[00:00.000 --> 00:00.460]  Cool.
[00:01.920 --> 00:05.700]  I'm going to turn on my own little recording here so that I have my own record as well.
[00:09.560 --> 00:10.200]  All right.
[00:10.480 --> 00:11.880]  My name is Lydia Lawrenson.
[00:12.140 --> 00:14.640]  I am super excited to be here.
[00:15.480 --> 00:21.480]  I've never been to HOPE before, and people say it's like the cool version of DEF CON.
[00:22.400 --> 00:24.600]  So I feel very honored to be speaking here.
[00:26.840 --> 00:30.500]  And the talk that I'm going to give you today is Media Vibe Coding in the Long Tail.
[00:32.400 --> 00:40.740]  So just as a quick preview, my interest here is will something truly new become possible in media due to Vibe Coding?
[00:43.460 --> 00:48.960]  I'll talk a bit about kind of the recent history of media, a little bit since like the 60s, but it's kind of overall.
[00:49.380 --> 00:52.200]  And a little bit about how the tech has changed.
[00:52.520 --> 01:05.480]  And then in particular, I'm really interested in how Vibe Coding specifically, but you know, any technological change, recently is affecting how communities and institutions like journalism function.
[01:09.170 --> 01:11.910]  So the reason I was invited to HOPE...
[01:14.650 --> 01:22.110]  I haven't been working in the tech industry directly for a little while, but I recently attended a Skillshare for Vibe Coding.
[01:22.110 --> 01:26.110]  One of my buddies, I live in the Bay Area, so it's like very easy to attend this sort of thing.
[01:27.810 --> 01:28.870]  And go to the Skillshare.
[01:30.070 --> 01:31.190]  Friend's like, what do you want to build?
[01:31.410 --> 01:32.690]  I'm like, gosh, I don't even know.
[01:33.090 --> 01:34.270]  And I was like thinking about it.
[01:34.270 --> 01:46.570]  And I realized, you know, I've thought, because I've done a bunch of work in journalism and other forms of media, I'm really interested in the question of fact checking and whether we can like make it more efficient or cheaper or whatever.
[01:46.750 --> 01:47.730]  Currently, it's very expensive.
[01:48.190 --> 01:52.910]  So a lot of smaller organizations can't do very much of it, unfortunately, a lot of the time.
[01:53.910 --> 01:57.690]  So I had this experience where I go to the Skillshare and I'm like, what about an AI fact checker?
[01:57.870 --> 01:59.250]  We're like, okay, let's try this.
[01:59.250 --> 02:03.230]  And my friend kind of walked me through this process that he came up with.
[02:03.870 --> 02:08.990]  There was this old New Yorker article that I happened to remember that had detailed the fact checking process.
[02:09.750 --> 02:13.390]  So we literally just took this article about the fact checking process.
[02:13.590 --> 02:14.650]  We dumped it into Claude.
[02:15.750 --> 02:17.050]  Claude wrote a spec.
[02:17.330 --> 02:18.910]  We were like, Claude, how would you write a spec?
[02:19.490 --> 02:20.750]  Claude was like, here's the spec.
[02:20.890 --> 02:21.610]  We read the spec.
[02:21.750 --> 02:23.450]  We were like, this is actually a pretty good spec.
[02:23.810 --> 02:25.590]  Then we dumped the spec into Replit.
[02:25.730 --> 02:27.010]  Replit made the program.
[02:27.470 --> 02:28.790]  It took less than two hours.
[02:28.790 --> 02:31.390]  I spent almost that entire time playing with my friend's baby.
[02:32.110 --> 02:35.330]  And hitting buttons in Replit that were like, resolve error?
[02:35.590 --> 02:36.310]  Check, yes.
[02:38.650 --> 02:41.330]  So this was the fact checking program that it made.
[02:44.370 --> 02:47.370]  And again, it just designed all of this itself.
[02:47.970 --> 02:49.990]  I did almost nothing here.
[02:50.930 --> 02:54.650]  It was like, you can input an article and I will give you fact check results.
[02:54.850 --> 02:56.250]  Here are the facts that I have found.
[02:56.250 --> 02:56.670]  You know?
[02:57.430 --> 02:59.950]  So all of this UX was designed by it as well.
[03:00.090 --> 03:02.070]  Like it wasn't just the facts that were checked and everything.
[03:03.730 --> 03:09.270]  And I used the sample text here as I have a post on my blog about my time in the Peace Corps.
[03:09.550 --> 03:13.310]  So I just like, I know that subject really well.
[03:13.390 --> 03:16.950]  So I assumed that I would be able to catch it on factual errors.
[03:16.950 --> 03:18.570]  So that was the text that I tested it on.
[03:18.970 --> 03:20.530]  So here's some of the Peace Corps text.
[03:20.710 --> 03:21.990]  And then it, you know, checks all the claims.
[03:22.150 --> 03:23.230]  It's like, here's the verified claims.
[03:23.250 --> 03:23.930]  Here's my confidence.
[03:25.990 --> 03:26.270]  Okay.
[03:28.310 --> 03:31.870]  So I was like, whoa, I used to think AI might be overhyped.
[03:32.010 --> 03:33.190]  I do not think that anymore.
[03:33.570 --> 03:33.970]  You know?
[03:35.350 --> 03:38.990]  This AI fact checker is not super good at its job, to be clear.
[03:39.330 --> 03:42.730]  It hallucinated many fake answers to the fact check.
[03:43.370 --> 03:51.490]  And what I personally found more disturbing about its hallucinations were that its hallucinations were very convincing, even though I knew the topic extremely well.
[03:51.710 --> 03:56.390]  And so it wasn't just that it hallucinated the fact check and was like, true, untrue, whatever.
[03:56.730 --> 04:02.490]  It hallucinated citations that sounded really plausible and were on real websites.
[04:02.850 --> 04:05.310]  So I would like go to the website and I would try to find the citation.
[04:05.450 --> 04:07.970]  The citation wouldn't be there, but it would be the actual website.
[04:07.970 --> 04:09.590]  I was just like, whoa, okay.
[04:10.490 --> 04:12.410]  So it doesn't work just to be clear.
[04:12.610 --> 04:18.090]  This is not replacing human fact checkers anytime soon, which is great because I love human fact checkers.
[04:18.210 --> 04:21.850]  But as soon as I saw it, I was like, this is crazy.
[04:22.130 --> 04:27.150]  I would never have like put this together in my mind myself if I hadn't happened to be here having this conversation.
[04:27.550 --> 04:29.290]  So I wrote this Substack post about it.
[04:32.130 --> 04:42.590]  And I actually have a history with one of my old blogs where I was a pseudonymous blogger for a long time and between like 2008 and 2013 under the name Clarice Thorne.
[04:43.170 --> 04:47.690]  And I spent a lot of time in sort of the infosec world, both because that was like my friends.
[04:47.690 --> 04:53.870]  And also because I like the closest thing I had to a professional guild was the Electronic Frontier Foundation.
[04:54.430 --> 04:56.710]  Like those are the people who helped me maintain my pseudonymity.
[04:58.330 --> 05:04.150]  So a lot of those people had lost track of me after I stopped writing under that name.
[05:04.690 --> 05:11.730]  But just sort of coincidentally, I had this like extremely disastrous public humiliating broken engagement recently.
[05:12.090 --> 05:18.350]  And because of that, my name was in the media and a bunch of people connected my birth name with Clarice Thorne.
[05:18.530 --> 05:23.010]  And one of those people searched for my blog and found this post about vibe coding.
[05:23.730 --> 05:26.590]  And that person was like, oh, Clarice, you should go to HOPE.
[05:26.850 --> 05:28.350]  So that's how I ended up here.
[05:30.690 --> 05:42.450]  And while I was sort of researching vibe coding after this occurred, you know, I found this quote from this guy, Brandon Wilson, who I don't know, but I was reading one of his articles about vibe coding.
[05:42.970 --> 05:51.550]  And his whole take on this is basically these systems will make talented creatives extremely productive and they will dominate the software industry for the foreseeable future.
[05:51.770 --> 05:52.970]  That's like his take.
[05:53.150 --> 05:54.910]  Like, he thinks it's over for developers.
[05:54.910 --> 05:57.110]  I'm not so sure, but that's his take.
[05:59.150 --> 06:04.350]  So a little bit more about me and just to explain why I speak confidently on some of these subjects.
[06:04.770 --> 06:06.150]  I'm not exactly a hacker.
[06:08.050 --> 06:14.970]  However, I have done a lot of work on media and technology and I have often done hacker like things, I suppose.
[06:14.970 --> 06:19.270]  Some of my friends have actually told me I'm basically like a hacker software engineer.
[06:19.550 --> 06:23.150]  I wouldn't presume that I am one, but like I make websites sometimes.
[06:25.130 --> 06:28.570]  I started as a game designer because I loved MUDs in the 90s.
[06:28.710 --> 06:30.170]  I was really into text based games.
[06:31.610 --> 06:33.890]  I worked in Chicago's oldest bookstore for a while.
[06:34.050 --> 06:38.410]  I have done many web experiments, including a famous blog and other blogs that are not so famous.
[06:38.750 --> 06:43.730]  I ended up working in startup tech, primarily in UX research and content strategy.
[06:43.730 --> 06:45.130]  I've also worked in journalism.
[06:46.290 --> 06:49.150]  And I launched my own print magazine in 2019.
[06:49.490 --> 06:50.330]  It's right here.
[06:50.570 --> 06:54.930]  If anybody would like to see it or buy a copy, this is the magazine.
[06:55.350 --> 07:01.330]  The same friend who recommended that I come to HOPE actually saw this magazine and was like, you should bring your magazine to HOPE because people there are going to love it.
[07:01.410 --> 07:03.130]  So please come look at it.
[07:03.730 --> 07:08.330]  And in particular, if you are an international buyer, I recommend that you buy it from me now in person.
[07:09.970 --> 07:11.850]  But we do ship internationally as well.
[07:15.440 --> 07:20.160]  So just at a high level, I want to talk a little bit about what media is.
[07:22.220 --> 07:25.580]  People basically want any product, anything to do things for them.
[07:25.740 --> 07:26.580]  They wanted to do jobs.
[07:26.680 --> 07:28.040]  They hire a product to do a job.
[07:28.760 --> 07:31.720]  Here are some of the jobs of media products, right?
[07:32.000 --> 07:37.060]  For a game, most people hire the game to do entertainment and inspiration.
[07:37.060 --> 07:41.220]  Sometimes they also get community or they earn money.
[07:41.700 --> 07:46.440]  But typically, like, you know, the greatest games are these like extraordinary artworks.
[07:46.540 --> 07:51.900]  So it's like most people who love games are playing games for a similar reason as like an art aficionado would go to an art museum.
[07:52.160 --> 07:53.780]  It's for entertainment and inspiration.
[07:56.880 --> 08:00.180]  Something like a social network, you know, like Facebook or Twitter or whatever.
[08:01.000 --> 08:04.140]  People typically are joining for community and informational purposes.
[08:04.140 --> 08:05.620]  It's like, what is happening?
[08:05.760 --> 08:13.780]  Like, I don't know if this is still the tagline of Twitter, but I know like the original thing that Twitter used to show you if you were looking at it was like, see what's happening.
[08:14.020 --> 08:18.860]  That was like literally their version of what they thought the product was for.
[08:19.260 --> 08:20.680]  Obviously, it's also for community.
[08:21.020 --> 08:26.280]  For the most part, people join these products, hire them to message each other, organize, etc.
[08:27.280 --> 08:29.460]  Although a lot of them added things onto those court functions.
[08:34.710 --> 08:42.450]  So, one thing that I think is really notable about this group of people and the people who I tend to like to hang out with is that we all love and are committed to the beauty of the Internet.
[08:42.670 --> 08:44.210]  We are inspired by it.
[08:44.370 --> 08:45.270]  That's why we're here.
[08:46.790 --> 08:47.490]  Internet media.
[08:47.490 --> 08:55.130]  On my bad days, I think it's all gonna be money and addiction and it's over and we're all screwed and whatever, you know.
[08:56.050 --> 08:58.630]  But I think we all know in our hearts that it's more than that.
[08:59.050 --> 09:10.490]  And so, on my good days, what I'm trying to talk about here, like I know that what it's really about, any form of media on the Internet or off, is inspiration, ideally true information, and community.
[09:10.970 --> 09:14.010]  These are like really beautiful things, right?
[09:16.470 --> 09:17.810]  Most of us are here for love.
[09:17.990 --> 09:18.970]  I believe in this.
[09:21.390 --> 09:23.770]  So, there's a saying that form follows function.
[09:24.070 --> 09:30.090]  And I personally think that we are still learning how the Internet and code are remaking media.
[09:30.270 --> 09:32.610]  Like we are only at the beginning of this revolution, probably.
[09:33.030 --> 09:38.130]  Absent some kind of like societal collapse, like there's going to be a lot of development from here.
[09:39.530 --> 09:44.770]  One of the things I think is really interesting is that there are many holdovers from print media in online media.
[09:45.070 --> 09:48.470]  So, for example, if you're a writer, a lot of time you get hired by word count.
[09:49.150 --> 09:53.550]  Which is actually pretty weird because an online article can be any length.
[09:53.750 --> 09:55.750]  But typically, people are being hired by word count.
[09:55.990 --> 09:59.290]  And it actually ends up being like a little weird and sort of a dodge.
[10:00.870 --> 10:04.750]  Because like a lot of the time you're being hired by word count and your pay is tied to it.
[10:04.750 --> 10:06.210]  So, you'll be hired at a rate.
[10:06.350 --> 10:08.950]  It'll be like $300 for a thousand words.
[10:09.070 --> 10:13.490]  Then you write 5,000 words and they print the 5,000 words because there's no space limitation.
[10:13.810 --> 10:14.870]  And you still get $300.
[10:15.630 --> 10:19.070]  So, it's like this funny combination of like the new medium and the old medium.
[10:19.330 --> 10:25.690]  Like the entire idea of word count comes from this time when you had to fill a certain number of column inches or something in a newspaper.
[10:28.450 --> 10:33.490]  So, a corollary of this is that old media forms tend to define the new ones for a while.
[10:34.250 --> 10:37.490]  Which I think is just like one of the most interesting things about it.
[10:37.610 --> 10:43.230]  It's like you can have these huge disruptions, but everyone is still working with the assumptions from the old media forms for a long time afterwards.
[10:43.890 --> 10:45.530]  So, I'm going to show you some pictures.
[10:48.530 --> 10:54.990]  So, there is a media experiment that happened in the mid-60s to early 70s called Aspen Magazine.
[10:56.290 --> 10:57.870]  One of the things I love about New York...
[10:57.870 --> 10:58.370]  I don't live here.
[10:58.410 --> 11:00.770]  I live in the Bay Area currently, but I grew up around here.
[11:00.890 --> 11:06.150]  And one of the things I love about New York is there is so much stuff that you can just go do and find and see.
[11:06.310 --> 11:13.310]  And one thing that I learned before this trip is that the Cooper Hewitt Museum has a design archive and you can just go there for free.
[11:13.310 --> 11:15.950]  You can email them and show up in their design archive.
[11:17.050 --> 11:21.290]  And so, I emailed them and I was like, I would like to come look at your design archive.
[11:21.470 --> 11:24.610]  And specifically, I asked if they had old copies of Aspen Magazine.
[11:25.310 --> 11:29.890]  Aspen Magazine was a magazine in the mid-60s, early 70s that shipped in a box.
[11:30.030 --> 11:31.230]  This was like a really big deal to them.
[11:31.350 --> 11:32.570]  They were like, we're shipping this in a box.
[11:32.590 --> 11:33.370]  It's not a magazine.
[11:33.610 --> 11:35.170]  So, it's not this.
[11:35.570 --> 11:38.690]  It's a box this size and shape that has stuff in it.
[11:41.210 --> 11:42.430]  So, that's super interesting.
[11:42.430 --> 11:55.210]  But one of the things that I thought was particularly interesting once I actually made it up there and was in the Cooper Hewitt archive looking at this thing, is that a lot of stuff in this magazine was still basically the same as it would have been in a magazine.
[11:55.430 --> 11:57.530]  So, here we have an example of an ad.
[11:57.870 --> 12:01.910]  This was an ad for Xerox that was in this box magazine.
[12:01.910 --> 12:08.950]  So, they literally made this two-page spread that's exactly the same as it would have been in a magazine and they put it in this box.
[12:09.610 --> 12:11.710]  Even though like the whole point was to subvert the form.
[12:11.950 --> 12:13.850]  I thought that was truly fascinating.
[12:15.450 --> 12:16.990]  Here's some more pictures of Aspen.
[12:18.710 --> 12:23.190]  They did start to do a little bit more new stuff as time went on.
[12:23.190 --> 12:27.350]  So, this was the first issue.
[12:27.630 --> 12:29.890]  And this picture shows you...
[12:30.470 --> 12:31.730]  This is like a little booklet.
[12:31.930 --> 12:35.470]  One of their advertisers, IBM, made this about the IBM World's Fair.
[12:35.670 --> 12:36.110]  Or I think...
[12:36.110 --> 12:38.410]  It was the IBM booth at the World's Fair.
[12:38.610 --> 12:40.270]  So, IBM made this little booklet.
[12:40.550 --> 12:42.290]  They made this booklet about makeup.
[12:42.610 --> 12:43.590]  That's not IBM.
[12:43.750 --> 12:44.550]  Obviously, that's a different brand.
[12:46.050 --> 12:52.770]  And then the only thing that they really shipped as far as I can tell with the first issue that was actually really different was records.
[12:53.210 --> 12:55.810]  So, they did ship records in the box.
[12:56.410 --> 12:57.950]  But I mean, you could do that in a magazine.
[12:58.550 --> 12:59.670]  Magazines do things like this.
[12:59.770 --> 13:02.090]  You can ship a magazine and like stick a record to the pages.
[13:02.750 --> 13:03.250]  So, okay.
[13:03.310 --> 13:04.030]  This is the first issue.
[13:04.550 --> 13:07.450]  One of the later issues, they did start messing around with it a little bit more.
[13:07.650 --> 13:10.830]  So, in this box, you can see the box is a different size.
[13:10.990 --> 13:11.930]  They changed the size of the box.
[13:13.490 --> 13:15.850]  They started shipping more like paper games.
[13:16.070 --> 13:17.230]  They shipped like sheet music.
[13:17.890 --> 13:20.150]  This was a good example of a random thing that was in there.
[13:20.350 --> 13:26.650]  This is like a set of transparent pages that could be rearranged and like piled on top of each other in different ways.
[13:28.050 --> 13:29.010]  So, that was Aspen.
[13:29.650 --> 13:32.250]  Eventually, interestingly, Aspen closed down.
[13:32.250 --> 13:34.890]  At least the reason they said they closed it down.
[13:35.730 --> 13:36.870]  I don't know if this is true.
[13:36.870 --> 13:45.090]  But apparently, the reason they claimed that they closed it down was that the post office changed its rules about what counted as like a box that could be shipped.
[13:45.710 --> 13:47.190]  And they were like, we're done.
[13:47.310 --> 13:48.410]  We're not paying the extra charge.
[13:48.530 --> 13:48.630]  Whatever.
[13:51.310 --> 13:59.170]  So, this is an example of, I think, the form shifting a little bit, but not as much as you might expect given that they were doing something that was theoretically totally different.
[14:00.890 --> 14:06.070]  So, as a corollary to this, here are some screenshots of Unreality, which is a partially vibe-coded app.
[14:06.070 --> 14:11.090]  So, when I was putting this talk together, I asked some of my friends like, who's working on stuff that's vibe-coded that's interesting?
[14:11.430 --> 14:13.150]  And some of them sent me Unreality.
[14:14.350 --> 14:20.450]  Unreality is essentially a social network for reality professionals.
[14:20.690 --> 14:27.170]  So, this is like VR, augmented reality, even interior designers and so on have profiles in there.
[14:27.350 --> 14:29.490]  So, I was looking at it and I was like, okay, this is interesting.
[14:29.670 --> 14:30.410]  They have a critique.
[14:30.670 --> 14:34.210]  This critique is pretty clearly probably aimed at something like LinkedIn, right?
[14:34.650 --> 14:37.370]  They're like, we want a social network that's project-first.
[14:38.010 --> 14:39.750]  We want actual project updates.
[14:39.870 --> 14:42.790]  We want it to be built for collaboration and not for whatever LinkedIn is built for.
[14:43.550 --> 14:47.230]  We want it to have, like, skill-based discovery, which LinkedIn does have, right?
[14:47.350 --> 14:52.670]  But, so, they're taking some of the ideas of LinkedIn, which everyone's pretty familiar with at this point, and they're making this new app.
[14:53.070 --> 14:54.650]  Here are some examples of profiles.
[14:58.010 --> 14:59.110]  They do have the skills.
[14:59.430 --> 15:03.830]  It tends to be, obviously, like, a little bit more fine-grained than you would see on LinkedIn.
[15:06.410 --> 15:12.710]  But they have, like, you know, basic profile pages, lists of skills, like, types of reality that someone works with.
[15:12.750 --> 15:13.870]  This is an interior designer.
[15:14.490 --> 15:19.990]  This person, Tyler Hansen, is a little bit more of, like, a VR experience designer, you know?
[15:20.070 --> 15:21.830]  So, his skills are, like, mixed reality, whatever.
[15:21.830 --> 15:22.530]  Okay.
[15:22.990 --> 15:25.930]  So, I like this.
[15:26.050 --> 15:26.990]  I'm excited about this.
[15:28.790 --> 15:33.890]  It is interesting that this new thing can be made, but it is very similar to LinkedIn, right?
[15:34.270 --> 15:34.610]  Ultimately.
[15:36.790 --> 15:48.410]  So, I'm going to change gears a little bit and talk a bit about what sort of Internet media I ran across in the 90s, which was when I was a young teenager and first fell in love with the Internet.
[15:50.090 --> 15:51.650]  Gemzone 3 was my favorite game.
[15:52.250 --> 15:53.110]  It was text.
[15:53.390 --> 15:54.570]  You entered text commands.
[15:54.690 --> 15:55.510]  It was, like, go south.
[15:55.750 --> 15:57.790]  The game would be, like, you are in a snowy field.
[15:58.070 --> 16:01.050]  You are being eaten by a snow wolf or whatever.
[16:01.590 --> 16:02.790]  It was the first time I played the game.
[16:02.870 --> 16:05.930]  The first time I played the game, I got attacked by a ghost wolf and died immediately.
[16:06.650 --> 16:07.610]  But I came back.
[16:09.310 --> 16:15.250]  Another website that existed was called the Trojan Room Coffee Machine, which may have been the first webcam.
[16:15.470 --> 16:19.670]  I am genuinely interested how many people here already know about the Trojan Room Coffee Machine.
[16:20.050 --> 16:20.530]  Okay.
[16:20.970 --> 16:21.350]  Wow.
[16:22.050 --> 16:23.110]  I feel a little old.
[16:23.210 --> 16:23.570]  It's okay.
[16:23.770 --> 16:24.430]  It's going to be okay.
[16:25.550 --> 16:31.750]  The Trojan Room Coffee Machine was, like, super epic because nobody had ever seen anything like it before.
[16:33.750 --> 16:37.730]  And basically, all it was was a bunch of researchers who had a coffee machine in their office.
[16:38.090 --> 16:41.350]  And they were like, we want to know whether the coffee machine is empty or full.
[16:41.510 --> 16:47.730]  I guess there was some sort of, like, internecine warfare that was linked to, like, whether the coffee machine was full or empty at any given time.
[16:47.990 --> 16:53.070]  So finally, someone just puts a camera on it and they link the camera to a website.
[16:53.070 --> 16:57.230]  And it's like, look, you can monitor the real-time status of this coffee machine.
[16:57.630 --> 16:58.950]  No one had ever done this.
[16:58.950 --> 17:04.970]  And so this website went viral, literally just a live updated video of a coffee machine.
[17:05.110 --> 17:05.890]  That was all it was.
[17:06.830 --> 17:09.030]  It was one flight of stairs up from the engineer.
[17:09.530 --> 17:09.990]  That was it.
[17:10.730 --> 17:11.450]  Oh, was it?
[17:11.650 --> 17:11.990]  Was it?
[17:12.170 --> 17:13.150]  Tell me more about that.
[17:13.350 --> 17:14.670]  You go upstairs to look at the BNG.
[17:16.350 --> 17:16.710]  Right.
[17:17.350 --> 17:17.710]  Yeah.
[17:17.850 --> 17:19.290]  It's like, people would be like, there's no coffee.
[17:19.390 --> 17:20.990]  Why would I go upstairs if there's no coffee?
[17:21.050 --> 17:22.550]  I have to make a new coffee badge.
[17:22.910 --> 17:24.310]  Someone else should make the new coffee badge.
[17:24.450 --> 17:25.070]  So this was...
[17:25.070 --> 17:25.610]  Yes, exactly.
[17:25.930 --> 17:27.570]  So I think it's very funny.
[17:27.570 --> 17:28.370]  Anyway.
[17:29.730 --> 17:32.110]  Again, perhaps the first webcam, right?
[17:32.270 --> 17:35.290]  Obviously, the form of webcam then did many other things.
[17:35.410 --> 17:36.170]  It was not about that.
[17:37.230 --> 17:37.730]  All right.
[17:37.770 --> 17:39.190]  There was also a text site that I found.
[17:39.710 --> 17:40.450]  Age 12.
[17:40.730 --> 17:42.090]  How to pee while standing.
[17:42.410 --> 17:42.970]  You know?
[17:43.370 --> 17:44.790]  Made an angel fire website.
[17:45.030 --> 17:46.350]  It had a fire tile background.
[17:46.570 --> 17:47.410]  I shared links.
[17:47.570 --> 17:48.630]  This sort of thing, right?
[17:48.690 --> 17:49.750]  Like Amazon sold books.
[17:49.870 --> 17:50.750]  It didn't sell anything else.
[17:50.810 --> 17:52.290]  It only sold books, you know?
[17:52.550 --> 17:54.270]  Just like it was a more innocent time.
[17:54.790 --> 17:58.690]  Magazines and newspapers put the text and photos of their articles on websites.
[17:58.890 --> 17:59.810]  Because they didn't know what else to do.
[18:00.230 --> 18:00.950]  They were like, okay.
[18:01.570 --> 18:02.970]  We have this text and these photos.
[18:03.190 --> 18:04.150]  We'll put the articles online.
[18:04.910 --> 18:08.250]  The first banner ad was invented by the CEO of Wired.
[18:08.510 --> 18:10.610]  Because they couldn't figure out how to monetize the Internet.
[18:10.850 --> 18:12.510]  They were like, how are we going to monetize this thing?
[18:12.570 --> 18:13.030]  This is weird.
[18:13.210 --> 18:14.450]  We're going to try banner ads.
[18:14.450 --> 18:17.130]  Again, incredibly similar to old ads.
[18:17.310 --> 18:19.710]  Basically just a picture except on a website instead of in a magazine.
[18:22.230 --> 18:22.630]  Okay.
[18:22.910 --> 18:25.530]  So slowly the Internet became more structured after this.
[18:25.650 --> 18:28.170]  And then that structure itself was subverted and resubverted.
[18:28.910 --> 18:31.390]  Creating and maintaining your own website became harder and more expensive.
[18:31.550 --> 18:34.970]  It used to be that you, if you wanted a blog, you would put text on a URL.
[18:35.210 --> 18:36.030]  That was all you were doing.
[18:37.110 --> 18:39.670]  Eventually it was like, there's going to be second order tools.
[18:39.790 --> 18:41.650]  There was Blogspot, Blogger, etc.
[18:42.010 --> 18:45.550]  You could put text into this format and it would be put in a blog for you.
[18:46.130 --> 18:48.770]  And then there were third order tools like RSS feeds.
[18:48.950 --> 18:51.810]  There were ways to read a bunch of different blog posts at once to subscribe.
[18:52.030 --> 18:54.350]  And now we're in this world of Substack and all this other stuff.
[18:54.490 --> 18:56.110]  These little sub-worlds.
[18:57.630 --> 19:01.190]  So the meta point here is that you can start with a thing that is itself counterculture.
[19:02.190 --> 19:05.810]  And then once that thing establishes itself, counterculture starts on the thing.
[19:06.250 --> 19:06.330]  Right?
[19:08.050 --> 19:14.970]  Just for fun while I was editing this talk, I actually checked to see if Cyber Girl, which was the first blog I ever found, still exists.
[19:15.210 --> 19:18.570]  I remember reading her blog in like 1999 or something.
[19:18.730 --> 19:19.610]  And I was like, what is this?
[19:19.710 --> 19:20.370]  Like, who is this woman?
[19:20.450 --> 19:22.010]  And why is she putting her thoughts on the Internet?
[19:22.170 --> 19:22.410]  You know?
[19:22.870 --> 19:24.470]  So I searched for her blog.
[19:24.690 --> 19:27.730]  And I found a blog that claimed to be the greatest WordPress site in all the land.
[19:27.790 --> 19:30.590]  And I clicked on it and it is just this white box.
[19:31.170 --> 19:34.070]  But it is WordPress because it has the subscribe button down here.
[19:34.750 --> 19:39.690]  So anyway, if anyone knows Cyber Girl or like has a way to get in touch with her, I'm like genuinely interested.
[19:39.930 --> 19:44.750]  If she's like doing something here, I can like read the source code or something, but I didn't have a chance to do that.
[19:47.250 --> 19:51.190]  But anyway, life finds a way is part of the overall point here.
[19:51.470 --> 19:55.770]  There are going to be all kinds of weird usages of structured platforms once you have the structure.
[19:58.030 --> 20:00.870]  Another of my favorite examples is on Instagram.
[20:02.110 --> 20:06.490]  In a lot of areas of the world, cell phones moved faster than laptops.
[20:06.770 --> 20:09.110]  And so the first thing that people got was cell phones, right?
[20:09.210 --> 20:11.950]  This includes goat herders in places like Afghanistan.
[20:12.770 --> 20:21.150]  And so there was a while back where Instagram had this problem where people were using Instagram to like sell goats and sell arms and stuff like that.
[20:21.250 --> 20:25.210]  Like people would upload like a picture of like a Kalashnikov and be like guns for sale.
[20:25.750 --> 20:28.070]  Instagram was like, we're not sure if we want this to happen, right?
[20:28.770 --> 20:30.630]  Instagram certainly wasn't making any money, right?
[20:30.730 --> 20:32.870]  This was just how people naturally used it.
[20:32.990 --> 20:39.570]  They were like, oh, I'm like, you know, in the mountains with my goats and my guns and I would like to sell them to other people.
[20:39.570 --> 20:41.550]  And this is my like natural tool usage, you know.
[20:43.570 --> 20:47.110]  So there's also examples where the structure sort of spins out and becomes its own thing.
[20:48.230 --> 20:50.830]  Like there used to be a lot of crafty people who sold on eBay.
[20:51.410 --> 20:59.790]  Again, I may be one of the only people in this room who remembers this, but like there are people who like sewed beautiful personal clothing and stuff and sold it on eBay.
[20:59.790 --> 21:01.770]  And like for the most part, that doesn't happen anymore, right?
[21:01.870 --> 21:03.910]  They sell on like crafty specific platforms.
[21:06.970 --> 21:11.750]  Importantly, I think the things that people imagine wanting change.
[21:11.970 --> 21:15.130]  There's sort of this process where there's these slight subversions.
[21:15.290 --> 21:18.430]  There's these moments of surprise that people have where they're using the old format.
[21:18.970 --> 21:24.610]  And then the new form solves like some questions or creates new questions.
[21:29.190 --> 21:37.250]  So interestingly, when I went up to the Cooper Hewitt, I love this librarian who helped me because I was like, all I want is, you know, old copies of Aspen.
[21:37.850 --> 21:40.090]  Librarian was like, I'm also going to bring you some copies of Desk.
[21:40.210 --> 21:44.190]  They just like picked up this vibe where they were like, I think you might be more interested in this thing.
[21:44.190 --> 21:51.030]  So they brought me this magazine from the late 90s, early aughts called Nest, which is a super wacky magazine.
[21:51.030 --> 21:51.970]  I had never heard of it.
[21:52.050 --> 21:53.510]  And I am like in love with it now.
[21:53.630 --> 21:56.170]  And they were doing a bunch of really crazy stuff with the form.
[21:56.270 --> 21:59.250]  And I found it honestly much more inspiring than Aspen.
[21:59.910 --> 22:01.590]  So here's some examples, right?
[22:02.130 --> 22:08.750]  I feel like the image is a little color unbalanced, but basically this is a picture of a cover of Nest.
[22:08.750 --> 22:13.930]  They printed a picture of a naked woman on the cover, which was like not okay in the 90s.
[22:15.170 --> 22:23.710]  They also shipped the magazine with a piece of fabric that was like tasteful and matched the magazine and buttoned over the woman.
[22:24.930 --> 22:29.870]  So they were also subverting the form, actually, even though it was like a totally normal magazine.
[22:29.870 --> 22:32.510]  Like you pick up the magazine and it looks like this, just a little bit different.
[22:33.410 --> 22:38.990]  This is another example where they changed the top of the page, like the whole magazine, the top of the page.
[22:39.130 --> 22:40.450]  They cut to look like mountains.
[22:40.450 --> 22:42.130]  This is like an America themed issue.
[22:42.710 --> 22:48.510]  I think we may be ready for a redesign of America also, American brand, American identity.
[22:48.970 --> 23:04.130]  And this was actually part of what I thought was really beautiful about this magazine was like they take the lyrics of America, the beautiful, and they're kind of doing this like this new thing with a lot of that branding in this magazine.
[23:05.230 --> 23:09.590]  But again, totally normal magazine, like a magazine made of paper, no box, whatever.
[23:11.630 --> 23:17.570]  While I was wandering around New York, I was also realizing, you know, obviously all kinds of art forms evolve and re-evolve.
[23:17.570 --> 23:22.890]  So this is like a little bit of a random example, but I just want to show you some things that inspired me because I think inspiration is important.
[23:23.470 --> 23:26.170]  So I was at Remy Flower and Coffee, which is in the financial district.
[23:26.450 --> 23:29.830]  They have a fountain like built into their coffee shop.
[23:30.110 --> 23:35.730]  It's like this interior fountain that they've built to look like it has like plants and like a tree trunk.
[23:35.730 --> 23:36.450]  And I don't know.
[23:36.530 --> 23:38.510]  It's not what I expected to find in a coffee shop at all.
[23:38.570 --> 23:39.310]  They also sell flowers.
[23:40.010 --> 23:43.250]  And this is a Cooper Hewitt design library, which used to be somebody's mansion.
[23:43.470 --> 23:46.310]  So I believe it belonged to one of the Carnegie's or something like that.
[23:46.310 --> 23:54.730]  But if you go up there, one of the things that's so cool about it is just the venue itself is this like amazing mansion with all of these details.
[23:55.730 --> 24:00.870]  This picture is the fastener on the windows.
[24:01.190 --> 24:03.810]  Like they made these unbelievably beautiful fasteners.
[24:03.930 --> 24:06.190]  Like some craftsmen probably took forever doing this.
[24:07.010 --> 24:12.030]  And I was just thinking, you know, yeah, like this is totally just a different usage of this old form, which is the building.
[24:12.030 --> 24:12.610]  Right.
[24:14.430 --> 24:14.870]  Okay.
[24:15.030 --> 24:22.130]  So we're going to shift gears slightly again and talk, introduce this other concept, which probably almost everyone here is already familiar with, which is the long tail.
[24:22.350 --> 24:27.230]  I do want to define it, though, real quick, because this is being broadcast.
[24:27.230 --> 24:28.950]  So there may be people who watch this who are unfamiliar.
[24:29.910 --> 24:37.850]  So in 2008, Kevin Kelly, who was one of the original staff on Wired magazine in the early 90s, wrote a post called 1000 true fans.
[24:38.030 --> 24:49.150]  And his hypothesis was that creatives on the Internet could potentially support themselves with 1000 true fans instead of going through a publisher.
[24:49.930 --> 24:55.450]  All they had to do, he pointed out, was get 1000 true fans each to give them $100 per year.
[24:55.550 --> 24:59.390]  And then they make $100,000 a year, which is like a decent living in most places.
[25:00.950 --> 25:05.610]  So in the context of Internet media, this is typically what people mean when they talk about the long tail.
[25:05.970 --> 25:11.590]  It's not that you have to be super famous or even particularly famous in Kevin Kelly's estimation.
[25:11.590 --> 25:13.770]  It's just that you need a thousand true fans.
[25:14.030 --> 25:14.830]  That's all you need.
[25:20.430 --> 25:27.170]  So one thing I think is interesting here, and that I can sort of attest to from my own history, is that my blog is Clarice Thorne.
[25:27.670 --> 25:30.270]  It was too early to be monetized in this way, actually.
[25:30.690 --> 25:31.870]  Like Kickstarter didn't exist.
[25:32.070 --> 25:32.970]  Substack didn't exist.
[25:33.350 --> 25:35.510]  I was like self-publishing books and selling them.
[25:35.690 --> 25:39.630]  And I was able to get those books pretty high on like Amazon bestseller lists and stuff.
[25:39.630 --> 25:41.450]  So like I published books.
[25:41.730 --> 25:46.810]  One of my books was number one in both of its Amazon categories for an entire week because so many people bought it.
[25:47.570 --> 25:55.690]  But there wasn't like a way for me to just straight up monetize the blog aside from advertising, which I didn't want to do because I didn't want to sell out.
[25:56.370 --> 26:03.050]  Which in retrospect is like endearing given where the Internet is now.
[26:03.170 --> 26:05.190]  But I had like very strong principles around this.
[26:08.350 --> 26:10.970]  So now that the long tail is present, right?
[26:11.110 --> 26:13.730]  There's people like me who can do things like have substacks.
[26:16.790 --> 26:19.810]  Communities can just drive conversations in new ways, right?
[26:19.950 --> 26:22.330]  They can network, they can have their own stuff more.
[26:24.270 --> 26:26.650]  Previously, there had to be a gatekeeper, there had to be some kind of publisher.
[26:26.910 --> 26:28.350]  That's much less true now.
[26:28.570 --> 26:31.850]  I mean, it was possible before to like become a publisher on your own.
[26:31.850 --> 26:33.370]  And it was just like kind of a different process.
[26:35.210 --> 26:39.110]  So obviously, one of my favorite examples here is independent magazines like mine.
[26:39.290 --> 26:46.070]  And a lot of them exist in a new way, reaching this different, this like more focused long tail of subscribers.
[26:46.130 --> 26:47.570]  I actually started this on Kickstarter.
[26:48.490 --> 26:55.870]  Another magazine that I think about a lot is Drift, which is a independent magazine about coffee.
[26:56.350 --> 27:03.290]  And they reached the long tail of like every psycho coffee aficionado in the world, which is actually a lot of people.
[27:03.430 --> 27:08.630]  If you're finding all of the people who are psycho about coffee and are willing to like buy magazines about coffee, that's a lot of people.
[27:09.450 --> 27:15.590]  And they just ship directly to those people and they sell to coffee shops and they have like a magazine business as a result of that, you know.
[27:17.610 --> 27:22.590]  So I sometimes think of this in a way, this is a little idiosyncratic perhaps, as like a fun hack.
[27:24.790 --> 27:34.130]  Because the whole game is to sell an object for more dollars than it costs to manufacture it, which in the world of media these days feels revolutionary.
[27:35.410 --> 27:37.570]  Like, this is not confusing, right?
[27:37.850 --> 27:42.310]  Like, it's not unclear like what is being monetized, right?
[27:42.510 --> 27:43.650]  Like, no one is tracking you.
[27:43.830 --> 27:46.350]  Like, if someone buys this magazine, they're just buying the magazine.
[27:46.690 --> 27:47.930]  Sale price is $42.
[27:48.330 --> 27:51.770]  It costs me maybe $20-ish per issue to produce, right?
[27:51.890 --> 27:53.850]  And then I have to do storage and a bunch of other stuff, right?
[27:54.330 --> 27:55.990]  That's like the business.
[27:58.110 --> 28:01.970]  However, this is a little different, interestingly, from earlier magazines and newspapers.
[28:02.890 --> 28:04.390]  It's a luxury product, right?
[28:04.930 --> 28:08.290]  $42 is very expensive compared to a copy of, like, Vanity Fair.
[28:10.330 --> 28:15.110]  The point is not the information dissemination in the same way that it was before.
[28:15.470 --> 28:17.510]  So, like, newsprint, super cheap paper.
[28:17.710 --> 28:23.870]  Like, newspapers are very cheaply produced because it was like, we want to get this as cheap as possible because people are just buying this to read the text.
[28:24.790 --> 28:26.670]  That's not what's happening anymore, though, right?
[28:26.810 --> 28:31.090]  So there is this change in the form that sort of has come along over a longer timescale.
[28:32.290 --> 28:36.430]  Okay, so we are finally at Vibe Coding specifically.
[28:38.070 --> 28:42.550]  And I am excited about a lot of things about Vibe Coding, potentially.
[28:42.670 --> 28:44.730]  I'm also a little freaked out by it, but I guess everyone is.
[28:45.570 --> 28:47.910]  And here are sort of my top questions.
[28:48.870 --> 28:57.190]  The first one is, like, it's going to be so wild to have new, weird, independent creatives just making whatever they want on the Internet.
[28:57.890 --> 29:01.690]  Like, little daemons that follow their commands.
[29:01.770 --> 29:02.810]  Like, who knows what's going to happen?
[29:03.690 --> 29:06.010]  How will this interlate with old-school hacking and programming?
[29:06.770 --> 29:08.430]  How can small communities use this?
[29:08.630 --> 29:12.210]  How different is it going to become in non-Western cultures, like the goat herders that I mentioned?
[29:13.170 --> 29:16.550]  And what happens next with, like, large media institutions, right?
[29:16.550 --> 29:22.150]  So some of these are institutions that we are attached to, perhaps, and would like to survive.
[29:22.430 --> 29:24.110]  And there are interesting questions there.
[29:24.350 --> 29:28.990]  There's also just questions about how the institutions change themselves in order to survive.
[29:33.810 --> 29:43.590]  So when I first started getting interested in vibe coding, one of my friends sent me this talk by Molly Appleton called Barefoot Developers and Homecooked Software, which I really recommend.
[29:43.950 --> 29:46.170]  You can look it up and just look at it on the Internet.
[29:46.370 --> 29:48.630]  I believe she delivered this talk in Berlin in 2024.
[29:50.290 --> 29:54.910]  And it was just so interesting to me how much she has thought through these ideas.
[29:55.810 --> 30:05.390]  But Molly's idea here is that, you know, we already have, like, this broader software, but what if there was local software, right?
[30:05.530 --> 30:11.190]  And this could be considered, like, home-cooked software, the same way you have a home-cooked meal.
[30:12.070 --> 30:14.770]  A lot of people just go out to eat all the time, whatever, it's fine.
[30:14.770 --> 30:21.910]  But actually, cooking your own food or even growing your own food has a lot of advantages, especially for a certain type of personality that wants a really customized world.
[30:22.870 --> 30:26.570]  Or for people who have, like, special dietary needs or something like that.
[30:26.990 --> 30:29.090]  So I think the metaphor sort of does itself.
[30:30.110 --> 30:31.810]  This graphic is from Molly's talk.
[30:32.110 --> 30:41.230]  And it sort of looks at this broader circle of the communities and then goes into, for me, she points out that people could even make much more software just for themselves.
[30:41.690 --> 30:44.430]  If you're a developer, you're totally used to thinking this way.
[30:44.590 --> 30:46.670]  I am not used to thinking this way.
[30:48.310 --> 30:52.410]  Like, I've always been totally bottlenecked on developer time for any of my weird web projects.
[30:52.410 --> 30:54.150]  Or I have to use, like, off-the-shelf software.
[30:54.370 --> 30:59.050]  So even starting to think about this this way, like, I could just build something is new for me.
[31:02.410 --> 31:05.010]  And there's a lot of stuff that was never built because the money was never there.
[31:05.070 --> 31:06.470]  But that doesn't mean people don't want it.
[31:08.470 --> 31:10.710]  Okay, obligatory slide, problems of vibe-coding.
[31:11.390 --> 31:12.290]  Security is terrible.
[31:12.490 --> 31:14.010]  I do not need to tell this audience this.
[31:14.150 --> 31:15.670]  I do feel the need to point it out.
[31:15.910 --> 31:24.510]  I have heard that vibe-coded stuff, if you admit that something is vibe-coded online, apparently, like, there are some hackers who will take that as, like, an invitation to hack it.
[31:25.750 --> 31:29.370]  So this doesn't necessarily matter for really obscure things, right?
[31:29.490 --> 31:32.930]  So if you're, like, a villager in Africa or something, like, whatever.
[31:33.190 --> 31:34.390]  Is someone going to hack your thing?
[31:34.530 --> 31:34.950]  Probably not.
[31:34.950 --> 31:36.930]  I mean, maybe if it scales up, it becomes a big thing.
[31:37.070 --> 31:38.030]  But, like, how likely is that?
[31:39.110 --> 31:41.350]  So security through obscurity may help here.
[31:43.090 --> 31:47.390]  The other thing, as I already noted, LLMs lie.
[31:48.250 --> 31:50.750]  Or hallucinate would be the polite way to say it constantly.
[31:51.310 --> 31:53.150]  But they also seem very motivated to help.
[31:53.350 --> 31:57.030]  They are incredibly motivated to make you think that they are helpful.
[31:59.030 --> 32:00.310]  And they're really good at it.
[32:00.510 --> 32:06.170]  And so I really just want to reiterate, like, my AI fact-checker was not actually that good, but it wanted to be.
[32:06.370 --> 32:07.970]  And it almost convinced me that it was.
[32:08.330 --> 32:09.750]  It was pretty close, honestly.
[32:10.270 --> 32:13.090]  Like, there were a bunch of the citations that I almost didn't even check.
[32:13.330 --> 32:18.490]  But then just because I think, like, unusually detail-oriented, I was like, okay, I guess I better just, like, check all of these.
[32:18.490 --> 32:19.390]  And then I was like, wait.
[32:20.530 --> 32:26.870]  Is it possible that this, like, 27-page PDF that it's cited on the Peace Corps website doesn't actually exist?
[32:28.910 --> 32:33.130]  So that impressed me because I'm generally slightly paranoid about misinformation on the Internet.
[32:33.130 --> 32:36.350]  But, like, it was, like, good enough to, like, almost get past my defenses.
[32:37.370 --> 32:41.430]  So I think people have to be very on top of their mental game with LLMs.
[32:41.430 --> 32:44.050]  So it isn't just about classic security.
[32:44.050 --> 32:46.050]  It's also about cognitive security.
[32:50.940 --> 32:57.060]  So one of the other questions I'm super excited about is how this helps with community resilience.
[32:58.920 --> 33:00.580]  A lot of things are getting worse.
[33:00.960 --> 33:01.960]  Hopefully things will get better.
[33:02.140 --> 33:05.520]  But many things are getting worse, including, you know, hurricanes.
[33:07.660 --> 33:14.680]  I recently had this experience after Hurricane Helene where a Facebook acquaintance posted on Facebook and she was asking for research help.
[33:14.820 --> 33:16.080]  So I volunteered to help.
[33:17.820 --> 33:23.440]  So in her area of Asheville after the hurricane, there had been some kind of chemical spill.
[33:24.040 --> 33:26.020]  They didn't know what the chemical was.
[33:26.200 --> 33:32.700]  All they knew was that sometimes when people drank the water or tried to wade through the mud to rescue their friends, the water burned their skin.
[33:32.700 --> 33:33.840]  It burned their mouths, etc.
[33:34.480 --> 33:35.900]  She was like, what do we do?
[33:36.840 --> 33:38.540]  She was like, I don't have time to research this.
[33:38.700 --> 33:40.820]  But I was like sitting at home, you know, in the Bay Area.
[33:40.980 --> 33:43.140]  I was like, sure, I will happily help you out, you know.
[33:45.160 --> 33:53.620]  And so I spent like an afternoon trying to research, like, how do you deal with this problem if you're a disaster survivor?
[33:53.860 --> 33:54.000]  Right.
[33:54.100 --> 34:00.800]  And it's actually a very complex question because there are standard ways to deal with water purification under normal circumstances.
[34:00.800 --> 34:03.500]  But this wasn't normal circumstances like normal circumstances.
[34:04.160 --> 34:08.720]  Like when I was in the Peace Corps, I could like boil my water or bleach it or put iodine in it and then it was fine.
[34:08.940 --> 34:15.970]  But if it's a chemical spill and you don't know what the chemical is, then any single one of those actions could actually be counterproductive.
[34:16.620 --> 34:19.980]  Like really almost the only thing you can do in that circumstance is distill your water.
[34:20.600 --> 34:26.380]  If you want to be really safe from whatever this chemical is, a lot of these people also didn't have time to distill their water.
[34:26.540 --> 34:31.120]  So then it was like this question of like, what do I even actually recommend for these people who are in this disaster area?
[34:31.300 --> 34:39.900]  And ultimately, the only recommendation I felt like I could reasonably make was like only use the water that's being shipped in in bottles for anything important.
[34:39.900 --> 34:43.120]  And wear hazmat suits, obviously, if you're like wading through the mud.
[34:47.020 --> 34:52.360]  But obviously, this is a very tragic situation and I was very disturbed while trying to help with it.
[34:53.800 --> 35:04.480]  But I was also interested in the sort of community problem that it posed, which was like, how do we get this information as concisely described as possible and disseminate it?
[35:04.480 --> 35:04.680]  Right.
[35:04.820 --> 35:07.300]  And so the tools we used were Google Docs, Facebook and Twitter.
[35:07.440 --> 35:08.680]  You know, that was what was available.
[35:09.380 --> 35:13.220]  People were like messaging this Google Doc to each other and posting it on Facebook and stuff.
[35:14.120 --> 35:18.480]  But I think that what this demonstrates is more and more we need networks that are small, fast and strong.
[35:19.340 --> 35:23.580]  And that's only going to get more true as situations like this escalate.
[35:25.660 --> 35:30.000]  And so there's this question of like, what kind of software supports that kind of community resilience?
[35:30.680 --> 35:37.080]  I think that the networks that we have are functional in a certain way, but they aren't solving that problem, really.
[35:37.700 --> 35:42.740]  Probably the closest thing we have to a major network that's doing that, that's working on that problem specifically, is Reddit.
[35:43.000 --> 35:44.440]  But Reddit's also doing a bunch of other stuff.
[35:44.900 --> 35:53.720]  So there's this idea that kind of goes around some parts of the software community, which is personal CRM.
[35:54.460 --> 35:56.120]  This is another question I'm actually interested about.
[35:56.280 --> 35:59.700]  Has anyone else thought about this idea at any length in this room?
[35:59.700 --> 36:01.460]  Okay, that's actually really interesting.
[36:01.820 --> 36:02.440]  Okay, yeah.
[36:02.740 --> 36:07.080]  So I looked into personal CRM a while back, because I was really interested in this question.
[36:07.540 --> 36:11.000]  CRM stands for customer relationships management.
[36:11.440 --> 36:15.060]  And typically what people use it for is they're like salespeople, right?
[36:15.320 --> 36:19.580]  So the software giant Salesforce makes CRM software for sales.
[36:20.540 --> 36:23.040]  They charge thousands of dollars for this.
[36:23.140 --> 36:23.820]  It's very expensive.
[36:24.040 --> 36:25.500]  It's also like not that fun to use.
[36:25.500 --> 36:30.420]  So people will like use it for their jobs, but they often won't use it at home to like reach out to their communities.
[36:31.160 --> 36:36.840]  And there have been various attempts by people to make personal CRM, where it's like, I want to stay in touch with my friends more.
[36:36.940 --> 36:38.240]  I want to stay in touch with my family more.
[36:38.360 --> 36:44.840]  Like, how do I like keep track of these connections in a different way so that I have a stronger, more resilient community experience, basically.
[36:48.480 --> 36:51.500]  And this is colloquially sometimes called personal CRM.
[36:51.920 --> 36:53.720]  Almost all of these products have failed.
[36:55.300 --> 36:56.880]  Some like pretty spectacularly.
[36:57.520 --> 37:00.320]  And so there's this question is like, is it because it's never been tried?
[37:00.520 --> 37:02.060]  Has like real personal CRM never been tried?
[37:02.120 --> 37:02.860]  Or is it something else?
[37:02.860 --> 37:10.040]  So I personally kind of think maybe real personal CRM has never been tried, but that's because of developer and venture capital bottleneck.
[37:10.180 --> 37:17.800]  Because when people make personal CRM products that work, typically they've taken VC and the VCs make them pivot towards enterprise.
[37:18.120 --> 37:25.020]  And so even if you had a product that worked really well for your little niche community, the VCs are going to make you turn it into the next sales force.
[37:25.620 --> 37:27.020]  So that's my diagnosis.
[37:27.560 --> 37:29.780]  Some people think that the problem is harder than that.
[37:30.040 --> 37:38.260]  But my overall point here is like, you could have a world where someone like me or some other like weird community nerd makes the next great personal CRM.
[37:38.580 --> 37:46.300]  At least figures out some of the UX problems using vibe coding instead of having to develop it from scratch, get the VC money, figure all of that out, you know?
[37:50.300 --> 37:54.260]  So another question is about how culture affects media structure and usage.
[37:55.140 --> 38:00.780]  I am very interested in how non-Western cultures might create structurally different things.
[38:01.140 --> 38:04.000]  I did a bunch of research on that after I got back from the Peace Corps.
[38:04.500 --> 38:09.520]  As I already mentioned, cell phones and therefore apps leapfrogged laptops in many areas of the world.
[38:09.660 --> 38:12.840]  So there's lots of people in this world who use apps who have never used a laptop.
[38:13.300 --> 38:17.160]  So a lot of their assumptions are very different.
[38:18.800 --> 38:21.260]  Cultural style can also matter a lot.
[38:21.260 --> 38:27.700]  So something I learned while I was doing this research, some of it is a little more surface level or about presentation.
[38:28.060 --> 38:34.160]  So Chinese marketers are more prone to using pictures of teams than individuals because China is a more communalist kind of style of culture.
[38:35.800 --> 38:48.100]  One of the ones that I think is more sort of structurally interesting, people from direct communication cultures, this would include cultures like Israel, the United States, were found to be better at building websites optimized for Google.
[38:48.400 --> 39:01.020]  So like if someone's trying to make a website whose structure is like easily read by the Google search bots, they're going to be probably better at it if they're from Israel than if they're from like South Africa.
[39:05.100 --> 39:08.780]  So this obviously opens up a whole bunch of other questions that are not really what this talk is about.
[39:08.820 --> 39:18.360]  But I am just really interested in this question of like, is there something completely structurally different that happens then in these different environments, these different cognitive environments, these different spiritual environments?
[39:18.360 --> 39:24.320]  like all of that may lead to like super different products in the weird astral space of the Internet.
[39:24.760 --> 39:32.680]  As a side note, I really recommend Jeff Ryman science fiction novel, Air, which when I when I was in Peace Corps, I think I read it then or soon afterwards.
[39:33.060 --> 39:36.660]  And it is a story of technological adoption in the third world.
[39:36.920 --> 39:38.820]  And it is one of the most faithful.
[39:40.040 --> 39:46.340]  Like it's it's a science fiction novel, but it's one of the most faithful portrayals of what that is actually like that I've ever seen.
[39:48.560 --> 39:49.940]  This is another question everyone has.
[39:50.800 --> 39:52.720]  Where is journalism going, right?
[39:53.000 --> 39:55.300]  What's going to happen to musicians, book writers, etc.
[39:56.060 --> 40:02.080]  So one output of the current situation is that many creators are trying to become influencers instead of employees.
[40:04.180 --> 40:05.080]  What does that mean?
[40:05.300 --> 40:06.760]  We're still in the middle of this change, right?
[40:06.880 --> 40:07.580]  It's not done yet.
[40:08.440 --> 40:15.040]  We may end up with a new style of publisher where they do representation as well as publishing.
[40:15.040 --> 40:16.520]  It's like there is this question.
[40:16.620 --> 40:19.380]  If you're a superstar journalist, why do you sign on with a major newspaper?
[40:19.960 --> 40:21.840]  Like in some cases, it's really clear.
[40:22.000 --> 40:24.860]  It kind of depends if you like like certain parts of the job or not.
[40:25.040 --> 40:31.040]  If you don't like the marketing and you just want to write and have your column published, maybe it makes sense for you to stay at the New York Times or whatever.
[40:31.420 --> 40:34.500]  If you like the marketing, maybe there's no point.
[40:34.680 --> 40:35.920]  Maybe you just go off on your own.
[40:36.200 --> 40:39.500]  Maybe then the publishers have to offer the journalists more.
[40:39.500 --> 40:41.400]  Maybe they offer them speaking representation.
[40:41.640 --> 40:43.500]  Maybe they do other value ads to keep them.
[40:45.400 --> 40:50.320]  So one of my friends, I was practicing this talk with him and he was like, yeah, I guess theory A is that it's over for journalists.
[40:50.680 --> 40:53.840]  But theory B is that it's over for journalism companies, right?
[40:54.320 --> 40:55.700]  Again, we haven't seen this play out.
[40:56.020 --> 40:59.160]  My guess is a lot of the big companies are going to adapt and be okay.
[41:00.240 --> 41:01.440]  But it will be different.
[41:02.460 --> 41:14.480]  And then one of the things that I find perennially fascinating is, and is one of the things that I'm really interested in doing more R&D on at my magazine, is often communities can figure out stuff faster than individuals.
[41:14.780 --> 41:25.880]  So there's this phrase, the Internet detective, which was popularized in sort of the something awful days, when like something awful Internet detectives would like find someone for good or for ill, sometimes for ill.
[41:25.880 --> 41:33.840]  But Internet detectives are really good at finding things out usually, often better than journalists, often better than anyone else.
[41:34.060 --> 41:37.940]  Like sometimes some fact or situation is trying to get figured out online.
[41:38.040 --> 41:40.700]  And it's figured out in like 30 minutes by like some guy.
[41:42.760 --> 41:50.160]  And so one of the questions that I find really interesting here is like, what if you had a journalism company that was more integrated with a community?
[41:50.600 --> 41:52.640]  Like right now, a lot of journalists are pretty separate.
[41:52.760 --> 41:54.500]  They kind of live in this like ivory tower.
[41:54.500 --> 41:57.980]  I might make fun of it a little bit, but that's kind of the feeling, I think.
[42:00.180 --> 42:05.980]  But what if you had a situation where like, you had upvoted community posts where like real fact checking was applied to them?
[42:06.220 --> 42:11.220]  You sort of have this on Twitter right now with the community notes, but it's not like a professional fact checker.
[42:11.320 --> 42:17.000]  You could imagine a situation where there was like a community figuring stuff out, and then it was being fact checked by a professional fact checker.
[42:17.580 --> 42:21.700]  Or you could do like collaborative articles with the Internet detectives, right?
[42:22.620 --> 42:24.300]  I think stuff like this is really exciting.
[42:24.300 --> 42:26.720]  And we have like not even come close to figuring it out yet.
[42:27.080 --> 42:36.680]  And one of the reasons it excites me is that a lot of what I think is happening with epistemology on the Internet is happening like in part because we haven't figured some of this out.
[42:36.680 --> 42:39.780]  Like we have all these issues with truth and misinformation and all of that.
[42:40.280 --> 42:49.240]  I think it would be really cool if there was a different attitude towards including more people in the process of information gathering.
[42:51.440 --> 42:53.980]  Okay, so this is my last slide.
[42:54.940 --> 42:56.760]  These are the core questions that I asked.
[42:58.500 --> 42:59.860]  My name is Lydia Lawrenson.
[43:00.160 --> 43:01.220]  I have this magazine.
[43:02.060 --> 43:04.260]  We can do Q&A or we can just call it.
[43:04.420 --> 43:05.320]  Does anyone have any questions?
[43:13.510 --> 43:14.990]  Why is it called vibe coding?
[43:15.550 --> 43:16.730]  Why is it called vibe coding?
[43:19.350 --> 43:24.010]  So I believe the phrase was coded by, I'm not going to pronounce his name correctly.
[43:24.570 --> 43:24.930]  I'm sorry.
[43:26.270 --> 43:27.070]  Andre Karpathy.
[43:27.770 --> 43:28.770]  It's like some guy.
[43:29.290 --> 43:31.530]  There's a tweet that he, there's a tweet that he wrote.
[43:31.670 --> 43:33.530]  I can find it on my phone real quick and read it to you.
[43:34.070 --> 43:38.790]  But he's, I think he's at one of the big AI companies.
[43:41.090 --> 43:52.370]  And he had some tweet, like towards the beginning of this year, where he was talking about how like, he no longer codes like, based on like, you know, the old stuff.
[43:52.450 --> 43:52.610]  Oh, yeah.
[43:52.710 --> 43:53.250]  Okay, here we are.
[43:53.250 --> 43:55.850]  There's a new kind of coding I call vibe coding.
[43:56.010 --> 43:56.790]  This is Andre.
[43:57.130 --> 43:57.490]  Andre.
[43:58.390 --> 43:59.990]  Where you fully give in to the vibes.
[44:01.170 --> 44:04.750]  You embrace exponentials and forget that the code even exists.
[44:05.450 --> 44:08.130]  It's possible because the LLMs are getting too good.
[44:08.790 --> 44:12.630]  Also, I talk to Composer with Super Whisper, so I barely even touch the keyboard.
[44:12.990 --> 44:17.770]  I ask for the dumbest things like decrease the padding on the sidebar by half because I'm too lazy to find it.
[44:18.230 --> 44:21.210]  I accept always I don't read the diffs anymore.
[44:21.470 --> 44:24.670]  When I get error messages, I just copy paste them in with no comment.
[44:24.850 --> 44:25.970]  And usually that fixes it.
[44:26.190 --> 44:28.830]  The code grows beyond my comprehension.
[44:30.330 --> 44:35.510]  Sometimes the LLM can't fix a bug, so I just work around it or ask for random changes until it goes away.
[44:38.710 --> 44:41.510]  Anyway, as anyone who's worked in software development is aware.
[44:41.690 --> 44:44.690]  That is a very different attitude from standard programming typically.
[44:46.150 --> 44:47.070]  Okay, next question.
[44:47.190 --> 44:47.710]  You were in the back.
[44:48.250 --> 45:05.630]  Yeah, I mean, so my understanding is that at least present, a lot of the technology enabling this is not really accessible in the community space because it involves huge capital outweighs for owning a ton of GPUs to train these models or just having access to in-organized data sets.
[45:06.510 --> 45:16.690]  What do you think about that causing problems potentially for the tenability of sort of grassroots adoption of this technology?
[45:17.030 --> 45:23.990]  Yeah, I do think that we are in an interesting time right now where many of these products are probably cheaper than they're going to be in the long term because...
[45:23.990 --> 45:24.550]  They're subsidized.
[45:24.590 --> 45:25.470]  Because they're subsidized.
[45:26.710 --> 45:32.110]  With that said, the experiment that I described doing, I did for like...
[45:32.110 --> 45:33.370]  I think...
[45:33.370 --> 45:35.010]  So all the tools that I used were free.
[45:35.290 --> 45:39.030]  The only place I actually had to put money into the system was for API calls.
[45:40.890 --> 45:42.410]  When I was using Replit.
[45:42.710 --> 45:44.010]  Like, I used Replit for free.
[45:44.170 --> 45:45.170]  I used Cloud for free.
[45:45.850 --> 45:46.730]  Replit wrote...
[45:46.730 --> 45:49.110]  What Replit wrote was using API calls in OpenAI.
[45:49.410 --> 45:51.710]  So I had to give OpenAI money for the API calls.
[45:51.870 --> 45:53.170]  I maxed it out at ten.
[45:53.350 --> 45:54.470]  It hasn't hit the maximum.
[45:54.550 --> 45:56.190]  I've used this thing a bunch of times.
[45:57.850 --> 45:58.250]  So...
[45:59.450 --> 46:05.510]  It could get more expensive and still be pretty accessible to community users, I think.
[46:05.630 --> 46:06.370]  At least in America.
[46:06.630 --> 46:09.870]  Obviously, if we're talking like rural tribesmen, that's like a totally different story.
[46:10.470 --> 46:10.870]  But...
[46:10.870 --> 46:14.190]  I mean, you also have like issues including who owns...
[46:15.110 --> 46:15.230]  Yeah.
[46:15.930 --> 46:16.550]  Oh, yeah.
[46:16.710 --> 46:21.870]  I mean, and one of the things I think is really interesting, for instance, is like a lot of people are using LLM's as their therapists right now.
[46:21.870 --> 46:22.970]  So they're like...
[46:23.890 --> 46:28.290]  And like, this is really interesting from like, again, a cognitive security perspective.
[46:28.470 --> 46:30.950]  Like, first off, the LLM is not a human with human motivations.
[46:31.330 --> 46:36.790]  It also does not have like professional standards.
[46:37.710 --> 46:39.290]  Like, and so...
[46:39.290 --> 46:41.650]  I mean, I understand distrusting therapists.
[46:41.850 --> 46:43.470]  I think sometimes therapists really mess up on this.
[46:43.550 --> 46:44.850]  And there have been some high profile cases.
[46:45.130 --> 46:51.350]  But like, I don't think that like a random database in a company you barely know is like safer than a therapist to tell your problems to.
[46:51.350 --> 46:53.410]  So there's a whole bunch of stuff here.
[46:53.650 --> 47:02.010]  I mean, one thing on the expense thing, though, I think it will be so valuable to have this territory that it's going to be subsidized for a long time.
[47:02.210 --> 47:10.670]  Like, Facebook did a really interesting thing called Facebook Zero for a while, where they subsidized all Facebook use in developing countries for years.
[47:10.670 --> 47:12.210]  I think they might still be doing it.
[47:12.370 --> 47:17.570]  So like, if you were just like a random person and Eswatini using your cell phone to access Facebook, you could do it for free.
[47:17.570 --> 47:18.350]  They just paid for it.
[47:19.110 --> 47:26.670]  So I could see a version of that where one or more of these companies do it really cheaply or freely because they want to really control and dominate the vibe coding space.
[47:29.950 --> 47:30.650]  Go ahead.
[47:30.990 --> 47:31.510]  And then you can go.
[47:32.110 --> 47:36.970]  I'm really interested in what you said about it's open for journalism companies, not journalists.
[47:36.970 --> 47:50.990]  I'm finding that really compelling in other contexts, too, like in movies and stuff where, like, actually, I don't know, I think, like, artists and writers and whoever are just going to keep going, like, whether they're paid or not, a lot of us are doing this work.
[47:51.150 --> 48:04.270]  Anyway, so it's like, I don't think that the, maybe, like, big players are kind of going to all start using AI to cut costs and therefore, like, publish the same stories and, like, optimize, right?
[48:04.270 --> 48:18.890]  Like, all the content is just going to keep being optimized, but actually, like, more and more people, like, might just realize that they're, like, in that optimization story versus, like, in stories that are way more interesting and kind of more on the margin.
[48:18.890 --> 48:21.170]  So I feel like the margins will just grow.
[48:21.750 --> 48:37.930]  But I'm curious, like, yeah, what you kind of see happening with that and what that looks like, I guess, for these institutions, like, whether you agree with that or if, like, I don't know what that means for the margins, I guess, also.
[48:38.610 --> 48:41.090]  Yeah, yeah, you covered a lot of territory in your comment.
[48:41.490 --> 48:54.650]  I guess the overall thing that I'm just perennially obsessed with is, like, I mean, I think you're right, people are inspired, like, people just want to make stuff, right?
[48:54.650 --> 48:58.510]  Like, like, ultimately, this is why I'm in this business, right?
[48:58.770 --> 49:00.330]  I would probably do it for free.
[49:00.470 --> 49:02.050]  Like, please buy my stuff anyway, you know?
[49:02.490 --> 49:06.690]  Like, I just, like, have done this for a long time.
[49:06.850 --> 49:07.650]  I really love it.
[49:07.910 --> 49:17.150]  You know, I am perennially in this world of trying to find the right business model to make it possible for me to keep working.
[49:17.150 --> 49:20.710]  I think most people don't have that kind of relationship to their careers.
[49:20.830 --> 49:26.910]  But a lot of journalists, writers, and media people have that relationship to the work that we do.
[49:27.270 --> 49:33.590]  And the question becomes much more like, how do I literally just find the money and the stuff that lets me keep working?
[49:35.410 --> 49:40.130]  Like, one example that I find always really moving when I think about it is,
[49:43.330 --> 49:52.990]  I don't know if this is still happening, but there was apparently a while in Mexico where there was this sort of like loose group of moms on Twitter, who would post where gang violence was happening.
[49:53.330 --> 50:02.050]  So they were essentially like independent journalists literally just being like there's a shootout at the gas station, you know, because there was so much cartel violence in their areas.
[50:03.030 --> 50:08.530]  Cartels didn't like this, they preferred a situation where people were confused about where violence might happen and when.
[50:08.530 --> 50:18.630]  And so the cartels would occasionally like gruesomely murder these moms, but they were doing this work entirely because they cared about their kids and their communities, right?
[50:18.730 --> 50:22.890]  That was why they took this risk, which they knew was a risk in order to get that information out there.
[50:23.570 --> 50:30.870]  And like, obviously that's like never going to die, you know, like that impulse, as long as they're humans, will be with us.
[50:32.810 --> 50:33.710]  But I don't know.
[50:34.190 --> 50:35.950]  I don't know how we're going to scaffold it.
[50:35.950 --> 50:38.570]  I was going to say you're over time, but we don't see the next speaker.
[50:38.730 --> 50:39.770]  So feel free to...
[50:39.770 --> 50:40.810]  Okay, sweet.
[50:40.890 --> 50:41.190]  Thank you.
[50:41.770 --> 50:43.070]  The next speaker is here.
[50:44.890 --> 50:45.810]  You had a question.
[50:46.450 --> 50:53.030]  I guess it's, I want your take on a classic problem, which is a basis of trust, right?
[50:53.110 --> 50:56.970]  Because you talked about reputation systems, things of that nature.
[50:58.710 --> 51:01.170]  Getting back to kind of Molly's point, right?
[51:01.630 --> 51:04.770]  Do it for yourself and do it for your family and do it for your community.
[51:05.590 --> 51:11.430]  How do you balance that against sort of a social trend towards bad actors?
[51:11.430 --> 51:16.950]  How do you think that fits into the idea of vibe coding as an empowering force?
[51:17.210 --> 51:17.750]  Yeah.
[51:18.090 --> 51:30.950]  Well, I mean, I think humanity is in this sort of permanent state of having to balance against bad actors, you know, both internally among humans and forces like disease or whatever.
[51:32.910 --> 51:37.510]  With vibe coding specifically, I mean, I assume that the security stuff is really going to escalate.
[51:37.850 --> 51:39.270]  That's kind of one of the most obvious things.
[51:39.470 --> 51:44.550]  It's just like there will be better off the shelf security tools.
[51:44.710 --> 51:47.110]  Probably there will be better off the shelf security attacks.
[51:47.310 --> 51:50.850]  It'll rebalance in some kind of weird new equilibrium.
[51:51.390 --> 51:52.190]  I don't know.
[51:52.690 --> 51:55.510]  Again, I'm sure there are many people in this room who know way more about that than I do.
[51:57.250 --> 52:01.550]  When I think about trust, my whole thing is always just like
[52:06.380 --> 52:13.720]  a lot of bad actors are pretty continuously trying to spoof the trust of other people.
[52:13.980 --> 52:17.600]  But there is a real basis of trust that exists.
[52:18.300 --> 52:25.440]  Like and I feel like that part of the conversation often gets like lost in ways that I often find very annoying.
[52:25.440 --> 52:31.600]  Like I've been part of many journalism conversations where it's like, we'll be more trusted if we make a website that looks more trustworthy.
[52:31.920 --> 52:38.600]  And I'm like, or we could go back to basics and figure out why people trust journalists and do more of that.
[52:38.740 --> 52:40.280]  You know, like be trustworthy.
[52:40.480 --> 52:41.420]  This is like my whole thing.
[52:42.420 --> 52:44.420]  And sometimes I feel like I really talk to people who get it.
[52:44.520 --> 52:45.880]  And sometimes I feel like I really don't.
[52:48.060 --> 52:52.120]  But again, it's like I think there's these sort of higher motivations, the inspiration and the connection.
[52:52.120 --> 52:54.140]  And that is all founded on trust.
[52:54.420 --> 52:58.560]  And so one of my whole things in my recent career is just like, how do I do more of that?
[52:58.680 --> 52:59.980]  Like, how do I work with trusted people?
[53:00.160 --> 53:01.240]  How do I be trustworthy?
[53:01.980 --> 53:05.560]  How do I just like do more of this thing that is obviously good?
[53:07.220 --> 53:11.140]  And try not to get too lost in questions of like bad stuff.
[53:11.820 --> 53:12.280]  All right.
[53:12.420 --> 53:12.920]  Thank you guys.
[53:12.920 --> 53:13.820]  Thank you.