediting 147 00:11:09.540 --> 00:11:14.980 real reporting and presented through social platforms that in many cases deliberately 148 00:11:14.980 --> 00:11:19.780 shape the subset of available information to evade responsibility for misinformation 149 00:11:19.780 --> 00:11:25.400 and also to suit sort of our own worst selves desires. 150 00:11:25.400 --> 00:11:28.140 So this is not great. 151 00:11:28.140 --> 00:11:34.220 And then we also have just the massive proliferation of information that I feel like we've just 152 00:11:34.220 --> 00:11:38.100 about lost a way to talk about maybe because we're so buried in it. 153 00:11:38.100 --> 00:11:40.020 The science folks bring it up. 154 00:11:40.020 --> 00:11:44.460 And that was extremely heartening to hear them acknowledging that's what they do, that 155 00:11:44.460 --> 00:11:46.020 the knowledge crafting. 156 00:11:46.020 --> 00:11:52.940 But this phenomenon of exponential hyperproliferation and information overload is a real one with 157 00:11:52.940 --> 00:11:58.540 a large and solid body of evidence about its genuinely damaging psychological and emotional 158 00:11:58.540 --> 00:12:01.020 and physical effects. 159 00:12:01.020 --> 00:12:05.940 And we've been making filters and developing practices to cope with this with too much 160 00:12:05.940 --> 00:12:11.900 information since we invented writing and complaining about it the whole time. 161 00:12:11.900 --> 00:12:15.900 The history of information overload is amazing. 162 00:12:15.900 --> 00:12:22.060 But our latest filters including web search and social media algorithms and now the LLMs 163 00:12:22.060 --> 00:12:28.300 are not necessarily filtering our flood in ways that actually serve us well. 164 00:12:28.300 --> 00:12:33.860 And many of us too have lost faith in some of the sort of stolid older filters because 165 00:12:33.860 --> 00:12:39.020 we can see that they're not representing the reality that we find in eyewitness videos 166 00:12:39.020 --> 00:12:40.900 and first person accounts. 167 00:12:40.900 --> 00:12:45.700 So maybe we decide to let go of those filters too and the more filters we opt out of the 168 00:12:45.700 --> 00:12:48.900 more we expose ourselves to the flood. 169 00:12:48.900 --> 00:12:54.460 But at least when we do that we're getting access to more knowledge. 170 00:12:54.460 --> 00:12:59.020 So this is a question social scientists have spent a lot of time and funding trying to 171 00:12:59.020 --> 00:13:00.500 figure out. 172 00:13:00.500 --> 00:13:05.900 And the emerging consensus is that for most people most of the time keeping up with current 173 00:13:05.900 --> 00:13:13.140 events via our social feeds produces something very near to zero gains in our knowledge about 174 00:13:13.140 --> 00:13:15.340 civics and politics. 175 00:13:15.340 --> 00:13:19.420 We see a lot of things go by and we feel a lot of feelings about them. 176 00:13:19.420 --> 00:13:24.700 But by and large we are not actually building much knowledge in our human brains. 177 00:13:24.700 --> 00:13:28.460 But I feel kind of cautious about this research because it blends together a lot of different 178 00:13:28.460 --> 00:13:30.060 ways to use social media. 179 00:13:30.060 --> 00:13:36.860 And I think those of us in this room are probably outliers in how we use it. 180 00:13:36.860 --> 00:13:40.620 In most cases like news, the researchers are just talking about a specific thing which 181 00:13:40.620 --> 00:13:48.540 is links to news stories being passed around and not like mega threads of analysis or videos 182 00:13:48.540 --> 00:13:51.220 by particularly gifted explainers. 183 00:13:51.220 --> 00:13:54.180 But even with those caveats the numbers are not good. 184 00:13:54.180 --> 00:14:00.220 There's a very recent and thorough meta analysis on whether we gain political or civic knowledge 185 00:14:00.220 --> 00:14:08.500 from social media exposure that has persuasive results that on the whole we really just don't. 186 00:14:08.500 --> 00:14:11.780 Which doesn't mean that outliers don't exist. 187 00:14:11.780 --> 00:14:15.700 But it does mean unless we're only building for outliers we should recognize that the 188 00:14:15.700 --> 00:14:21.980 combination of social media and human brains does not on average produce workable knowledge 189 00:14:21.980 --> 00:14:24.000 about what's happening. 190 00:14:24.000 --> 00:14:27.520 So that's bad but there's worse. 191 00:14:27.520 --> 00:14:34.200 We also have pretty solid although somewhat less evidenced research that on the whole 192 00:14:34.200 --> 00:14:38.240 we don't just fail to learn things from our social feeds. 193 00:14:38.240 --> 00:14:45.360 We also come away from them feeling more confident in our knowledge that we don't have. 194 00:14:45.360 --> 00:14:51.080 Which is also closely connected to my all time least favorite research on cognition 195 00:14:51.160 --> 00:14:55.200 which is on the illusion of explanatory depth. 196 00:14:55.200 --> 00:14:58.000 Who knows about the illusion of explanatory depth. 197 00:14:58.000 --> 00:14:59.520 Oh great. 198 00:14:59.520 --> 00:15:01.640 Okay we'll get to that later. 199 00:15:01.640 --> 00:15:08.480 For now I want to underline that not knowing things we might need to know is bad but lacking 200 00:15:08.480 --> 00:15:17.080 the epistemic humility to know that we might need to go learn them is probably worse. 201 00:15:17.080 --> 00:15:23.800 And then there's the emotional dimension of the flood. 202 00:15:23.800 --> 00:15:28.600 Which I want to talk about in strictly human terms and without laundering it through social 203 00:15:28.600 --> 00:15:35.160 science which is not to pick on the highbrow moral panic about the internet folks. 204 00:15:35.160 --> 00:15:36.160 Although it is. 205 00:15:36.160 --> 00:15:41.440 But I think many of our dominant critical lenses on our networks and what they do to 206 00:15:41.440 --> 00:15:48.440 society are essentially laundering genuine affective feelings trouble through selective 207 00:15:48.440 --> 00:15:53.160 and often irresponsibly represented social science and neurological research to present 208 00:15:53.160 --> 00:15:57.440 themselves as unshakably authoritative policy prescriptions. 209 00:15:57.440 --> 00:16:04.400 And I took the anxious generation slide out of this talk but please note that it was there. 210 00:16:04.400 --> 00:16:08.520 There are responsible ways to work with the science and I try to be really cautious when 211 00:16:08.520 --> 00:16:12.200 I an epistemic trespasser do it. 212 00:16:12.200 --> 00:16:17.000 But we don't even need it for conversations among ourselves because the affective stuff 213 00:16:17.000 --> 00:16:23.400 is real and we experience countless ephemeral sort of news shaped bids for our attention 214 00:16:23.400 --> 00:16:28.440 every day and most of us who hang out online experience them through highly emotive feeds 215 00:16:28.440 --> 00:16:32.960 that lean toward volume and speed and never ending context collapse. 216 00:16:32.960 --> 00:16:36.240 So for a lot of us it just feels bad. 217 00:16:36.240 --> 00:16:39.320 And this is the saturating flood. 218 00:16:39.320 --> 00:16:41.420 And then we get the shocks. 219 00:16:41.420 --> 00:16:46.880 And in the landslide essay I focus pretty narrowly on the Trump regime's attack on the 220 00:16:46.880 --> 00:16:52.280 idea of truth and on every viable model of building knowledge. 221 00:16:52.280 --> 00:16:59.220 But the more important attack from the administration is that they are just doing deranged nonsensical 222 00:16:59.220 --> 00:17:03.460 bullshit all the time. 223 00:17:03.460 --> 00:17:09.480 Thank you. 224 00:17:09.480 --> 00:17:14.180 And what that does to our ability to understand the world is still not as bad as what they 225 00:17:14.180 --> 00:17:20.960 are actually doing to people and to economies and to countries and to sovereignties. 226 00:17:20.960 --> 00:17:26.740 But it's still very bad because the only way out of this barring international intervention 227 00:17:26.940 --> 00:17:31.780 is for Americans to pull our brains together and act. 228 00:17:31.780 --> 00:17:35.380 And I don't know what your personal networks look like but a lot of people in almost every 229 00:17:35.380 --> 00:17:39.980 part of mine in all my overlapping circles spent a lot of the last year in a state of 230 00:17:39.980 --> 00:17:46.940 real disorientation about things like how is this not breaking the law and what is the 231 00:17:46.940 --> 00:17:49.420 law even? 232 00:17:49.420 --> 00:17:53.340 And this was true for almost everyone except for the folks who knew in advance that they 233 00:17:53.340 --> 00:17:57.900 would be targeted and we're paying a different kind of attention. 234 00:17:57.900 --> 00:18:03.580 But also unfortunately one bad regime is not the only kind of shock that we're experiencing 235 00:18:03.580 --> 00:18:05.900 or will experience. 236 00:18:05.900 --> 00:18:11.200 So I've spent the last few months reading more deeply in the research and also continuing 237 00:18:11.200 --> 00:18:16.860 to plug away in the info minds that I'm breaking and bashing on some possibilities for building 238 00:18:16.860 --> 00:18:21.900 ourselves out of this mess which is what I want to talk about next. 239 00:18:21.900 --> 00:18:27.860 But first I want to briefly say why I think this work belongs here at this protocol conference 240 00:18:27.860 --> 00:18:32.820 more than just about anywhere else. 241 00:18:32.820 --> 00:18:38.020 When I write about the fallout of our global networks some of the points I come to echo 242 00:18:38.020 --> 00:18:43.000 ones that many much fancier critics have covered but I tend to reach different conclusions 243 00:18:43.000 --> 00:18:48.740 than they do because I am focused on builders and they are mostly writing toward individualistic 244 00:18:48.740 --> 00:18:51.460 remedies or toward regulation. 245 00:18:51.460 --> 00:18:56.820 And I have been alive for a lot of the internet so far and I have absolutely no faith in either 246 00:18:56.820 --> 00:18:59.860 of those pathways at this time. 247 00:18:59.860 --> 00:19:02.420 We are not going to disconnect from each other. 248 00:19:02.420 --> 00:19:04.500 It's just not going to happen. 249 00:19:04.500 --> 00:19:11.860 And our regulators are mostly speaking for my own country working from cartoon understandings 250 00:19:11.860 --> 00:19:16.060 of network mechanisms and skating to where the puck was like between two and 12 years 251 00:19:16.060 --> 00:19:22.820 ago which leaves us, the people in this room and people like us who hold the technological 252 00:19:22.820 --> 00:19:26.940 means of production and would like to use it to do something different. 253 00:19:26.940 --> 00:19:30.880 And I will be honest, like saying this kind of thing which I have been saying in a lot 254 00:19:30.880 --> 00:19:35.500 of places for a long time has felt like kicking people who were already down because they 255 00:19:35.500 --> 00:19:41.180 are under resourced and over-taxed open source developers or they are stuck inside corporations 256 00:19:41.180 --> 00:19:46.900 with their own perverse incentives or people who have a lot of really good ideas but haven't 257 00:19:46.900 --> 00:19:52.380 had the technical means to turn those into something they can show other people. 258 00:19:52.380 --> 00:19:56.500 But there are two things happening right now that maybe make this moment different for 259 00:19:56.500 --> 00:19:57.740 some of us. 260 00:19:57.740 --> 00:20:02.780 The first is that decentralized protocols are a live force now with a bigger user base 261 00:20:02.780 --> 00:20:05.260 and more attention than we have ever seen. 262 00:20:05.260 --> 00:20:13.320 The atmosphere is a dev-friendly pro-experimentation zone with so much room for trying things. 263 00:20:13.320 --> 00:20:18.020 The second factor is that writing code has always been difficult and time-consuming and 264 00:20:18.020 --> 00:20:21.780 therefore quite expensive along one axis or another. 265 00:20:21.780 --> 00:20:29.860 And that appears to be changing quite quickly which can go several ways and will. 266 00:20:29.860 --> 00:20:34.180 But the models and the generators seem to be making a lot of previously high learning 267 00:20:34.180 --> 00:20:40.980 curve high time commitment prototyping and experimentation vastly easier to get into. 268 00:20:40.980 --> 00:20:45.220 And when I wrote this I felt like that was still sort of speculative and then we moved 269 00:20:45.220 --> 00:20:48.860 closer to the conference and I was like oh it's really happening I got here and I was 270 00:20:48.860 --> 00:20:52.580 like oh okay. 271 00:20:52.580 --> 00:20:58.540 And this gives people who have been thinking hard about the effects of our networks a 272 00:20:58.540 --> 00:21:04.700 chance to kind of rotate in and start prototyping and building and also I think gives long time 273 00:21:04.700 --> 00:21:11.220 developers more time and freedom to put into stuff that's actually good for us as organisms 274 00:21:11.220 --> 00:21:15.020 and communities of organisms. 275 00:21:15.020 --> 00:21:22.820 Therefore I want to bring you some research threads and hopes and provocations which we 276 00:21:22.860 --> 00:21:30.020 will mostly adapt from the wonderful world of kelp. 277 00:21:30.020 --> 00:21:37.540 I need to turn your attention to macro algae or macro algae for the Brits in the audience. 278 00:21:37.540 --> 00:21:42.980 Specifically to kelp and specifically to the subtlest and least showy parts of kelp which 279 00:21:42.980 --> 00:21:46.980 are its hold fasts. 280 00:21:46.980 --> 00:22:00.100 Hold fasts and thank you to my daughter June for this illustration. 281 00:22:00.100 --> 00:22:04.820 Hold fasts are the structural support systems that kelp and other micro algae and also 282 00:22:04.820 --> 00:22:11.140 a bunch of little sessile which is a beautiful word for non mobile sea creatures evolved 283 00:22:11.140 --> 00:22:16.580 to cling on to the sea floor or even to like bigger shellfish with such strength that they 284 00:22:16.580 --> 00:22:21.260 can hold on and flourish in deep water and in the roughest most turbulent intertidal 285 00:22:21.260 --> 00:22:22.260 zones. 286 00:22:22.260 --> 00:22:27.020 And the reason by the way that I tend to pull so heavily from the natural world and our 287 00:22:27.020 --> 00:22:31.740 interactions with it is that the world has kind of figured a lot of shit out in a lot 288 00:22:31.740 --> 00:22:37.180 of ways and I think leaving all of those models of complexity on the table is goofy and also 289 00:22:37.180 --> 00:22:40.260 kelp is awesome. 290 00:22:41.020 --> 00:22:46.780 So kelp and its hold fasts do three things that I think could be very useful for us to 291 00:22:46.780 --> 00:22:47.780 think about. 292 00:22:47.780 --> 00:22:54.740 They anchor entire communities into stability and they create permeable respite and sturdy 293 00:22:54.740 --> 00:22:59.980 shelter for all kinds of weirdos both first up in their fronds and down at the interface 294 00:22:59.980 --> 00:23:03.460 between the bedrock and the sea. 295 00:23:03.460 --> 00:23:08.220 So the first thing I want to bring is the idea of anchoring networks and humans. 296 00:23:08.260 --> 00:23:16.260 I think we could work toward reciprocal integration of rich sources of human knowledge. 297 00:23:16.260 --> 00:23:20.820 We know that news organizations have tried to adapt to the distribution changes that 298 00:23:20.820 --> 00:23:26.300 our networks have forced on them but the network's needs and affordances keep shifting so like 299 00:23:26.300 --> 00:23:31.020 getting good at SEO worked for a while with constant recalibration and then AI summaries 300 00:23:31.020 --> 00:23:37.200 have just cratered that experience for news orgs which is going to shut down newsrooms 301 00:23:37.200 --> 00:23:39.480 and already is. 302 00:23:39.480 --> 00:23:43.700 Getting integrated into social platforms worked pretty well until meta's properties started 303 00:23:43.700 --> 00:23:48.240 de-prioritizing news and Twitter turned into X and started hiding links. 304 00:23:48.240 --> 00:23:53.120 And I know that folks here today are psyched about getting journalists and news orgs onto 305 00:23:53.120 --> 00:23:57.400 the protocol but we're going to have to make it worth their while and demonstrate that 306 00:23:57.400 --> 00:24:02.520 there's a long game here that's going to be beneficial to them rather than a short period 307 00:24:02.520 --> 00:24:08.720 of reciprocity followed by a shift to either extraction or exclusion. 308 00:24:08.720 --> 00:24:11.940 And part of that is going to be based on whether or not people in the protocol are actually 309 00:24:11.940 --> 00:24:14.620 engaging with news and knowledge. 310 00:24:14.620 --> 00:24:19.840 As I suspect, although I have no data for this, is higher on blue sky and various other 311 00:24:19.840 --> 00:24:25.160 skies than on other platforms. 312 00:24:25.160 --> 00:24:32.080 And also we need to tell whether people in the protocol or encourage people in the protocol 313 00:24:32.200 --> 00:24:37.040 are engaging with that rather than treating links and headlines like trading cards to 314 00:24:37.040 --> 00:24:42.080 signal membership or confirm the truth of our already stated positions. 315 00:24:42.080 --> 00:24:48.520 So this is a long way to say we should try to get people to actually read what they're 316 00:24:48.520 --> 00:24:52.280 sharing. 317 00:24:52.280 --> 00:24:54.960 And the current state there is not great. 318 00:24:54.960 --> 00:24:58.200 Nature analysis of 35 billion Facebook posts. 319 00:24:58.200 --> 00:25:04.680 I know it's Facebook but that had links showed that three quarters of the time people shared 320 00:25:04.680 --> 00:25:05.680 links. 321 00:25:05.680 --> 00:25:10.640 They hadn't clicked through let alone read the article they were sharing. 322 00:25:10.640 --> 00:25:16.000 And the presence of sharing cues, just the existence of a cue, might you want to share 323 00:25:16.000 --> 00:25:17.040 this? 324 00:25:17.040 --> 00:25:23.200 Also makes people measurably worse at discerning true things from false things. 325 00:25:23.200 --> 00:25:29.560 Or as the researchers put it, the social media context interferes with truth discernment. 326 00:25:29.560 --> 00:25:34.960 So social media encounters as we see them on the big centralized platforms seem to push 327 00:25:34.960 --> 00:25:37.440 interactions with news. 328 00:25:37.440 --> 00:25:43.000 Not just from knowing things to knowing kind of about things, but from knowing about things 329 00:25:43.000 --> 00:25:46.480 to signaling that we might know about them. 330 00:25:46.480 --> 00:25:49.480 Which is getting really far away from the knowledge itself. 331 00:25:49.480 --> 00:25:52.800 Which I mean I've been on the internet, that tracks. 332 00:25:52.800 --> 00:25:58.080 If we wanted to move toward better integration and more reciprocity with knowledge collecting 333 00:25:58.080 --> 00:26:05.160 institutions because if no one's clicking it's not helping them, what would we do? 334 00:26:05.160 --> 00:26:12.760 So for blue sky and black sky and north sky and probably various other skies, y'all are 335 00:26:12.760 --> 00:26:17.840 already ahead by not doing things that are actively hostile to knowledge. 336 00:26:17.840 --> 00:26:22.600 Like hiding links and deprioritizing news and analysis. 337 00:26:22.600 --> 00:26:26.880 There's probably also room to think about some remediation. 338 00:26:26.880 --> 00:26:32.480 I will start here with my least favorite intervention, which is the nudge. 339 00:26:32.480 --> 00:26:36.080 The nudge to read before sharing I find this highly annoying. 340 00:26:36.080 --> 00:26:41.280 But I think it's interesting that it does seem to change behavior, at least in the short 341 00:26:41.280 --> 00:26:42.280 term. 342 00:26:42.280 --> 00:26:46.440 Twitter's comms team claimed in 2020 that when they did a little, want to read this 343 00:26:46.440 --> 00:26:51.500 before retweeting it, nudge, they got a 40% boost in article opening. 344 00:26:51.500 --> 00:26:53.820 We don't know what went into people's brains, but we know that happened. 345 00:26:53.820 --> 00:26:55.620 That's not nothing. 346 00:26:55.620 --> 00:27:00.780 And as skeptical as I am about nudges, all friction is perhaps not terrible. 347 00:27:00.780 --> 00:27:06.540 There's a study from just this past winter that looked quite carefully at friction, the 348 00:27:06.540 --> 00:27:13.220 nudge, and found that a very, very light touch of friction, so just a very occasional intervention, 349 00:27:13.220 --> 00:27:17.620 plus a tiny quiz, really improved information quality. 350 00:27:17.620 --> 00:27:24.260 Like a third, just doing the friction made people stop sharing things as much. 351 00:27:24.260 --> 00:27:27.220 It didn't improve the quality of the knowledge on the network. 352 00:27:27.220 --> 00:27:32.420 But the share friction plus a little bit of engaging analytical thought actually increased 353 00:27:32.420 --> 00:27:37.580 the quality according to the study's measures by about a third, which is so much. 354 00:27:37.580 --> 00:27:43.660 But I should say the caveat is that this study was done using a simulation. 355 00:27:43.660 --> 00:27:50.940 So it's kind of like reading the headline that's like, scientists cure cancer in mice. 356 00:27:50.940 --> 00:27:52.780 But even as with the mouse effect, right? 357 00:27:52.780 --> 00:27:55.620 They're interesting mice. 358 00:27:55.620 --> 00:28:01.260 So adding friction is never going to fully reverse the directional movement of a system 359 00:28:01.260 --> 00:28:08.800 that leans toward virality and toward first-mover advantage and toward signaling versus understanding. 360 00:28:08.800 --> 00:28:13.620 So what would it look like to move that problem upstream? 361 00:28:13.620 --> 00:28:17.860 What would a system look like that actually centrally attended to community knowledge 362 00:28:17.860 --> 00:28:19.540 and reciprocity? 363 00:28:19.540 --> 00:28:24.260 What would it do when a user in such a system encountered something that appears to be, 364 00:28:24.260 --> 00:28:28.840 for instance, news or analysis? 365 00:28:28.840 --> 00:28:33.460 Maybe it would open an article or the video or whatever and suggest finding a good clip 366 00:28:33.460 --> 00:28:37.580 or a quote to share, which could improve the quality of shared information and perhaps 367 00:28:37.580 --> 00:28:40.060 get more of it into the sharer's brains. 368 00:28:40.060 --> 00:28:44.420 Maybe it would automatically add sources mentioned in post that people bookmark or 369 00:28:44.420 --> 00:28:50.340 a fave to a reading or watch list that they can return to to help fight the anxiety-producing 370 00:28:50.340 --> 00:28:54.460 ephemerality that keeps information seekers doom-scrolling. 371 00:28:54.460 --> 00:29:00.220 Maybe it could offer a tiny, even machine-generated summary and some context cues about the source 372 00:29:00.220 --> 00:29:05.300 that provide enough trust signal from people with broadly strong reputations for a given 373 00:29:05.300 --> 00:29:10.540 local cluster to pull attention away from worthless content farms and misinfo peddlers 374 00:29:10.540 --> 00:29:14.500 and toward journalists and creators with integrity. 375 00:29:14.500 --> 00:29:19.420 Or maybe, God help us, it would ask us to answer a question about the material in the 376 00:29:19.420 --> 00:29:22.180 article, which I actually, I've flipped. 377 00:29:22.180 --> 00:29:23.180 I'd love that. 378 00:29:23.180 --> 00:29:29.180 I'm fully ready to leaderboard knowledge as a countermeasure to the current mode of leaderboarding, 379 00:29:29.180 --> 00:29:30.900 the lowest common denominator. 380 00:29:30.900 --> 00:29:31.900 Let's go. 381 00:29:32.580 --> 00:29:36.580 The right move obviously depends on a specific project's values and goals. 382 00:29:36.580 --> 00:29:41.900 I think this is the right level to consider it back in the architecture of a social internet 383 00:29:41.900 --> 00:29:46.260 system rather than as aftermarket fixes. 384 00:29:46.260 --> 00:29:53.020 Having seen yesterday's science track, those folks are way ahead of everyone else on this. 385 00:29:53.020 --> 00:29:59.420 It's so good. 386 00:29:59.420 --> 00:30:02.540 Next cool kelp fact. 387 00:30:02.540 --> 00:30:08.380 Kelp makes slower, calmer zones. 388 00:30:08.380 --> 00:30:15.580 For us, I think we could make slower, calmer zones that help our people keep them from 389 00:30:15.580 --> 00:30:19.180 being overwhelmed by network tides. 390 00:30:19.180 --> 00:30:21.940 There are two ways to look at this. 391 00:30:21.940 --> 00:30:25.940 First I want to say, kelp forests are super bio-diverse. 392 00:30:25.940 --> 00:30:27.740 They're a very rich community. 393 00:30:27.740 --> 00:30:35.340 They're up there with coral reefs on the biodiversity and what we call productivity 394 00:30:35.340 --> 00:30:36.780 as a system. 395 00:30:36.780 --> 00:30:43.900 They do two things roughly that help them be such an important sheltering force. 396 00:30:43.900 --> 00:30:50.260 The first is up in the water column where the kelp fronds and blades calm and slow and 397 00:30:50.260 --> 00:30:55.900 deflects the very strong currents that surround kelp forests, creating calmer and less turbulent 398 00:30:55.900 --> 00:30:59.140 spaces where other organisms can thrive. 399 00:30:59.140 --> 00:31:02.100 The second is down in the hold fasts themselves. 400 00:31:02.100 --> 00:31:07.180 The structures that wrap around and latch on to the sturdiest anchor they can are actually 401 00:31:07.180 --> 00:31:10.380 the most thriving part of the community. 402 00:31:10.380 --> 00:31:15.900 They shelter hundreds of mobile and sessile, invertebrate species, so they're like the 403 00:31:15.900 --> 00:31:18.660 critical infrastructure for kelp busy town. 404 00:31:18.660 --> 00:31:21.700 I want to take these one at a time. 405 00:31:21.700 --> 00:31:26.820 First up in the water column where the tides are moving fast, but perhaps our network's 406 00:31:26.820 --> 00:31:34.940 affordances could provide a little bit of cognitive and emotional traffic calming. 407 00:31:34.940 --> 00:31:38.660 Something I've been watching for a long time and especially this year is that every time 408 00:31:38.660 --> 00:31:43.900 something happens, whether it's someone dropping an album or a natural disaster happening or 409 00:31:43.900 --> 00:31:50.660 ice shooting someone in my country, there is an instant hyper-proliferation of chatter 410 00:31:50.660 --> 00:31:55.620 about it on the networks that comes online around the same time as live posting and eyewitness 411 00:31:55.620 --> 00:31:58.260 video or spot analysis. 412 00:31:58.260 --> 00:32:02.520 Then you get the local and specialist coverage and then you get the national and international 413 00:32:02.520 --> 00:32:07.220 coverage and the next morning the think tanks wake up and get moving on it. 414 00:32:07.220 --> 00:32:12.340 This is always expanding, but we don't really have a corresponding moment of synthesis or 415 00:32:12.340 --> 00:32:20.620 contraction of all of that stuff into something that is intelligible to a reasonable person. 416 00:32:20.700 --> 00:32:26.380 This is wild to me, but we can't just go to our networks or really anywhere, speaking 417 00:32:26.380 --> 00:32:31.580 in general knowledge terms, I understand the scientists may have this better, and find 418 00:32:31.580 --> 00:32:38.300 just an event node that connects all the sources of relevant information and the opinions of 419 00:32:38.300 --> 00:32:41.100 people we care about in one place. 420 00:32:41.100 --> 00:32:45.740 We have reporters who can do this high-speed gallop through the relevant context at the 421 00:32:45.740 --> 00:32:49.300 beginning of every article they write and then the next day they wake up and they write 422 00:32:49.300 --> 00:32:50.580 another article. 423 00:32:50.580 --> 00:32:52.220 It's all super atomized. 424 00:32:52.220 --> 00:32:54.820 There's nothing to bring it together. 425 00:32:54.820 --> 00:33:00.340 If you want to build a knowledge molecule, you have to do it with your own brain or you 426 00:33:00.340 --> 00:33:07.500 can glom on to someone in the creator or interpreter class, some of whom are genuinely world-class, 427 00:33:07.500 --> 00:33:13.700 super gifted analysts and some of whom are basically semi-deranged street preachers and 428 00:33:13.700 --> 00:33:18.020 not everyone can tell the difference all the time. 429 00:33:18.140 --> 00:33:23.460 Uplifting and reciprocating our gifted interpreters is necessary and I hope we're going to hear 430 00:33:23.460 --> 00:33:26.140 more about that this weekend. 431 00:33:26.140 --> 00:33:31.380 But it also seems to me that we really could get the machines involved in bundling up all 432 00:33:31.380 --> 00:33:36.180 of this noise into intelligible collections of material that we could actually learn through 433 00:33:36.180 --> 00:33:41.100 and refer to so that it's not so insistently ephemeral and scattered. 434 00:33:41.100 --> 00:33:47.180 And if this sounds like resuscitating the dream of structured data, it kind of is. 435 00:33:47.180 --> 00:33:56.660 But I want to put it in service of ordinary people using tech we didn't have before and 436 00:33:56.660 --> 00:34:03.020 signals from social that can make this stuff more than like a terrible index. 437 00:34:03.020 --> 00:34:07.540 And it's odd to me that we have all of these recommendation algorithms working in variously 438 00:34:07.540 --> 00:34:13.540 pro and antisocial ways, but as far as I can tell, we haven't effectively brought our 439 00:34:13.540 --> 00:34:20.500 social tech to the work of orientation or reorientation in any structured way. 440 00:34:20.500 --> 00:34:26.740 As again the science cluster is ably demonstrating, we can develop features that cluster and amplify 441 00:34:26.740 --> 00:34:28.700 the knowledge people assemble. 442 00:34:28.700 --> 00:34:33.340 We can do it with machine learning and human curation, probably both of them. 443 00:34:33.340 --> 00:34:38.500 We could give our communities the tools to do this for their areas of interest or their 444 00:34:38.500 --> 00:34:41.580 physical localities or their specializations. 445 00:34:41.580 --> 00:34:48.940 And we could extend existing work on community moderation to rule certain knowledge associations 446 00:34:48.940 --> 00:34:50.500 in or out. 447 00:34:50.500 --> 00:34:55.400 But if we only do this work for highly motivated specialists, we're missing the chance to bring 448 00:34:55.400 --> 00:34:58.460 that fire to everyone else. 449 00:34:58.460 --> 00:35:02.820 And I will now embarrass my old journalism world friend Tyler Fisher and say that I think 450 00:35:02.820 --> 00:35:07.100 his work with CIL is admirable and a really good start towards this. 451 00:35:07.100 --> 00:35:12.140 And some things about the next area I want to talk about, which is mitigating the cognitive 452 00:35:12.140 --> 00:35:21.140 and emotional effects of processing so many different emotional registers in one place. 453 00:35:21.140 --> 00:35:28.540 I think we could give our embodied human selves a break if we were to find a way to split 454 00:35:28.540 --> 00:35:34.260 rhetorical registers at the point of reader viewer contact. 455 00:35:34.260 --> 00:35:38.500 We go online because we're bored or we're lonely or we want to see what people are talking 456 00:35:38.500 --> 00:35:43.060 about either in a professional or a very unprofessional way. 457 00:35:43.060 --> 00:35:49.900 But what most of us get when we come online is this amazing puree of jokes and infighting 458 00:35:49.900 --> 00:35:55.920 and scary news and hustling and entertaining nonsense and dumb memes and video evidence 459 00:35:55.920 --> 00:35:59.240 of war crimes. 460 00:35:59.240 --> 00:36:02.800 I think this is genuinely not good for us. 461 00:36:02.800 --> 00:36:07.680 There's a lot of evidence about this, but honestly, it's kind of weird. 462 00:36:07.680 --> 00:36:10.040 I wanted it to be stronger. 463 00:36:10.040 --> 00:36:14.320 There is a study about how rough context switching is for our ability to understand things, but 464 00:36:14.320 --> 00:36:15.940 it's mostly focused on format. 465 00:36:15.940 --> 00:36:22.080 So it's hard to go from video to text, which is fine, but not the same thing. 466 00:36:22.080 --> 00:36:28.480 There's this really fun paper with three studies from a couple of professors of advertising 467 00:36:28.480 --> 00:36:34.600 that found that scrolling Instagram for 30 seconds essentially makes us weak to ads, 468 00:36:34.600 --> 00:36:39.240 but especially to ads that have high social signals. 469 00:36:39.240 --> 00:36:43.880 So we're switching from a more analytical mode to heuristic processing, and if a lot 470 00:36:43.880 --> 00:36:49.100 of people appear to have liked the thing, we're more likely to just buy whatever nonsense 471 00:36:49.100 --> 00:36:51.000 it is. 472 00:36:51.020 --> 00:36:56.160 The same paper also includes another experiment showing that people who scroll their social 473 00:36:56.160 --> 00:37:01.540 feed for a very brief period also immediately got worse at math and sentence diagramming. 474 00:37:01.540 --> 00:37:02.540 So that's fun. 475 00:37:02.540 --> 00:37:05.160 But that's just one paper. 476 00:37:05.160 --> 00:37:09.160 And there are a few others, but they're not necessarily generalizable, so I want to be 477 00:37:09.160 --> 00:37:10.160 careful. 478 00:37:10.160 --> 00:37:13.940 I think we do have two really helpful bodies here of indirect evidence. 479 00:37:13.940 --> 00:37:20.480 One is the quite beefy body of work on transportive communication, stuff that transports us, which 480 00:37:20.480 --> 00:37:26.280 is behind a bunch of the power of storytelling stuff. 481 00:37:26.280 --> 00:37:32.200 But essentially things that feel transportive, including a lot of first-person testimony, 482 00:37:32.200 --> 00:37:35.400 really reduce skepticism and analytical thought. 483 00:37:35.400 --> 00:37:39.760 And there's a solid pool of evidence and theory there, kind of just waiting for someone to 484 00:37:39.760 --> 00:37:43.800 step in and work on the effects of trying to move between the transportive and just 485 00:37:43.800 --> 00:37:49.720 quotidian newsy stuff registers in rapid succession over and over again in the same 486 00:37:49.720 --> 00:37:51.920 place. 487 00:37:51.920 --> 00:37:57.720 But the second thing we have is, again, the evidence of our senses. 488 00:37:57.720 --> 00:38:04.000 I think there are a few people having a good time in the political news zone on the social 489 00:38:04.000 --> 00:38:05.600 internet. 490 00:38:05.600 --> 00:38:10.840 I think there are journalism and policy people who mostly treat the networks like Brief Mean 491 00:38:10.840 --> 00:38:14.880 LinkedIn, which works out well for them. 492 00:38:14.880 --> 00:38:20.320 And then there are a lot of people, including people in positions of extraordinary power, 493 00:38:20.320 --> 00:38:24.840 who treat the experience of modern life and the suffering of others as a spectator blood 494 00:38:24.840 --> 00:38:26.040 sport. 495 00:38:26.040 --> 00:38:31.140 And I do not think that that mode in particular is good or even neutral. 496 00:38:31.140 --> 00:38:36.080 I think we will not fix the problems with our governance if we continue to allow it 497 00:38:36.080 --> 00:38:39.920 to metastasize through our civic life. 498 00:38:40.000 --> 00:38:44.540 And for the rest of us, this stuff is just exhausting and alienating, and it does things 499 00:38:44.540 --> 00:38:46.640 to our brains that we don't like. 500 00:38:46.640 --> 00:38:50.280 And I want to briefly bring in Joshua Myrowitz. 501 00:38:50.280 --> 00:38:52.840 Do we have other Myrowitz heads in the crowd? 502 00:38:52.840 --> 00:38:54.360 He wrote No Sense a Place in 1986. 503 00:38:54.360 --> 00:38:58.360 Oh, my God, I'm going to write so many blog posts about Joshua Myrowitz for you. 504 00:38:58.360 --> 00:39:04.680 So he's this media scholar whose ideas informed Dana Boyd and Alice Marwick's work on context 505 00:39:04.680 --> 00:39:05.680 collapse. 506 00:39:06.440 --> 00:39:14.480 But he was writing in the mid 80s, which makes this a little unsettling to me. 507 00:39:14.480 --> 00:39:22.120 He wrote, the current merging of situations, by which he means places, doesn't give us 508 00:39:22.120 --> 00:39:27.960 a sum of what we had before, but rather new synthesized behaviors that are qualitatively 509 00:39:27.960 --> 00:39:29.000 different. 510 00:39:29.000 --> 00:39:33.680 If we celebrate our child's wedding in an isolated situation where it is the sole experience 511 00:39:33.760 --> 00:39:37.160 of the day, then our joy may be unbounded. 512 00:39:37.160 --> 00:39:41.960 But when on our way to the wedding, we hear over the car radio of a devastating earthquake 513 00:39:41.960 --> 00:39:46.920 or the death of a popular entertainer or the assassination of a political figure, we not 514 00:39:46.920 --> 00:39:54.240 only lose our ability to rejoice fully, but also our ability to mourn deeply. 515 00:39:54.240 --> 00:39:59.280 Unbounded joy and unmitigated misery cannot coexist in the same place and time. 516 00:39:59.280 --> 00:40:06.320 The situations merge, the hot flush and the icy stare blend into a middle region cool. 517 00:40:06.320 --> 00:40:09.680 And for Myrowitz's cool, I would substitute numb. 518 00:40:11.680 --> 00:40:17.040 Now some communities in the decentralized space have tried to deal with this by asking 519 00:40:17.040 --> 00:40:22.080 people to either content warn their posts based on the whole panoply of their readers' 520 00:40:22.080 --> 00:40:26.400 potential sensitivities or to divide their expression into separate accounts. 521 00:40:26.880 --> 00:40:30.480 And I think these people are working from good instincts, but from the wrong end of 522 00:40:30.480 --> 00:40:35.800 the problem, given how alienating and anti-convivial it is to be instructed to preprocess your 523 00:40:35.800 --> 00:40:39.320 emotional utterances for people you may not even be speaking to. 524 00:40:39.320 --> 00:40:43.240 But I think we could make this work on the other end. 525 00:40:43.240 --> 00:40:47.680 And I want to say if we do it successfully, it will be because we recognize that we cannot 526 00:40:47.680 --> 00:40:54.880 rely on like civility forward sentiment analysis that for instance reads black Americans speaking 527 00:40:54.880 --> 00:40:58.640 in a neutral register as aggressive. 528 00:40:58.640 --> 00:41:03.400 Doing it well would involve grounding in community and doing a lot of training work which would 529 00:41:03.400 --> 00:41:06.320 have to be shared across a willing group. 530 00:41:06.320 --> 00:41:08.680 I don't want to undersell the difficulty. 531 00:41:08.680 --> 00:41:16.000 But being able to come online and see a mourning experience that starts with like, hi, friends, 532 00:41:16.000 --> 00:41:20.360 it's puppy pictures and selfies and not have the hard news come online until like after 533 00:41:20.360 --> 00:41:22.720 the caffeine kicks in. 534 00:41:22.800 --> 00:41:24.880 I think this would be transformative. 535 00:41:24.880 --> 00:41:29.400 And like for me when I'm working, I would love to be able to filter in posts that match 536 00:41:29.400 --> 00:41:36.880 the register I'm working in and hold the puppies and beefs for later. 537 00:41:36.880 --> 00:41:42.680 I would like to think about this less as feed switching and more like a radio dial. 538 00:41:42.680 --> 00:41:47.640 Always the people that we like, but with the option to filter different registers in and 539 00:41:47.640 --> 00:41:52.440 out to suit different emotional capacities and allow for some separation. 540 00:41:52.960 --> 00:41:58.040 So those are some thoughts about working on this stuff in kind of lateral ways up in the 541 00:41:58.040 --> 00:42:00.240 big open world of the water column. 542 00:42:00.240 --> 00:42:03.240 And now I want to talk about the other thing. 543 00:42:03.240 --> 00:42:08.400 The most vibrant part of the kelp forest is in the hold fasts which act like roots but 544 00:42:08.400 --> 00:42:12.160 kind of in reverse instead of pulling resources out of the soil. 545 00:42:12.160 --> 00:42:17.320 They push benefits out to every living thing around them by providing a firmly anchored 546 00:42:17.320 --> 00:42:24.160 structure that offers complex and variegated shelter for both the charismatic invertebrates 547 00:42:24.160 --> 00:42:29.480 with all the little flowy parts and also the tiny snails and worms and all the other beloved 548 00:42:29.480 --> 00:42:32.240 lobes of the kelp forest. 549 00:42:32.240 --> 00:42:37.840 So one more Joshua Myrowitz quote from 1986. 550 00:42:37.840 --> 00:42:42.640 For every moment of our lives there are things happening someplace that would upset us, that 551 00:42:42.640 --> 00:42:46.760 would involve us, that would drain our energies and engage our feelings. 552 00:42:46.760 --> 00:42:50.440 There are more things that if present in our view we would feel compelled to respond 553 00:42:50.440 --> 00:42:56.560 to or do something about than could possibly fit into the same experience of a single lifetime. 554 00:42:56.560 --> 00:43:01.720 Situation segregation, however, acts as a psychosocial shock absorber. 555 00:43:01.720 --> 00:43:06.480 By selectively exposing ourselves to events and other people we control the flow of our 556 00:43:06.480 --> 00:43:08.480 actions and emotions. 557 00:43:08.480 --> 00:43:14.260 Compassion, empathy, and even ethics may be more situationally bound than we often care 558 00:43:14.260 --> 00:43:15.260 to think. 559 00:43:15.740 --> 00:43:18.140 I love this guy. 560 00:43:18.140 --> 00:43:22.700 I feel this in the way that I participate in the online world. 561 00:43:22.700 --> 00:43:27.860 I think we need space to think and feel together that isn't always fully in the current and 562 00:43:27.860 --> 00:43:33.380 I feel like probably everybody here gets this. 563 00:43:33.380 --> 00:43:37.500 It also produces the benefit, also I think we understand the benefit of not having to 564 00:43:37.500 --> 00:43:42.980 send people off protocol into discords and group chats whenever they need to do that 565 00:43:42.980 --> 00:43:44.380 work. 566 00:43:44.380 --> 00:43:49.340 The spaces work that the Blue Sky team has been publishing about is going to unlock so 567 00:43:49.340 --> 00:43:53.260 many affordances for pro social things that go well beyond private accounts. 568 00:43:53.260 --> 00:43:57.220 I cannot tell you how excited I am about this. 569 00:43:57.220 --> 00:44:02.300 I also want to make a note, I think letting people huddle up inside an anchoring structure 570 00:44:02.300 --> 00:44:09.300 also undercuts both the need for and the value of mobbing as territorial defense. 571 00:44:09.300 --> 00:44:13.860 I think that's something very real on the networks right now and that also came up in 572 00:44:13.860 --> 00:44:15.860 the science track yesterday. 573 00:44:15.860 --> 00:44:21.460 If you and your community have a cave or a building or a little sheltering trees to step 574 00:44:21.460 --> 00:44:26.820 into to get out of the broad featureless plane with everyone in it filled with many other 575 00:44:26.820 --> 00:44:32.060 communities some of whom will always hate you, you're not going to have to constantly 576 00:44:32.060 --> 00:44:38.500 make up for your lack of shelter by continuously contesting every millimeter of cultural boundary. 577 00:44:38.500 --> 00:44:40.300 You can just sidestep it. 578 00:44:40.300 --> 00:44:44.900 This is so low level that I think it's honestly hard to find ways to talk about, but I think 579 00:44:44.900 --> 00:44:48.060 we have to solve these problems. 580 00:44:48.060 --> 00:44:55.680 I would like to suggest as a completely unqualified nonanthropologist that the entire known history 581 00:44:55.680 --> 00:45:01.060 of humans coming together in groups larger than a few families has involved making buildings 582 00:45:01.060 --> 00:45:05.960 of one kind or another and then elaborating those buildings fairly quickly with rooms 583 00:45:05.960 --> 00:45:10.560 and divisions because our structures out in the physical world don't just shield us 584 00:45:10.560 --> 00:45:11.560 from the environment. 585 00:45:11.560 --> 00:45:16.680 They also give us necessary shelter from each other which probably brings the murder rate 586 00:45:16.680 --> 00:45:21.120 down among other things. 587 00:45:21.120 --> 00:45:26.700 I said earlier that I would revisit the illusion of explanatory depth which is essentially 588 00:45:26.700 --> 00:45:32.400 a pervasive cognitive con that all our brains are running all the time by leading us to 589 00:45:32.400 --> 00:45:37.120 believe that we understand things about the world that we actually do not. 590 00:45:37.120 --> 00:45:42.960 We don't know that we don't know until we have to explain something or fix something 591 00:45:42.960 --> 00:45:46.200 and then it's not there. 592 00:45:46.200 --> 00:45:50.880 If I had an extra 10 minutes I would absolutely ask you each to turn to the person next to 593 00:45:50.880 --> 00:45:56.120 you and explain how a flush toilet works. 594 00:45:56.120 --> 00:45:57.640 I've worked on flush toilets. 595 00:45:57.640 --> 00:45:58.640 I was so wrong. 596 00:45:58.640 --> 00:46:02.300 It was very upsetting. 597 00:46:02.300 --> 00:46:08.640 This is usually okay as long as we have reliable sources to refer back to but it's extremely 598 00:46:08.640 --> 00:46:09.640 bad when we don't. 599 00:46:09.640 --> 00:46:14.620 I read a lot about this in the landslide essay and I'll link the relevant research 600 00:46:14.620 --> 00:46:19.960 but the upshot is that for people who care about knowledge we need to know that we don't 601 00:46:19.960 --> 00:46:25.940 know very much in our brains actually and that our knowledge is not just embodied but 602 00:46:25.940 --> 00:46:31.180 also genuinely social and networked which means that the quality of knowledge in our 603 00:46:31.180 --> 00:46:37.840 networks when it goes down we all get less knowledgeable and we all get less capable 604 00:46:37.840 --> 00:46:41.500 and I would argue that we all get less free. 605 00:46:41.500 --> 00:46:46.820 This is why everything I'm suggesting today centres on treating ordinary people as subjects 606 00:46:46.820 --> 00:46:52.380 with agency who deserve workable access to good information in reasonable environments 607 00:46:52.620 --> 00:46:58.540 rather than treating them as consumers whose purpose is to be a manipulable object. 608 00:46:58.540 --> 00:47:02.560 The science folks from yesterday are already demonstrating the protocol can be used to 609 00:47:02.560 --> 00:47:07.740 do knowledge consolidating and brain preserving work but if we don't bring those goods to 610 00:47:07.740 --> 00:47:13.660 everyone at all levels we're going to lose our democracies all of which are predicated 611 00:47:13.660 --> 00:47:21.680 on the existence of at least minimally informed civic community who are subjects with agency. 612 00:47:21.680 --> 00:47:24.720 So I've got one more cool kelp fact for you. 613 00:47:24.720 --> 00:47:29.080 At least you'll have the kelp facts even if nothing else carries forward. 614 00:47:29.080 --> 00:47:33.680 Kelp are way older than we thought until a few years ago we thought they were only 14 615 00:47:33.680 --> 00:47:38.080 million years old because we found a kelp fossil and it was 14 million years old. 616 00:47:38.080 --> 00:47:43.560 But then this amateur fossil collector down in Washington state cracked open some rocks 617 00:47:43.560 --> 00:47:48.080 he found on the beach and found something that looked a lot like a fossilized hold fast. 618 00:47:48.080 --> 00:47:54.640 So he sent it down to the ancient kelp dating nerds at UC Berkeley and the new fossils were 619 00:47:54.640 --> 00:47:58.480 around 32 million years old. 620 00:47:58.480 --> 00:48:04.700 And this is important because it means that the kelp evolved way before all the species 621 00:48:04.700 --> 00:48:09.560 that now shelter in their fronds and their hold fasts and became a foundation species 622 00:48:09.560 --> 00:48:13.800 without which all that other life would have been impossible. 623 00:48:13.800 --> 00:48:18.960 But another way their presence and the affordances they offered bent the future around them in 624 00:48:18.960 --> 00:48:23.320 ways that we now find beautiful and valuable and good. 625 00:48:23.320 --> 00:48:28.240 To me the point of working toward knowledge anchoring and against continuous turbulence 626 00:48:28.240 --> 00:48:34.360 and trouble is not just short term as a respite or a little pocket oasis for us and our friends 627 00:48:34.360 --> 00:48:39.440 but toward a genuinely more workable way to think and know together. 628 00:48:39.440 --> 00:48:46.240 Where together is going to include billions of humans and also many forms of machine assistance. 629 00:48:46.240 --> 00:48:52.280 And this moment is our window for bending the future around our hopes and values that 630 00:48:52.280 --> 00:48:57.560 cherish knowledge and communal knowing and that orient us toward real remedies for some 631 00:48:57.560 --> 00:49:05.080 of our most ancient information troubles and pains in service of living together in peace 632 00:49:05.080 --> 00:49:06.080 and mutual respect. 633 00:49:06.080 --> 00:49:10.080 Thank you. 634 00:49:37.080 --> 00:49:47.320 Truly, truly, leaning toward the wrong mic, if you're working on this stuff, if you care 635 00:49:47.320 --> 00:49:49.160 about working on this stuff, let's talk. 636 00:49:49.160 --> 00:49:50.160 That's why I'm here. 637 00:49:50.160 --> 00:49:53.800 I gave this talk so they'd let me in the door. 638 00:49:53.800 --> 00:49:56.720 But I want to talk to people who are actually trying to fix this. 639 00:49:56.720 --> 00:49:58.720 Thank you. 640 00:49:58.720 --> 00:49:59.720 We have no time for Q&A. 641 00:49:59.720 --> 00:50:00.720 I didn't feel that. 642 00:50:00.720 --> 00:50:01.720 I want a couple of minutes. 643 00:50:01.720 --> 00:50:02.720 This is how these things go. 644 00:50:02.720 --> 00:50:08.720 Thank you.