Beginnings, Digital Gardens and AI Delusions | Decapsulate
Decapsulate

Beginnings, Digital Gardens and AI Delusions

·
skip to discussion

We get excited about the stats on episode 1 and talk a little about podcast logistics. Tris then does a deep-dive into Digital Gardening and Robin talks about the ethics of TESCREAL.

📖 CHAPTERS

  • 00:00 The podcast has landed
  • 00:37 Feedback on feedback
  • 04:40 TRIS’S CONSPIRACY CORNER
  • 08:04 Saving Robin’s life with Digital Gardening
  • 27:55 TESCREAL
  • 46:02 Typing with LLMs

Ours

External

🧑 CREDITS

Decapsulate is a NAMTAO Production (namtao.com)

It is hosted by:

This work is BrainMade (https://brainmade.org)

Transcript

DC3

Updates & Housekeeping

[00:00:00]

Tristram: So here we are, episode three.

I have been having a great time setting up all the infrastructure. I’ve crunched the stats and between YouTube and the podcast feed, there have been so far 4,000 plays of episode one, which is insane.

Robin: Is amazing.

Tristram: we a week into it? That’s wonderful.

Robin: but do we

Tristram: I.

Robin: Is that does that include crawlers and things or do we know their real

Tristram: Hmm.

I think we’re okay with this data because it falls off organically, like huge numbers of people listened on day one, two, and then it has fallen off more organically. Whereas a crawler, I would expect periodic or something more like that.

And certainly not MP three, like crawling, an MP three is a bit weird, But anyway, the point is that these 4,000 people are incredible. Presume the listener is one of those peopleThat is, astonishing and I’m very grateful as I’m sure you are.

Robin: you. Yes.

Tristram: Yeah.

Robin: I never

Tristram: So

Robin: to be as popular as it’s been already.

Tristram: yes.

Robin: We haven’t really figured out the format of the show yet.

Tristram: Yeah I think it’s quite nice to figure it out, almost with the listener, although in a very, one way relationship there, but we’ll figure it out as we go. I’ve got a bit of housekeeping here, so thanks everyone who commented and has been chatting on the Discord, this is all fantastic.

The feedback’s been wonderful. And we are especially grateful to those who’ve signed up to the Patreon patreon.com/deencapsulate or it’s all on the website. The patrons get at the moment, early episodes,

Robin: website,

Tristram: so when it,

Robin: is deencapsulate.com,

Tristram: yes you’re a natural robin, you’re a natural. You’ve just gotta say it all the time, repetitive, but we, if we don’t do it, who else will?

Robin: Early episode. So when, for example, when this episode drops in the public feed, patrons will have access to episode four already. Yeah

Tristram: We’ve also got a merch store. You go to encapsulate and click on store. There is exactly one item. Available for purchasing. And that is the square dashed logo thing. I’m not even sure if it is the exact final design, but what did you say this was gonna be Robin?

Like an early

Robin: first edition.

Tristram: low Yes. First edition before Tris has quite got the logo symmetrical. I’ve been learning Vector illustration using Inkscape, the fantastic open source tool that is in some ways better than an illustrator.

we get almost no commission from this. This is more about a fun thing for the community to do to publicly let you set your commission. And I think the price is already, like the base price is already, correct for that sort of thing. I’ve set it as low as possible.

To support us, Patreon is the best option, but you also don’t need to do a thing. You can just keep listening, keep recommending things to your friends. As I said in episode one, this is the only way podcasts can work is word of mouth. There’s no algorithms. Although I have now submitted the show to about a dozen different podcast indexes and stores and so forth.

I think we’re gonna talk about feedback.

yes.

Robin: about the fact that this is we’re, we are doing this together with the user,I thought that would be a beautiful segue into talking about feedback because that’s how, it’s two way.

Tristram: Right, well Reminded. So there are a few ways you can give us feedback, but the best way for our organization, and so that I see it and Robin sees it and we both can be involved in the conversation, is on the website to capsulate.com. If you go through to each individual episode in addition to the full transcript, we’ll go about that later.

In addition to the full transcript, at the very bottom of the page, there is a comment box, but this comment box is powered by GitHub discussions. So this has a huge advantage in that all of the discussion is actually in one place. If you look at the discussions on GitHub, you can search through them, you can attach files, you can write in markdown, all of the good stuff that we have come to expect from GitHub.

But it’s also this nice little comment box. And this is powered by Guus, G-I-S-C-S app, which is an open source GitHub discussions comment system. Really happy about that. And most importantly, probably for the listener, this emails me when you comment. Now, it may or may not email Robin. Your notification settings are your own, but I will at least see it

Robin: No, I

Tristram: we can chat about it.

Robin: my notifications for it. I hadn’t thought of that yet, but I’ll make sure I do it.

Tristram: Yeah, selective notifications. I’m a big fan of notifications off by default in general, but turning on very important things, like your partner your teachers, that sort of thing. So if you’re listening to this, that’s how to give us feedback at the bottom of the episode page on the main site, comment using the discuss or go straight to the repo if you’re familiar with GitHub, and start making some new threads.

Give us some suggestions, some [00:05:00] corrections or whatever. There were quite a lot of mistakes in the episode one video I used descrip, which is what we’re using to record this and do the initial edit which is a great tool.

The problem is it makes mistakes into the video, and I can’t fix them because YouTube do not let you, do not let you make changes or re-upload a video. I think it is possible. If you are a very fancy YouTuber, like millions of subscribers and you’ve got a personal connection to a YouTube manager or something, you can beg them to do a re-upload

But that is not for mere mortals like us. I dunno if I’ve ever told you my theory, Robin, about why YouTube doesn’t let you edit videos, which is strange.

Robin: No I don’t think so.

We’ve immediately gone off topic, but I think it is, it’ll be interesting to our listeners at least. Let’s try and keep it short.

Tristram: Yeah, exactly. We got stuff to get through. The, I think the biggest reason that YouTube do not let you edit your videos is not for like fraud reasons so that you could get a load of videos and then change the video to make it look like people liked something that it wasn’t or anything like that.

It is because as with everything Google Do, it all comes back for advertising. The way advertising works in podcasts is that Robin and I could approach a podcast advertising network and they could splice in their adverts into parts of our episodes. I really. Don’t think we will do that for advertising, because I personally hate those.

I prefer host red ads, but it is possible to automate this and keep things fresh so that you listen to it one week, you get one ad, you listen to it. The next week you get the more updated ad, which is presumably better for everyone. YouTube runs their own ad platform, and if they let you edit your video, they would let in competing ad platforms

Robin: Oh,

Tristram: because you could splice your own ads later on into videos.

You see what I mean?

Robin: People do put ads in videos anyway, but you are saying they can’t be changed after you’ve put them in

Tristram: Yes. That’s a big feature of these podcast ad networks is that they will keep your ads up to date, which increases yield for everyone,

Robin: yeah.

Tristram: I feel that is a problem for podcasting, at least in the way I’d want to do it, because people give over their podcast to these advertising networks to then put these ads in between. And I think often, I particularly listened to quite a lot of leftist, political shows

Aspace communism shows. Yes, I know.

Robin: and there’s quite a lot of things like, so I was listening to like the intercepted podcast, and only adverts in this podcast at the beginning of the end were two. British voiced, and therefore, not very clearly notfrom the original feed for Starbucks. And a big a big topic of intercepted in general is like Gaza and Israel and Palestine and all that kind of stuff.

And Starbucks are one of the, one of the boycott targets. So it was rather jarring, I sent

Tristram: oh.

Robin: about it but you lose control of exactly what adverts on your podcast, which concerns me a bit.

Tristram: I would like to promise to the listener at this stage that we will never do something like that and we will run Host Red Ads if and when we are lucky enough to be able to get income through the show. It will be Host Red Ads by Robin and me from companies that we genuinely love and are not like destroying the planet.

And if you disagree, Robin will delete this from the edit and nobody will ever hear this, but I bet you agree.

Robin: I certainly agree.

Tristram: Good, good.

Digital Gardening

Tristram: Have I got a great, wonderful little slice of what kind of feels like the cozy internet for you. So I’ll start off with some definitions. Digital gardening is a way of working in the open. So instead of working on a, let’s say, a blog post for many months and getting it all perfect and then publishing it,

Robin: Who could you be referring to?

Tristram: Anyone, anyone could do this perfectionism is a great curse in our society instead of working on that privately and trying to get it as good as you can.

Digital gardening is where you work in the open in a interconnected series of notes and you potter by which I mean you do a little here, do a little there. You’ve got a lot of ideas, a lot of documents, a lot of posts, a lot of thoughts, half done, and you bounce between them, improving them all

Robin: Yeah.

Tristram: bits.

Robin: you do a little bit of weeding here. You kinda bend this round so it can

Tristram: That’s it.

Robin: To the vine. To, to the like trellis a bit better over there,

Tristram: ah, the dream.

Robin: a while. Yeah.

Tristram: It is absolute hobbit core. Publishing, like very cozy, very just a dream. It feels very much like the scrappy websites of old, they were always under construction, weren’t they? You never have an under construction website. Now it’s always, this is my perfect brand.

Please subscribe to my substack. It’s very polished and boring. if you were to visit a digital garden [00:10:00] like namtao.com, the first thing you’ll see is not posts reverse chronological lauder, like a blog. If you’ll excuse me, like if you go to robin winslow.co uk,

Robin: Oof calling me

Tristram: uh,

Robin: Oh

Tristram: I’m not saying that is bad.

I’m saying you have got a blog, whereas a digital garden is a web of linked pages,

Robin: yeah.

Tristram: very much like what the web was designed for. The clue is in the name. It’s not a timeline, which are, I dunno about you, but I am real tired of timelines these days. But it’s more something that you’ve gotta explore.

Things aren’t super signposted. You might have to go to one page, then another, and then circle back and get lost a little bit. It’s not very monetizable. It’s not very contenty but it is lovely. That is digital gardening. I will make sure in the show notes we’ve got a link to Maggie Appleton’s page on the history and ethos of the digital garden, I believe is what that is called.

Digital gardening. There’s a lot of videos on YouTube of people who’ve taken up digital gardening who were inspired by Maggie’s amazing blog post In it be very cleverly signposts. How finished each page on her site is by describing them as seedlings ideas that have been planted buds, the seeds that are started sprouting and trees maybe, fully formed ideas that she,

Robin: is so amazing. This is like exactly what I need.

Tristram: I knew it would be. That’s why I’ve brought it here for you. So this is lovely. This is exactly what you want and it fits really nicely into obsidian namtao.com is based on the Obsidian Digital Garden plugin, which lets you put in the front matter a checklist that says, Hey, publish this page, and then handles all of the linking, uploading of images.

And it even supports like mermaid diagrams and incredibly obsidian data view, which is a third party plugin, not core to obsidian, but it might as well be core to obsidian. It’s very nice. perhaps I’ll find a few other examples of digital gardens.

Robin: So sorry, what’s the relationship between namta.com and obsidian?

Tristram: namta.com is created from markdown documents inside my obsidian fault.

Robin: So it’s created for markdown documents in the same way that any static site is. Is there like tech, as in from that markdown that leads to the website. Does that share anything with obsidian or is it just two different systems?

Tristram: It

Robin: the same

Tristram: is extreme. It’s extremely deeply integrated. There is an obsidian plugin called Digital Garden, and this gives you like a command in obsidian to publish single pages or all pages that have the checkbox DG publish in the properties in the MET data.

And this pushes up to GitHub using the GitHub, API not.

Using Git. The command line uses the GitHub, API, to directly add these files to the repo. This is clever ’cause it means it works on mobile where you don’t have access to the Git client. So it uses the GitHub, API to push these files there. And then on the GitHub side, there is a GitHub action, which crunches those markdown files into a website like any other static site generator.

It’s just that these are designed to work together. You just, clone and copy their repo and just get going. so the advantage of this over a more traditional static site generator is that in a traditional static site generator, you can have unpublished markdown, but it has to be in the GitHub repo.

It might have a little checklist that says, don’t, this is a draft, but it’s still in public. People can still look at it on the GitHub repo

Robin: But with the digital garden plugin, you keep all your notes private in your obsidian vault, which can be private on Dropbox using obsidian sync, whatever you like. And you only push the documents as you want to make them public.

Sorry. Not by putting ’em in a folder

Tristram: no by putting this front matter property called DG Publish. And it’s a Boolean True. False.

Robin: puts it in gi.

Tristram: Exactly.

So it’s just what I was looking for because I want all of namtao.com, which has my video scripts and everything, including lost terminal scripts, lost terminal characters.

Like everything. I like it. It’s the whole thing for all of my projects. I want those to also be on GitHub in markdown, that people can give me poor requests for if I make mistakes or if they want to improve them or whatever.

but I also wanted privacy because this is just my normal vault with obsidian.

one of the big rules that I learned very early on is don’t have multiple vaults, have one vault and figure it out.

Robin: yeah.

Tristram: Because interlinking is what we’re doing here. I didn’t expect such a strong reaction from you, Robin, but I’m delighted. ’cause this, in hindsight, this is exactly what you need, right?

Robin: I was actually, I was finding it quite funny when you said because it uses the gout API, [00:15:00] therefore it works on mobile and it might seem like such a trivial point unless you’ve tried to do this thing of to like, I don’t know, blog on the train on your phone or whatever. it’s so important. it’s such a centrally important feature. I can’t even it’s, yeah. Anyway, sorry. how long have you been keeping this from me?

Tristram: Maybe

six months.

I can only apologize.

Robin: it.

Tristram: Look, it actually gets better

Robin: acceptable.

Tristram: the heading I’ve got for this section is digital gardening and Aze Carston. So I’ve realized that they are the same thing. I don’t think I’m the first person to figure this out,

Robin: Yes.

Tristram: but they’re extremely. Compatible because in z Augustine, you take your fleeting notes and thoughts, ideas that you’ve had by while reading or listening or watching something, and you process them into permanent notes sometimes called atomic notes.

And these are single units of teachable knowledge. And they let me have a look at my atomic notes. So I’ve got an atomic note that says, using copilot GPT to automate the easy stuff is the opposite of doing code cutters. And that sort of single thread, that single idea struck out of some ideas I had whilst watching some videos or reading stuff or just like doing the dishes or whatever.

And I filtered it through and made this atomic note. Now that is not a blog post, but it also is something that I was happy enough to post on Mastodon perhaps a month or so ago. And so there’s something between a thought that you have had and a blog post and sometimes that’s

Robin: blog post.

Tristram: Exactly. So there’s this continuum and I think that you don’t actually need to. Post these things on Twitter or Mastodon, you could have them on your digital garden ready for people to browse through if you are happy with these atomic ideas. And they’re not too private. But if they’re well fleshed out and you’ve got some references, as usually happens when you’re making these robust atomic notes, they’re naturally branched.

They’re naturally linking to each other. They flow through and ideas string themselves together. And you get like a sequence of atomic notes that build upon each other. Let’s look at that atomic note. Using copilot to automate the easy stuff is the opposite. Getting code casters.

I’ve got a note before that says, AI can only reduce complexity, and those are linked threaded through. And so that one idea follows another, and they eventually, if I kept adding on thoughts to this string of ideas, this string of atomic notes, if I kept stringing more and more together, I might eventually have a video.

A video script or a blog post idea. this is literally how my last AI video happened. I just had these little ideas that came together and I developed them and I built them into atomic notes that I was pretty pleased with. They seemed to stand alone on their own, and I strung them together.

And eventually I had the outline of my script. AI is Not for You. Which is a few videos ago, and it feels like I could probably do this in public, in my digital garden. I could publish my atomic notes with links between them and people could browse around and find a little bit of meaning for themselves should they wish to.

Robin: Yeah, this is what I’ve always wanted to do, and some have never found the structure that supports it ish. Like I’m not very private and I don’t really want to be very private. And I believe in openness and I would love it if, I think also you don’t need to, general, you don’t really need to be scared of putting things out there because a lot of people who don’t publish, and this might be a thing that changes with the new TikTok generation or whatever, but like people who don’t publish, can be very scared of outputting anything at all. when you actually start to do it, I think you realize that there’s so much information out there that nobody cares. Like predominantly nobody cares. Like you have to be quite,

Tristram: Yeah.

Robin: Known or say saying something interesting in some way for anybody to care, right?

So for the most part it’s gonna just it’s inconsequential and so you don’t need to be scared of it. so I would love to be able to do exactly what you’re saying and throw up small thoughts. have a structure for them to be drawn together into larger thoughts.

Something that I think possibly I would distress me when you try and turn it into a website is like in obsidian, obviously I can create whatever mess I like because it’s only for me to come back to right

Tristram: Yes.

Robin: online. I feel a certain obligation to try and avoid link rot, right? If I had a fleeting thought that I throw up in someplace that you can find through looking through my website, and then someone finds it and they find it interesting, but then later I want to evolve it into something a bit bigger.

And so therefore I might turn that into this other thing. And that old thing might be gone now it’s a 4 0 4, right?

Tristram: I’ve got an answer for you,

Robin: would concern me.

Tristram: I’ve got an answer for you. If you’re using a robust [00:20:00] tool like obsidian, it renames links as you rename pages, and I admit that doesn’t solve deleting a page, you’ll then end up with a broken link there. But it’s not without a modicum of control.

As long as you don’t expect like a, I would imagine you’re not thinking of, oh, ha, having a 4 0 4 when I delete a page, that’s actually very normal.

You’re worried about having a 4 0 4,

Robin: Sure.

Tristram: right? You are worried about it like somehow links breaking when you move things around. Obsidian solves that by keeping all links working.

Robin: Does it, as in, it’s not on obsidian, like the, what did you call it? Is it called digital garden?

Tristram: It is just called the digital garden plugin.

Robin: it is the digital garden plugin that’s gonna have to take care of it because something has to be listening for requests and turn them into three oh twos or three oh ones, right?

Like

Obsidian

Tristram: yes.

Robin: that.

Tristram: What the plugin does is every time you publish it. Uses all of the links as they are currently in your obsidian vault. So if the links in your obsidian vault are still working, they will still work on the website.

So you

Robin: expecting links to break. What I’m saying is bookmarks will break

Tristram: Ah I see. So if I move something around, if I change its name, somebody will link to my site and it will no longer be there. Great.

Robin: yeah

Tristram: okay. so there are two ways to think about this. The first one is that there’s a method in just like in many static site generators, you can throw something into the front matter that says link slug or link URL or whatever.

And then you’ve got yourself a permanent link. So no matter where you change the do, that’s all fine. Problem solved.

Robin: Second option is that the very nature of the digital garden is that things are a little bit in flux, but because we are not separating this out into a thing that is all public and a thing that is all private, because it’s blended, we have got a much more granular way of doing things.

Tristram: If you go to namtao.com and the first option is no Boilerplates, and if you click read more, the ur l is namtao.com/nb. My digital gardenmy obsidian note for that is not called nb. It’s called something like No Boilerplate Home,

But I’ve got this perma link called nb, and then in the video section I’ve got some of my videos listed.

Each of those videos also has a PERMA link, because I think I’m very likely to link those from outside. But if you go back into something more relaxed, like lost Terminal, everything under, so Lost Terminal is nta com slash lt. That’s a perma link, but everything under it is dynamic because I.

Want the freedom to just move things around, change characters, names, change where people live, change the whole topography of it. But you’ll still be able to go to the lost terminal section of my garden because I’ve made a little, I dunno what the metaphor should be. A stone archway that says Abandon or hope.

No. The opposite. Get some hope.

So it’s read more under the lost terminals section on the front page. If you click read more at the bottom of that Green Hope punk sci-fi.

Robin: so it’s the read more and the fact that you’ve got an LT slash which you’re

Tristram: Yeah.

Robin: archway. And then you’re saying within this things move around, which means that if I click on LT season oh eight, or I’m trying to figure out what I could, okay, so you’ve got a note, you’ve got a keratin note section, right? So if I clicked on one of these, for example volunteer or whatever, and which is empty, for example. But let’s say you’ve written a few sentences in there, but then you decide to change her name. You are just saying there’s no promise that you’d necessarily find that it was here if you came back here.

Tristram: Am saying that there is a granular promise you can expect my garden to have a lost terminal section. In 10 years time, you can very reasonably visit my house, come into my garden and say, oh, I remember this lost terminal section. It’s the same place that I’ve, I remember it, but look at all these new flowers you’ve planted, and those, some of those old flowers, some of those old shrubs have moved.

Where’s that old oak tree? Oh, you cut it down.

Like I don’t like, I know that cool URLs don’t change, but there is a gotta be some granularity to that binary when it, when in the digital garden

Robin: to recti because, ’cause I love HTPI love the internet. I love the fundamental

Tristram: should.

Robin: Of the internet. And I suppose I feel like spiritually what you’re describing here is a bit the essence of what we think the internet should be like.

As in pure internet is the one where people are represented by their own domains rather than through

Some, conglomerate. In those domains they have, freedom of expression. You get to come along and experience the things that they wanna put out there. Freely. Because I feel like. The core standards of the internet and this idea of your own digital garden of self-expression are so overlapping. Like that’s what the internet’s there to enable. trying to therefore [00:25:00] reconcile some of the standards, with this vision, but I think you’re right that it’s probably just that I’m being too precious about broken links.

There are ones that you can rely on and

Are ones you can’t, and obviously you can make your 4 0 4 page helpful, right? So yeah. So I think it’s a point that we don’t need to focus on anymore, but I just, I, that’s where my mind went. But

Tristram: one of the things that you might be able to do is advance the signposting of each ’cause currently I’m breaking a rule of digital gardening that I’ve not signposted which of these pages are in progress and which of these progress are more like our buds or seedlings or trees or, I forget exactly.

So you could say when it is at the highest level of confidence, it gets a permalink.

Robin: yeah. Like a tree with a, it is a good metaphor because a.

Tristram: Yeah.

Robin: A tree, you don’t expect to disappear.

Tristram: Yes.

Robin: down a

Sacrilege, right?

Tristram: Yes. The tree is gonna survive a hurricane, potentially.

Robin: Might not that,

Tristram: Maybe. Yeah. A storm. Certainly. Drought, perhaps. Yeah, exactly. I do. You’d like it?

Robin: Yeah.

Tristram: I do.

Robin: I do and I will be playing with it also not very happy with the fact that my blog is a completely, inaccessible, chronological list. And I think it should be much more thematic and it doesn’t need to be neat. Like it could absolutely just be, I want you to be able to meander through it by topic, and I just haven’t made it like that, because I, in a way, I’m a bit too vain like I was really proud of the styling of my blog because it’s supposed to be elegant and usable straightforward and it’s, yeah, like the text is big enough.

it doesn’t use more colors than it needs. But it’s, trying to be elegant and make it as easy to read the content and access content as possible. in that mission it’s a bit like also the thing that blocks me from throwing content up because I’m being too perfectionist about it,

that’s why it’s not messy in the way that I think actually would be a much better representation me. So I will be, I hope I will find the time quite soon to recraft my blog as a digital garden because I love the idea.

Tristram: I love the idea is good.

Robin: keeping it from me for six months.

Tristram: I’m so sorry. It’s pretty cool. like every morning I do two hours of pottering before I get into my lost terminal writing.

Robin: it’s just.

Tristram: I just, and I literally very often will wake up before my alarm. Really excited to go do that because

I’ve blended my ze casting knowledge base, my project management and my digital garden all into one thing. And I like it’s like I’m gonna be working on all of this at the same time, without any real goal, but just knowing that I’m increasing the progress on lots of little things that are important to my life in this system.

It’s just the best thing ever.

Robin: That’s amazing. If you succeed in turning me into a sort of constant gardener of my website, I feel a bit like you’ll have saved my life. Like, I feel like I’m I’m I just, I don’t have the, I think I don’t have the space, the freedom to just in that way and everything is yeah. I would love that. if I could just, go and spend half an hour wandering through my web garden and pruning it and helping a bit grow here every day. That would be amazing.

Tristram: There we go. we’ll see how much of that we’re comfortable keeping in, but that was lovely. Let’s press on.

Okay.

me about Tess Riel.

Robin: Qui, yes, it’s a little bit of an ungainly title for a topic. So do you know fundamentally who Tim, Nick Gabriel is?

Tristram: No.

Robin: You may do, when I explain who she is timber Gabriel was a she worked for Google

And she was in the AI safety I think, and she published a paper, which is this stochastic parrots paper. heard of the term stochastic parrots? It’s a description of LLM generative AI systems.

Tristram: Oh, of course it is. Yeah, that makes sense.

so she, the paper basically says that AI is harmfulSure.

Robin: making some points about safety and things that Google should focus on.

And she was fired. And she very cleverly made use of fired by Google in her personal branding, I think.

Tristram: That’ll do it.

Robin: but she set up a research company after that. And then she worked with this philosopher called Emily Lee, or Emil, I dunno how to pronounce it p Torres. And they came up with this term, tere. I think why this is interesting is because you have to ask yourself, given that AI has had many different meanings, but at the moment is these generative sort of lamb based, algorithms.

Tristram: Huh.

Robin: if you understand, how incapable they are of solving many of the problems that people claim that they can solve, you have to ask yourself the question, how come so many incredibly wealthy investors have [00:30:00] put so much money into developing these AI systems? What is it that leads to that happening? Like was this an obsession of Silicon Valley? Why did people, represent such a significant fantasy that then diverted, and trillions of dollars and completely transformed the American economy through doing that based on the production of systems which were not gonna get anywhere near delivering on their stated

Tristram: right.

Robin: And so to answer that question, they’ve gone explored the ideology of the people who might invest. of the AI believers, the AI boosters and the, and particularly the AI investors, and so I suppose this is the reveal. TES real stands for transhumanism Exrop, pianism singularitarian, cosmism rationalism, rationalist ideology. Effective altruism and long-termism. And they draw out through this paper, which I would recommend if anybody is actually interested enough to read it. It is an easy to read paper. It’s very well written, and it’s not long. But it lays out the links between all of these overlapping ideologies and how they lead together. How significant advocacy or belief in these ideologies the line, the through line that links all of the different, aI boosters who have pushed the idea that we are heading for an AI apocalypse or that there’s the AI is this powerful thing that’s gonna take over the world that we really need to or that it’s gonna transform or it’s gonna be the thing that’s inevitably gonna transform humanity and save all the lives.

Like all these people who are pushing all of those fantasies that really cannot flow from, if you understand what in fact an LLM based generative algorithm is, right?

Tristram: Yeah. It’s also correct. Trained on the internet.

Robin: those, this bucket of ideologies is what pushes it. And you can trace it all the way through.

So the word that draws all these ideologies together, you could say as eugenics, because it’s basically an idea that you can craft yourself through whatever mechanisms, through manipulating things. You’re trying to build yourself a sort of perfect model human or a perfect model being that’s able to like, transcend humanity. and it’s similar to what eugenicists, would’ve been doing with the Arian race in Nazi Germany. There was a lot of long history of eugenics thought that. Eugenics was. The idea that there was racial traits in racial superiority was held as a scientific truth for a surprising amount of 20th century, right?

Not just focused around the 19 19th, thirties I’d say extending significantly beyondand it sort of infected the way that people thought about lots of different things. But this is the thread that ties all together.

So I just firstly, I think it’s really interesting to ask that question like, what happened in humanity that led to so much investment of such a significant sum of all of the product of humanity being invested in this complete fantasy, right?

Tristram: Yeah.

Robin: that’s a really interesting question to ask.

And then I think the answer is really interesting that these ideas really overlap and they’re all linked and they’re linked through this belief that you can transcend all of the flaws of humanity in this like creation. And they just believe in that so strongly that they will look for evidence and they will find every opportunity to reinforce it without asking too many questions. Do you know what I mean?

Tristram: Right.

Robin: think it’s really amazing.

Tristram: That I admit, I don’t know very much about this entire space. Other than the small bits about transhumanism and

what were the other, what were the other bits that, that

Robin: altruism, long-termism?

Tristram: yes. Like it’s always easier to imagine helping a future person than actually give money to a homeless person today. I don’t think of them particularly highly

Robin: yes.

Tristram: of altruism people quite apart from that crypto, bro.

Robin: journey of effective altruism, I think is fascinating because and it’s significantly based around Sam, Bankman Frees, that’s name, I think, right?

Tristram: Yes. Not the best poster child for it. Yeah.

Robin: But like effect altruism in its founding, I think was at least interesting.

So there’s, I can’t remember his full name, but there’s a guy called Peter, someone who’s like a philosopher. I think he was probably Oxford or something like that. and he’s sort of the founder or one of the fathers of this movement. And he’s basically saying, if you really try and quantify your altruism. Then you’ll see that you might think that you could help a homeless person in your city, but if you live in the west and you send that money to Africa, you could save way more lives. That’s quite an interesting basis. So that’s where you start.

If we were to follow that through and we were to calculate the different ways in which we could actually increase the way that we could help people, then that would be better because we could work out how to help more people. The thing is that then led, like [00:35:00] where that led is insane.

So the first insane place that people start to say, this is what Sam b for free did I think I’m gonna get a job in finance so I can make more money and then I can give it away.

Tristram: Yes.

Robin: and now there’s and that then obviously led to basically the most egregious and largest version.

I, I, what’s the right term? I don’t know if money laundering is the right term, but it’s the

Tristram: Raw was it

Robin: It is fraud, but it’s like this the inventing of money, the passing around between companies that are supposed to be independent. They’re like, yeah, just like writing fake numbers into books.

It’s exactly all the same stuff that like Enron was doing

Tristram: is. Is that securities fraud?

Robin: Probably, yeah.

Tristram: Is that what that is?

Robin: would, would’ve been doing securities fraud. But like all of this stuff was justified

Mind, I think because he’s I am, I’m a good place to be given a lot of money because I’m gonna be giving it away.

And I think he did give a lot of money away to some good causes, but would be arguable whether they were better causes than like other people with the money you might have given it to. So I don’t know, but like that, that the first thing is this idea that you can therefore get like one of the most unethical jobs just because you’re making a lot of money and then you give it away as like a way to do good, which seems pretty suspect to me in the first place. then obviously the other place EA goes, which is just absolutely nuts, is the idea of future, as you say, future lives versus present lives. Like we are going to imagine some kind of future where there’s a lot more humans and those ones would be hurt by, this thing. And they don’t turn it to climate change.

This is the really mind boggling thing. So

Tristram: oh no,

Robin: they don’t use it as an argument for climate change. They use it as an argument for why they have to explore into planetary travel or again, ai, right? We need ai ’cause it’ll save us from climate change. I guess there might be a bit of a climate change argument there, but you do it through ai.

Like the only way to solve climate change is to invest in ai, I think. Have you heard that argument?

Tristram: I have not. I would not believe it if somebody made it to me.

Robin: exists. This is definitely an

Tristram: Ugh.

Robin: people make.

Tristram: It does seem it’s the perfect ethos for an investor for whom every tomorrow is worth the unethical practices of today. Like it lines up perfectly with being somebody who just wants to see the number go up. And if someone’s telling you that, oh, that’s ethical. Marvelous. What a dream,

Robin: Yeah.

Tristram: I’m so happy for them.

Robin: Yeah, exactly. I think that’s exactly right. I think what’s interesting about this paper as well, is that she’s saying like on the left, where we recognize that capitalism is destructive, right?

Easy, therefore to say everything is the fault of capitalism. So all these people who are doing toxic things, they’re doing it because they’re driven by a pro profit motive. Now, I don’t think that’s true because I think that once you get beyond a certain level of wealth, the wealth no longer, may matter to you in some small time way.

Like you might want to keep your status as a trillionaire, right? So that you don’t lose face against your. Buddy you are competing with or whatever, right?

Tristram: Right.

Robin: broadly speaking, that’s not your big driver. big driver is something else. It’s something ideological, right? It’s not, I’m just doing everything because it makes me more money. I think you will inevitably be trying to make more money because you employ people who make sure that’s gonna be the case and whatever. They make smart investments. So what you are saying about being able to make an argument through effective altruism about investment, I think is right?

Like about it being good investment is right. But I think the thing about test reel is it explains the really impactful thing is are the people that have all the money, that have all the power and their actual driver is something else, right? So what is their actual driver? these are to draw that out. Do you see what I mean?

Tristram: It does. It sounds like I should read the paper to find that out. it does, it draws a thread between all of these different kinds people and explains what links them all and what motivates all of these different groups?

Is that, did I get that right?

Robin: Yeah. I mean it’s basically saying they’re overlapping. It’s saying that there’s actually a bit of a journey between them. But if you look at some of the people you might draw out, so you’ve got obviously SBF, you’ve got Sam Altman, you’ve got Hinton, Jeffrey Hinton.

You’ve got so all these AI boosters, if you inspect their ideologies, you find that they include some subset of this bucket and they overlap quite significantly. Do you see what I mean? and they evidence this, I think they, put it quite clearly to show that these are, defining beliefs of the set of people that are responsible for this direction in technology investment.

Tristram: Interesting.

Robin: Yeah, that’s the topic. I find it fascinating. I bring it to you, I suppose a little bit because I feel like broadly speaking, it’s my kind of thing and not your kind of thing. But I wanna see how interested you are in it. ’cause I love, sociology and philosophy and ideas and ideologies.

And also I guess bad people, understanding the things that harm us or lead us astray. Yeah.

Tristram: It’s, I have been very interested in, thank you very much for explaining this to me and I hope some of our listeners will find it useful as well. the [00:40:00] thing that. I am left with is that it’s wild. They didn’t want to solve climate change. Like the is such a low hanging fruit.

Robin: Yeah. as an overall statement that might not be fair. It’s just that there might be people there that really care about it, but SBF probably cared about it. It is just the way that he would want to go about it as suspect, maybe, I don’t know.

I dunno enough really just to comment.

Tristram: You certainly, don’t need to invest in space travel

Robin: yeah, certainly.

Tristram: Like you, the there is a very clear order of things to fix and space travel is not on that list.

Robin: Yeah.

Tristram: It’s like that is not to say that weather satellites are not very useful in combating climate change, but like space tourism probably not

Robin: certainly. No, that’s, but it’s the idea of planetary escape, isn’t it?

Tristram: right?

Robin: that one version, I suppose it’s if you try and map this out, and this is what these clever people do, and they clever, clever brains, right? Is you are like, look. Let’s say there’s a 30% chance that we fail to save the planet, and climate change destroys it and makes it completely uninhabitable, right? it’s an ethical good to save the human race, so therefore we should be investing some amount of money in the ability to move to Mars. Do you see what I mean?

Tristram: Yeah. It’s a preposterous argument. Mars is not very habitable, like even the worst ravages. I’m I think I, we’re now in my wheelhouse. ’cause we’re talking about a post environmental apocalypse, which is all I write about in Lost Terminal. I do a huge amount of research to try and make it right.

Even like the most grim, awful predictions don’t involve the planet. Not having any oceans. Mars doesn’t have any oceans. even if there is a catastrophe, like the, it’s actually the opposite problem. The ocean is gonna bring the party to us. It’s not that the ocean’s gonna go away, it’s gonna have too much ocean.

Robin: yeah.

Tristram: Like in Lost terminal, in my story, 10 meters sea level rise is what I chose in 2020. That was a wildly optimistic thing. In hindsight actually, we’re like, that would be, if every country immediately changed what they were doing and focused on climate change, we might only have 10 meters of sea level to rise.

I apologize in advance. For

Robin: a fact to drop, isn’t it?

Tristram: I’m sorry. I, especially, I’m sorry for our Dutch listenersWhen I think Amsterdam is three meters below sea level it’s gonna be a bad time for Hitachi friends. Which is especially sad because I believe that the Netherlands on the whole are amazing at investing ingood climate policies and so forth.

Certainly they’ve got the first iron powered brewery. So it’s like carbon neutral because they’re burning iron. I cannot go off topic about how much I love the Netherlands. Yeah, it’s cool.

Robin: I think it brings a little bit of wholesomeness to my topic

Tristram: We should just like solve climate change. It’s much

Robin: It

Tristram: like.

Robin: saying is yes, obviously we need to solve climate change if we’re gonna stay here, but also the best solving of climate change still means we’re going to have massive challenges in terms of environmental changes. And so therefore we have to solve climate change and work out how to deal with the changing environment.

Tristram: Having a

Robin: right?

Tristram: Yes, having Planet two is a great idea. That doesn’t remove Planet one, right? Like Planet one will still exist in this hypothetical utopia where we’ve become a multi-plan species. Like the solution is not, ignore the first planet with, conservatively 10 billion people on.

That doesn’t sound good, but it does sound good to the kind of person who imagines themselves going to planet

Robin: yeah. But it’s not just that they, the thing that I think is commonthe through line is that it’s the fantasy of it. Like I’ve always been struck

That the people who claim to be Our smartest, leaders, right? Have such a huge fantasy element to them. You know what I mean?

Tristram: Yeah.

Robin: all sorts of nonsense and then trying to make the world make that true for them, which is why I think the ideology thing is such a good point. It’s like ideologies are so defining.

I found this in the way that like Elon Musk complains about population decline in the US but none of his solutions make any sense, right? it’s so hard to figure out why the hell these people who presumably have an education and claim to be really smart can continually seem to be making such obvious, logical errors.

You know what you’re talking about it’s there’s Musk, there’s loads of stuff that loads of people think he’s really smart and they think he’s a genius until you are actually an engineer. It’s that thing about, if you read an article in a publication in your area you realize how inaccurate it is, and then everything else you hear, you think it’s accurate still, right?

Tristram: Yeah, so that’s the gelman amnesia effect.

Robin: Exactly. And that’s what baffles me about

Tristram: is it really baffling or is it that people will latch onto an ideology that already confirms what they already know to be true? If you’ve got a religion that says that. You are, you’re okay. You are born into the right thing. You are the right people. You’ve got the right [00:45:00] tribe.

People who are outside your tribe are wrong. That’s extremely persuasive. That’s extremely easy to be loved into. And if you’ve got something like effective altruism, which confirms your existing belief that you didn’t pay anyone anything, you should try to squeeze everyone around you for all the money, and you save it all.

That’s perfectly aligned because you will perhaps notionally save the money for some hypothetical future where you’ll be able to save a billion, billion people. And therefore, what you’re doing is ethical. That’s extremely convenient that it aligned with the values you already had. You didn’t have to change your mind about anything.

Robin: Yeah. But the thing about I, I don’t think following the money gets you there is because the fundamental thing about AI is that so many people who have invested so much hopeless dead ends, right? Like that’s the thing about it is like Elon Musk said, the only possible way to justify Tesla’s share price if. Become self-driving cars, right? that would explode the company’s value. And the share price would be justified. the only reason why people pay that amount in Elon Musk’s, whatever is because they believe that might happen or that will happen. And it’s not going to

Tristram: no, I, this is is this a post hoc argument? It’s just like circular argument. It’s rubbish.

Let, we must close this topic down and I will finish by looping us back to the idea of planet two space explosion and so forth

I forget who said this, everybody should go to space.

Not everyone should return,

Robin: We’ll go from a conversation about terrible things to do with ai into good ai, I believe is what you’re gonna tell us about, isn’t it? Tris?

Tristram: Yes. And I do hope the listeners stick with me for this. Don’t immediately say off. It should be very obvious by now that I’m an extreme AI skeptic. I made a t-shirt that said, don’t talk to me about ai. Like it should be very obvious. However, there are plenty of good uses for.

Large language models, and they’re all to do with very specific language features and so forth. Here is one that was impressed upon me by Amos. I’m sure they won’t mind me saying this, Amos from Faster Than Limea really good YouTuber and much better than I article writer. And now my collaborator ammos did a little bit of tweaking the website whilst I could see their shared screen.

And what Ammos was doing was using the Z editor combined with copilot like you can do, and did a little bit of what we would call derogatory vibe coding on the website. These were like some low level tweaks, like change this name everywhere, like small refactorings, like almost doesn’t need AI to code.

But this was like just we were doing something else and this was a really low priority thing. So fine, not my cup of tea, but it was okay. Here is where it astounded me. Ammos was using voice typing to instruct copilot inside of Zed. What to do. And this made me realize that actually there is a good use here.

Robin: Can I

Tristram: be

Robin: can I jump in for a second and

Tristram: please

Robin: a point of clarification. I know what copilot is.

Tristram: So Zed is a relatively modern new programming editor. It’s got the two main features that come to the front of my mind when I think about Z is one.

Robin: I,

Tristram: written in HTML and electron like VS. Code is, it’s written in native rust, which makes it extremely fast.

Like you think vs. Code is nippy enough, but that’s because we’re used to it. The speed is a feature.

Robin: okay.

Tristram: the second one is that it’s AI first. they’ve got a big AI plan and it’s why I didn’t really take it seriously because as I’ve said, I’m an AI skeptic, but the ability for you to say things like change all of these functions so that they’re wrapped in this other module and then go, if I was typing that it’s less compelling ’cause I’m having to type into an AI coding chat box.

But if I’m speaking that. Suddenly we’re doing something that I couldn’t do. And what that thing is, program using my voice, and this is something that recently has become all too important for me as I get RSI which is that ligament problem you have where typing hurts it’s like carpal tunnel that sort of thing.

’cause last year or maybe two years ago, I got RSI in my fingers, hand, wrist, forearm, elbow, both sides, because I was doing too much, perhaps pottering in my digital garden. I’m joking. That wasn’t the problem. The problem was that I was working for 18 hours all day, every day because I was self-employed.

And you can just blur the lines between business and pleasure. And I really just loved what I was doing. So it seemed like pleasure until six months in my body was like, Hey bro, we are not doing this anymore. So I’m doing much better because I have been lifting weights. That’s my tip for anyone listening [00:50:00] is just go to the gym, lift some weights, buy some weights, lift them.

That’s it. that’s all you need to do. At least it worked for me.

Robin: the

Tristram: I’m not a doctor,

Robin: this is

Tristram: blah, blah.

Robin: tangent, but is the RSI in the wrist or where is it?

Tristram: Fingers, hand, wrist, forearm, elbow, bilateral, catastrophic, both sides. Whole arm. Like I’ve got tennis, elbow and wrist and finger problem. Like it was, it happened all at once. Make ’cause my body was like, Hey, you’re not listening to us, so we’re gonna set everything on fire.

Please stop working so hard.

Robin: Wow.

Tristram: yeah, real nightmare. And I’ll be real with you. it is not an overstatement or an exaggeration to say that my A DHD meds gave me RSI. Now that is not, nobody should take away that you shouldn’t take a DHD meds or that they are dangerous.

But the line is very clear. A DHD meds allow you to work and focus on what you would like a little easier. That’s wonderful. But what I wanted to do was sit in front of my computer 18 hours a day and type and that is a real problem. Do be careful everyone. Don’t be stupid like your Uncle Tris.

Robin: I feel like I could divert into a topic psychology that I’m very interested in, but I won’t do it.

Tristram: That’s very generous of you.

Robin: time.

Tristram: Write it in our notes right now. Robin. Type it in, put ’em in for the next time. So I will turn to Zed and copilot and voice typing. here I had something that I could continue programming and hell probably anything, any kind of text I wasn’t using.

It’s a tool that you don’t need to use the generative part of it, you can be extremely prescriptive, but it’s a human computer interface. We already had solved voice recognition like in the nineties. Voice recognition had got to the point where it was great and we’ve now solved the comprehension for simple things.

You can’t with copilot. Get it to do complex work, complex refactoring, once your program gets to a certain level of complexity, I have been extremely unimpressed with what I’ve seen, but I’m not talking about that. Those are the problems. If things break down at high levels of complexity, but I’m not trying to replace that.

I’m just trying to replace typing for the days when I can’t type. Now those are few and far between these days, but being really prescriptive and saying things like those three paragraphs, move them down five lines comment them all out this would be too low level if you were to do this like day to day.

This is not what people want with vibe coding. They want make me a website that does x really high level maximum result and that doesn’t work very well, at least after the first prototype. Once you start refining nightmare because you are getting more specific and less general. And LLMs love being general and hate being specific because they’ve not been trained on specifics.

Or rather they have been trained on specifics but they’ve averaged it out and ground it down into this perfect sphere of unimpressive bullshit. But if you are being very specific and it’s like having that intern who doesn’t really know anything about what they’re doing, but they’re very enthusiastic and they will try to help you wherever they can for somebody who can’t type.

That is genius.

Discussion