Dreaming Against the Machine - June 23, 2026


Episode 11： Reverse Centaurs and AI, with Cory Doctorow

Topics
reverse centaurs and ai AI bubble danger radiology AI paradox sectoral bargaining
Episode 10： Better Off Regulated, with Ed Zitron Episode 12： The Cosmic Perspective, with Katie Mack

Episode Stats


Length

1 hour and 1 minute

Words per minute

181.29

Word count

11,177

Sentence count

572

Harmful content

Toxicity

23

sentences flagged

Hate speech

10

sentences flagged


Transcript

Transcript generated with Whisper (turbo).
Toxicity classifications generated with s-nlp/roberta_toxicity_classifier .
Hate speech classifications generated with facebook/roberta-hate-speech-dynabench-r4-target .
Topics generated with Qwen2.5-3B-Instruct.
00:00:00.000 Welcome back to Dreaming Against the Machine. I'm your host, Adam Becker. This week's guest
00:00:07.100 is Cory Doctorow, and I'm very excited to have him on the show. Cory is a science fiction author,
00:00:15.760 a nonfiction author, an essayist, blogger, commentator on the tech industry for decades.
00:00:22.500 I've been following his writing since I was in college, back when he was writing for the famous
00:00:28.480 blog boing boing and then read a few of his novels and from there i was hooked and over the last year
00:00:33.920 i have gotten the chance to meet him and be on a couple panels with him and i knew that he would be
00:00:39.620 a perfect person to have conversations with on this podcast about what a better future for the
00:00:46.160 tech industry and for all of us might look like so without any further chatter for me here is cory
00:00:54.380 doctorow
00:00:54.960 cory welcome to dreaming against the machine thanks adam it's a pleasure to be on how do i
00:01:03.420 even introduce you you are a science fiction author and uh activist for electronic rights
00:01:12.500 and essayist uh non-fiction author long-time blogger and general internet person uh at least
00:01:22.040 That's how I think of you.
00:01:23.320 That suits me fine.
00:01:24.680 Sure.
00:01:25.560 Yeah.
00:01:25.960 Yeah.
00:01:26.440 Why not?
00:01:27.940 Raconteur, gentleman adventurer, possible war criminal.
00:01:32.400 I don't know.
00:01:33.060 Yeah.
00:01:34.120 Blogs from a hot air balloon.
00:01:36.020 Right.
00:01:36.680 In a goggles and cape.
00:01:37.740 And has been what associated with the Electronic Frontier Foundation for how long now?
00:01:43.440 I'm just about to start my 25th year.
00:01:45.180 Wow.
00:01:45.860 Were you there when it started?
00:01:47.380 No, it's 35 years old.
00:01:49.560 Okay.
00:01:49.760 So I kind of grew up on it.
00:01:52.040 Right. My mentor was a science fiction writer named Bruce Sterling.
00:01:56.000 Oh, yeah.
00:01:56.420 And Bruce wrote a book about the founding of EFF called The Hacker Crackdown.
00:02:02.460 Basically, the year EFF got started. And I read that book and spent the next 10 years following EFF's work and being really excited about it. And 10 years later, found myself working for them.
00:02:11.880 Yeah, no, Bruce Sterling, he co-wrote that book with William Gibson, The Difference Engine, didn't he?
00:02:16.660 Yeah, The Difference Engine. That's right. Bruce is my daughter's godfather.
00:02:20.260 Oh, wow. Okay.
00:02:21.620 remarkable fella yeah well i've been following your work for a very long time i was reading uh
00:02:27.360 boing boing uh back in college and i knew when i started this podcast that you're one of the people
00:02:32.900 i wanted to bring on because if i'm trying to think about what a better future looks like
00:02:36.980 especially when i'm looking at the tech industry uh you're one of the people who has been thinking
00:02:43.240 about that in public longer and more carefully than anybody else i can think of and you have a
00:02:50.700 new book that is coming out the day this episode of the podcast drops, The Reverse Centaur's Guide
00:02:56.460 to Life After AI, How to Think About Artificial Intelligence Before It's Too Late. Well, why don't
00:03:01.960 we start with that? And I imagine the conversation will go many different places from there. So why
00:03:11.700 don't we start with what a reverse centaur is? It's a commonplace in automation theory and in
00:03:18.240 history of automation that when workers drive automation they tend to use automated tools to
00:03:23.400 improve the quality of their outputs and when uh capital chooses the way that automation is used
00:03:29.900 it's it's almost always to improve the quantity or the throughput and it's not because of like a
00:03:36.260 callous disregard for workers although there's some of that there it's it's because there's a
00:03:41.260 very material matter um if you acquire an asset and then you are depreciating it off your books
00:03:48.220 and you want to maximize the return before it depreciates to zero, you want to use it as much
00:03:52.640 as possible. And generally with machines, a human being is the bottleneck. The machine can usually
00:04:01.540 work longer and usually work faster than a person. But if the machine requires a person to do part of
00:04:07.660 the job, then the machine is going to make demands on the person that will go all the way up to the
00:04:14.440 entire capacity of that person how hard they can work and how long they can work and still demand
00:04:19.920 more if it's available to be given so in automation theory a centaur is a person assisted by a machine
00:04:26.340 which is to say it's usually a worker deciding how they're going to use automation right so it's you
00:04:30.080 on a bicycle uh you with a spell checker you with a debugger right and when i started writing
00:04:36.620 computer programs i didn't have a debugger i just would like literally have to write breaks into my
00:04:41.620 basic code on my apple 2 plus to figure out where my code was going wrong yep so those are all ways
00:04:47.360 in which a machine can be harnessed by a human and in which we can be configured like a centaur
00:04:53.480 which is to say a human brain directing a tireless body that can do things that exceed both the
00:05:00.380 endurance and the capacity of humans right that can work faster and harder than a human but is
00:05:05.680 directed by a human. And a reverse centaur is a human who is harnessed to a machine as a peripheral
00:05:11.760 for it, right? And so, again, this is not new. Like, there's a reason that when we look at
00:05:18.360 cinematic depictions of machines and humans in the class struggle, we get things like Charlie
00:05:23.200 Chaplin in modern times, or Lucy and Ethel trying to get chocolates into the chocolate boxes at the
00:05:28.300 end of the conveyor belt, and the conveyor belt keeps going faster and the chocolates are flying
00:05:32.040 everywhere because that's the foundational configuration, right? You're going to be worked
00:05:36.000 as hard as the machine can work you because you are there to do the part of the task the machine
00:05:42.640 can't do. The machine isn't there to help you do something that you're good at. And so in these
00:05:48.420 reverse centaur configurations where the machine is riding a frail human body, people aren't just
00:05:54.280 used, they're used up. So the most automated warehouses in America are Amazon's warehouses,
00:05:59.580 And they're also the most injury-prone warehouses in America, where workers are injured at about three times the national rate.
00:06:05.960 Those aren't coincidences.
00:06:08.100 There's a causal relationship.
00:06:10.080 You put seven figures into warehouse automation, you want to work those machines as hard as you can, which means that the people have got to be worked at the edge of their endurance and capability, which means that eventually they're going to miss a stitch and they're going to get squashed by a machine.
00:06:24.460 How do we push back against this?
00:06:26.980 and how, I mean, you're actually, you're reminding me of something that a previous
00:06:31.420 guest, Susanna Glickman, said about AI. She said, you know, you hear people saying things like,
00:06:37.320 oh, AI is making me, you know, do my job faster. And like, no, that's not true. My boss is making
00:06:44.480 me do my job faster. The AI is a tool and you could use it in different ways. So how do we
00:06:51.000 push back against this larger trend of reverse centaurs and how is ai playing into it i don't
00:06:58.660 think ai is an extraordinary technology i think ai is a normal technology which is to say there's
00:07:03.940 good ways of using it and bad ways of using it and if your boss tells you to use it they're probably
00:07:08.540 making you miserable and if you get to decide how to use it you'll probably make yourself happy
00:07:12.280 though not always because people sometimes make bad decisions that's like that's just every normal
00:07:16.600 technology works like that. But there is something extraordinary about AI, which is the size of the
00:07:22.800 financial bubble that it's harnessed to. And that is the most destructive thing about AI. It is the
00:07:28.320 thing that is upstream from all the destructive stuff that happens with AI, which is like the
00:07:32.720 suppression of wages, the immiseration of workers, the climate damage, the use of AI to automate
00:07:39.340 harm to people, whether that's targeting in Gaza and a genocide or Doge deciding who to fire or
00:07:47.760 the problems we have with customer service being outsourced to AI. All of those are downstream of
00:07:55.060 the fact that there's this massive financial bubble. And the size of the bubble can't be
00:08:00.280 overstated. When I wrote the book, the global expenditure on AI was about $700 billion. A year
00:08:06.760 later, it's 1.4 trillion and it's on track to double again. And that's a much bigger bubble
00:08:12.080 than, you know, South Seas. It's a bigger bubble than tulip bulbs, certainly a bigger bubble than
00:08:16.660 like Web3 or crypto or metaverse. You know, Mark Zuckerberg wasted $61 billion of his investors'
00:08:23.440 money on metaverse. He's already spent $150 billion in three years on AI, and it's another
00:08:27.980 $150 billion this year. And he says it's going to be more next year. And so where does this
00:08:33.560 extraordinary bubble come from and how do we stop it? That's the question I think we need to ask
00:08:37.820 because bubbles are very bad, partly because of the harms we've just articulated, climate, labor,
00:08:43.820 and so on and so on. But partly because the nature of a bubble is to have insiders tell lies about an
00:08:52.320 economic opportunity in order to steal money from ordinary investors. And in our world, the ordinary
00:08:58.140 investor is a worker who no longer has a defined benefits pension and has a market pension.
00:09:03.560 Which is to say that if you don't want to starve to death when you're old, you need to find something to invest your money in.
00:09:09.040 And these guys are tricking you into investing your money into something that they're just going to take and you're the bag holder for them.
00:09:15.380 So bubbles are really bad and how do we get rid of them?
00:09:17.660 Well, to understand that and to understand how to get rid of it, you have to understand the material and ideological basis of this bubble, which I'm happy to unpack.
00:09:26.980 I know I've just said a lot of things, so I want to get a chance to get a word in edgewise there if you'd like.
00:09:31.320 You and I were on a panel in L.A., what, I guess that was maybe two months ago, and you said some of this stuff there, and that was, I think I asked you to be on this podcast right after that because I thought, you know, I knew I wanted you on here, but I wanted you to say a bunch of that stuff on here.
00:09:48.260 Look, so the material basis for all of the tech bubbles of this century is that firms that have saturated their markets need to find a way to convince investors that they can still grow.
00:10:01.980 Yeah. And that is not because, per the old leftist saw, eternal growth is the ideology of a cancer cell. Wanting to continue to grow forever is, in fact, like a material thing that people want, not an ideological thing that people want, because firms that are growing have higher share prices relative to their earnings.
00:10:20.940 It's called the price-to-earnings ratio, and that's because a share in a company is a claim on its future earnings, and if your company is growing, its future earnings will be larger next year and larger the year after that, whereas if your company is mature and has steady state, then its earnings are going to be the same year on year or maybe move around a little in a noisy way but not see these big spikes or continuous growth.
00:10:42.780 So, of course, that share is worth less. And the reason that is materially important is, first of all, because the executives in the company who decide what the company is going to do are compensated in shares, which means that if the share price experiences a sudden decline because the market has revalued the company as a mature firm instead of a growing firm, they stand to lose a giant amount of money personally.
00:11:05.720 Yep. Right. So that in that case, what we're saying is that without growth, there's massive contraction. And also the way that firms love to grow in this century where we don't enforce antitrust laws is by buying other companies. And the cool thing about having a growth stock is your shares become highly liquid and you can buy other companies by giving them shares.
00:11:25.640 And the great thing about shares is they're a substance that is endogenous to the firm and can be produced on the premises by typing zeros into spreadsheets, unlike money, which is exogenous and can only be acquired by convincing a creditor or a customer or an investor to give you some money.
00:11:44.500 And if you try and make the money on the premises, the treasury department takes you away in handcuffs.
00:11:48.620 And so always better to be a company that has got a growth stock because then you can grow by buying other companies with your growth stock, which makes your stock keep growing.
00:11:59.920 So it's a self-licking ice cream cone.
00:12:02.420 And so firms really, really want to have a growth story.
00:12:06.800 And through this century, the growth stories started as fairly reasonable, but low probability.
00:12:14.700 So Facebook was about to become Google by doing the pivot to video and taking YouTube over.
00:12:20.540 Google was about to become Facebook by doing Google+.
00:12:23.260 Those are reasonable propositions.
00:12:26.420 Like, Facebook is a business.
00:12:27.920 We know because they have a balance sheet.
00:12:30.220 Whether Facebook will surrender that business to Google is, like, low probability.
00:12:34.680 But at least it's reasonable.
00:12:36.980 At a certain point, they got tired of having arguments about whether Facebook will become Google or Google become Facebook.
00:12:41.960 And they're like, no, we're going to become a thing that doesn't exist.
00:12:44.220 And, of course, that's less reasonable, but at least no one can credibly argue any more credibly than you can argue that that's a small opportunity versus a large opportunity because you've just made it up.
00:12:54.800 How big is the metaverse?
00:12:56.580 A quintillion dollars because I made it up.
00:12:59.520 How do you know the metaverse isn't worth a quintillion dollars?
00:13:02.100 I'm the world's leading expert on the metaverse.
00:13:03.760 I made it up, right?
00:13:06.020 I made up DAOs.
00:13:07.480 I made up Web3.
00:13:08.780 I made up cryptocurrency.
00:13:10.140 And the market opportunity is infinity.
00:13:12.420 and these guys don't need it to work although i think they'd like it to work all they need is for
00:13:18.300 there to be another one of these when the market loses confidence in this one they need to be able
00:13:23.200 to sort of run across the river on the backs of alligators without losing a leg and then we come
00:13:27.700 to the to the ai bubble and the ai bubble is like those other bubbles in that it's a growth story
00:13:32.660 about firms that have no room to grow but it's also much much bigger and so this raises the
00:13:39.800 question like, what is materially and ideologically different about AI? Materially, it's different
00:13:44.080 because it is realer than those other bubbles. AI is, from a computer science perspective,
00:13:49.760 interesting, right? Breakthroughs are always cool, and computer science gets those all the time,
00:13:55.560 right? They come up with a, they apply a technique in a new way and something new happens, and that's
00:13:59.960 what happened with AI 10 years ago. What is different about the AI breakthrough relative
00:14:03.940 to most of the CS breakthroughs that we experience is that it's scaled. So usually, you have a
00:14:09.720 thing and you do it slightly differently and you get a bigger output than you expected and you're
00:14:13.940 like that's great but you do it harder and the output doesn't increase linearly with it whereas
00:14:18.500 with ai for quite some time until pretty recently if if you just did more of the same you got
00:14:25.160 a commensurately larger output from ai which is like when you see that you you do everything you
00:14:33.220 can with it you want to find out just like you know you it's like finding some low-hanging fruit
00:14:37.660 and then finding a bunch of other low-hanging fruit next to it.
00:14:40.360 You're just going to keep picking that fruit
00:14:41.700 until you run at a low-hanging fruit.
00:14:43.080 So there's a lot of low-hanging fruit with AI, right?
00:14:45.540 That conceptually is very interesting.
00:14:47.200 But I don't think that accounts for it
00:14:48.380 because we still have these firms
00:14:49.920 that are now at a $1.4 trillion capital commitment
00:14:53.360 and that are grossing globally $50 billion a year.
00:14:57.940 And I know $50 billion sounds like a lot of money,
00:15:00.280 but you're an astrophysicist. 0.99
00:15:03.020 $1.4 trillion is a big fucking number. 0.99
00:15:05.360 Yes. 0.99
00:15:05.680 In a way that makes billions look like small numbers.
00:15:08.360 Yeah, yeah, yeah.
00:15:09.620 And so, why this exuberance?
00:15:12.800 And I think the exuberance is grounded in the ideological proposition that on the one hand, these major capital allocators, billionaires, Gulf states, and so on, are gripped by the fantasy of a world without people because they are gripped by the terror of what other people mean to them in the world.
00:15:29.600 Right?
00:15:29.840 If you are like a boss, you think you're driving, right?
00:15:33.140 You think you're in the driver's seat, but you know if you don't show up for work, everything happens just fine, whereas if the workers don't show up, the factory shuts down.
00:15:40.120 And so you suspect you might be in the backseat with a Fisher-Price steering wheel, and what AI dangles is the possibility of wiring that toy steering wheel into the drivetrain of the car. 0.99
00:15:49.780 So you have an amazing idea, and you prompt the AI, and it shits it out, and you don't have to have these ego-shattering confrontations with screenwriters who tell you that your screenplay is stupid, 0.53
00:15:59.380 with programmers who tell you your idea won't scale 0.97
00:16:01.480 or can't be secured or will run into problems
00:16:03.760 once it interfaces with the rest of the technological world
00:16:06.880 or that isn't in compliance.
00:16:10.020 You're a hospital administrator who finds a way to cut costs
00:16:12.860 and you don't have to contend with doctors
00:16:14.380 who have medical degrees who tell you you're going to kill people.
00:16:17.560 You know, this is the amazing thing about AI.
00:16:20.160 And so on the one hand, these guys quite like this, right?
00:16:22.460 If you're a billionaire, you already live in a fairly solipsistic world
00:16:26.660 where like all the actual flesh and blood people
00:16:29.000 you come into contact with just glaze you
00:16:31.020 from the moment they come into your company
00:16:33.040 until they part ways with you.
00:16:34.520 And on the other hand, you know,
00:16:37.200 like how could you hurt as many people
00:16:38.860 as you have to hurt to make a billion dollars
00:16:40.960 unless you don't really think other people's pain is real
00:16:43.640 and other people aren't real.
00:16:45.160 And, you know, you do what Elon Musk does,
00:16:47.740 which is call them NPCs, you know.
00:16:50.300 And so there's a certain degree of solipsism already there.
00:16:54.220 It's like, I think it's not a coincidence
00:16:56.360 as a ketamine is kind of the drug du jour.
00:16:58.920 I have chronic pain.
00:17:00.560 And, you know, I got a ketamine infusion once.
00:17:03.220 And like the overwhelming feeling of being in the K-hole
00:17:05.980 is that the whole world is something you imagined. 0.98
00:17:09.140 And that it's the whole world
00:17:10.820 is just taking place in your imagination.
00:17:12.400 It is the most solipsistic of all the drugs, right?
00:17:16.080 And then I think these people are also making a bet
00:17:18.260 that even if AI can't do away with the people,
00:17:22.220 that AI salesmen who can promise bosses
00:17:25.860 that this is a people removing technology
00:17:29.200 will be pushing on an open door
00:17:31.420 because bosses want this too.
00:17:33.960 And so you don't even have to believe
00:17:35.260 that it can get rid of the people.
00:17:36.840 You just have to believe that it can convince bosses
00:17:38.700 that it can get rid of the people
00:17:39.860 and that you can get out
00:17:41.160 before they figure out that it can't.
00:17:43.760 You know, you don't have to run faster than the bear.
00:17:46.060 You just have to run faster than the other investors.
00:17:48.380 Right, right, right.
00:17:49.360 And you make out okay.
00:17:51.580 Yeah.
00:17:51.740 No, this is...
00:17:53.740 this is related to like a question i get a lot when i run around and give talks and people ask
00:17:59.760 you know is ai going to take my job and i said well the good news is ai is not going to be able
00:18:04.160 to do your job as well as you do it uh or or really even close for almost any job uh the bad
00:18:10.900 news is in order for you to lose your job it doesn't have to someone just has to be able to 0.54
00:18:16.180 convince your boss that it can yeah yeah i i always say the fact that an ai can't do your job
00:18:21.780 doesn't mean that an AI salesman can't convince your boss to fire you and replace you with an AI
00:18:25.440 that can't do your job. Exactly. And, and so that is the, that's the, the, the reason that I think
00:18:31.840 the AI bubble is dangerous. And I think that is its material basis. And so if we're going to get
00:18:35.600 rid of the AI bubble, which is bad because people are losing their jobs. And so that's bad for those
00:18:40.460 people. It's also bad for anyone who depends on the work those people do. Um, and if we want to
00:18:45.680 get rid of the bubble because, because it's bad because it has environmental consequences because
00:18:50.100 it is ultimately designed to impoverish, scared old people who put their retirement savings into
00:18:57.920 what amounts to a stock swindle, right? Then we need to attack its material basis. We need to
00:19:03.740 attack the story that they tell to investors that gets the capital in the door. Because if we drain
00:19:10.700 off their capital, then the bubble goes away. Then all of these other things stop happening,
00:19:16.620 And it will be horrible and traumatic when it happens.
00:19:19.140 But the only thing worse than a $1.4 trillion bubble bursting is a $2.8 trillion bubble
00:19:23.840 bursting.
00:19:24.800 And that's the bubble that we're headed towards.
00:19:26.800 And so we have to get rid of this before it gets worse.
00:19:29.840 So how do we puncture that story?
00:19:31.320 I mean, that's some of what I have been trying to do with my work, because I think part of
00:19:38.080 the story here is a story about the existing AI being something that it's not, being more
00:19:45.860 like AI from science fiction,
00:19:48.220 like Commander Data from Star Trek or something,
00:19:50.560 rather than, you know,
00:19:52.520 the kind of compressed version of the internet
00:19:55.880 spitting out a kind of autocomplete on steroids
00:19:59.820 that is what we actually have.
00:20:02.160 I think that one of the ways to do this
00:20:04.180 is to resolve the seeming paradoxes of AI
00:20:06.780 because the paradoxes are used to confuse people.
00:20:12.680 So there is like one paradox of AI
00:20:14.600 is that you have programmers who use AI
00:20:17.020 who are skilled programmers
00:20:18.340 and who have historically been reliable narrators
00:20:20.960 of their own experience
00:20:22.340 who say that using AI is very good.
00:20:24.800 And you have other programmers who are skilled programmers
00:20:26.640 and reliable narrators who say it's very bad.
00:20:28.920 And when you actually examine the circumstances
00:20:31.540 under which they're working,
00:20:33.420 what you usually find is a centaur and a reverse centaur.
00:20:36.280 So by putting it in that framework,
00:20:38.300 then you get to something that resolves this paradox.
00:20:42.780 And then what falls out of that is, well, if the thing that AI is good for is making skilled workers work better, but without necessarily a reduction in cost, right?
00:20:54.660 But instead an improvement in quality, well, that's nice for us, but it's not going to produce the return on the investment you want.
00:21:00.740 Yeah.
00:21:00.880 You know, I am undergoing treatment for cancer.
00:21:06.600 I'm sorry.
00:21:07.300 Seems very treatable.
00:21:08.480 That's okay.
00:21:09.160 Good.
00:21:09.620 But I spent a lot of time talking to radiologists.
00:21:12.040 Yeah.
00:21:12.320 And so I am really very attuned to stories about AI and its ability to help radiologists.
00:21:20.520 Right.
00:21:21.000 And, you know, I was at the Kaiser Center this morning on sunset at the nuclear medicine clinic getting a PET scan, which is now on a radiologist's desk.
00:21:30.560 Uh, and, um, if I thought that there was someone working for Sam Altman down at the hospital administrator's office saying to him, look, right now you've got a million dollars worth of radiologists on the payroll.
00:21:44.620 It's 10 of them.
00:21:45.820 And they each do a hundred x-rays a day.
00:21:48.320 And here's what I propose.
00:21:49.980 You give me a half a million dollars a year and it's going to review the work of those radiologists.
00:21:55.420 And about twice a day, it's going to say, have another look.
00:21:58.620 and sometimes it'll be right
00:22:00.620 and you're going to save some lives
00:22:02.000 and your cost for radiology
00:22:06.400 are going to go up by 50%.
00:22:07.820 Your throughput is going to go down by 2%,
00:22:10.740 but you're going to save people's lives.
00:22:14.520 I'd be actually okay with that.
00:22:16.680 Yeah.
00:22:16.960 Right?
00:22:17.580 But the sales call that is going on
00:22:20.800 at hospital administrators all around America
00:22:23.020 is why don't you fire 90% of your radiologists?
00:22:26.340 Yeah.
00:22:26.740 Have the chatbot
00:22:28.620 do radiology at a rate that's too cheap to meter,
00:22:32.040 split the difference in the wage savings
00:22:34.100 with me, the AI salesman and my boss,
00:22:37.000 and make the radiologist sign the diagnoses
00:22:41.040 and blame them when people die.
00:22:44.320 Now, I've had radiologists write to me and say,
00:22:47.100 well, that's not how it's cashing out
00:22:48.200 because we have some labor power.
00:22:50.100 And so far, mostly we get to use the radiology software
00:22:52.720 the way we like it, which is great.
00:22:54.440 It's just how reverse centaurs work.
00:22:56.400 But that's not how you make back the cost of developing the radiology bot, right? You don't make back the cost of developing the radiology bot by giving more power to workers and having them do fewer outputs for more money, right? That might be how a public healthcare system that was well-funded would use this technology.
00:23:15.420 It's certainly not how the private sector is going to use it. And it's absolutely not how austerity crazed, you know, NHS or OHIP in Ontario are going to make use of it. They're only looking for cost savings.
00:23:26.520 You're reminding me sort of obliquely of something that Douglas Rushkoff said a few years ago before the current AI boom kicked off.
00:23:38.260 This story that he told of being brought to a room full of a few, you know, very wealthy tech moguls who asked him, how do we deal with societal collapse?
00:23:53.280 How do we use our money to insulate ourselves from societal collapse and guarantee the loyalty of security forces after the end of civilization?
00:24:02.500 What kind of bomb collar do I need to put on my mercenaries so they don't kill me and take my food after I move into the bunker?
00:24:09.080 Exactly, yeah.
00:24:10.020 And he said the best use of your money is to try to prevent the collapse of civilization.
00:24:14.620 and they just dismiss this as hopelessly idealistic
00:24:18.400 because it was, you know, less interesting
00:24:21.640 and would involve a hard look at their own actions.
00:24:26.340 But I keep thinking about that in conversations like this
00:24:30.640 because I keep thinking that these guys,
00:24:33.920 these tech billionaires and, you know,
00:24:36.020 the heads of the AI companies, the investors and whatnot,
00:24:38.900 see the ultimate promise of AI as a way of solving that problem.
00:24:44.920 Like, oh, what if we could have the bunkers
00:24:46.720 and we just have a completely automated defense force?
00:24:50.160 And so we don't need to buy the loyalty of our security forces.
00:24:52.580 Sure, it's the world without people.
00:24:54.280 Yeah.
00:24:55.220 No, it is very solipsistic.
00:24:57.380 And, you know, solipsism, I don't want to pretend that solipsism
00:25:00.180 is a vice confined to the rich, right?
00:25:03.200 I mean, hell really is other people.
00:25:07.080 Jean-Paul Sartre was not wrong.
00:25:08.900 Right. And, but the reason hell is other people is because other people are quite nice and yet they stubbornly refuse to acknowledge that you're right and they're wrong. And so when you want to do something that requires someone else, you have to either apply suasion or, or coercion to get them to do it. And that is very tedious.
00:25:29.500 And so, you know, I think the boom of AI boyfriends and AI girlfriends is the same impulse in different guys. It has different consequences. We're not firing all the girlfriends. So, you know, although I have to say that I think the fact that it's mostly AI boyfriends says something about who is harder to live with.
00:25:52.120 Yes. Yes, absolutely.
00:25:54.060 I think it's empirically harder to be the partner of a man than the partner of a woman. 0.97
00:25:57.880 Yeah, no, I think that that is true, but that's a separate conversation.
00:26:03.320 So that is a problem, right? And you see it up and down the stack, right? You see it with
00:26:08.620 Mark Zuckerberg saying, well, you have friends, which is great because they're a pain in the ass, 0.99
00:26:15.500 which means that you can't all agree when to leave Facebook. So, so long as I'm sure that 0.99
00:26:18.860 you hate me less than you love them, I can torment you in lots of ways with ads and slop
00:26:24.160 and manipulation and surveillance,
00:26:26.140 and you'll stay because you guys can't all agree
00:26:28.140 on where to go because you can't even agree
00:26:29.400 on what board game you're playing this weekend.
00:26:31.300 So you're just going to be stuck there.
00:26:33.260 But then he's like, you know,
00:26:34.140 the problem with your friends is
00:26:35.480 they don't want to optimize your friendship
00:26:38.060 to maximize engagement.
00:26:39.600 They keep insisting that the point of a friendship
00:26:42.300 is to be friends with people.
00:26:44.420 Clearly this is wrong.
00:26:45.760 And so I'm going to try replacing them
00:26:49.540 with like theater kids.
00:26:51.880 And if I like dangle remarkably small sums of money in front of theater kids, they will spend every hour that God sends trying to figure out how to maximize your engagement with the platform.
00:27:00.740 The problem is that an algorithmic feed of theater kids doing skits is fully fungible with any other algorithmic feed of theater kids doing skits.
00:27:10.380 And so you can go to TikTok or whatever.
00:27:12.900 And if it becomes non-fungible, right, if actually which theater kid you're watching matters, well, then that theater kid can start demanding things from Mark Zuckerberg.
00:27:20.420 So he's got a new plan. He's going to replace the theater kids with chatbots. And so we get rid of the theater kids, we replace them with chatbots, and we get social media without socializing, which is romance without romantic partners, which is a workplace without workers, which is screenplays without screenwriters, which is movies without actors.
00:27:39.360 You see this all the way through. And back to your book, I think that this is in a lineage with long-termism that says don't care about real people. Put all of your energy into making as much money as possible at the expense of living people.
00:27:59.840 and tell yourself that it's morally defensible
00:28:03.020 because you are going to create
00:28:05.940 an almost immeasurably small improvement
00:28:08.240 in the quality of life
00:28:09.460 of 10 to the 53 artificial humans
00:28:11.900 that we will compute on the bones of Venus,
00:28:14.840 which we will convert to computronium
00:28:17.180 10,000 years hence, right?
00:28:19.440 Which is the most solipsistic 0.99
00:28:21.380 goddamn ketamine-coded nonsense 0.99
00:28:23.260 I've ever heard, right? 1.00
00:28:25.440 That is a fantasy straight out of the K-hole.
00:28:28.580 no i can't disagree uh no i i just i'm i'm thinking about the angry emails i'm going to
00:28:38.280 get from ketamine users who listen to this podcast though that venn diagram might be zero
00:28:42.240 ketamine's super fun i i like look i i have i have i have experienced not just like what you
00:28:48.240 get if you do a bump of ketamine that you buy from some guy and then hopefully test with a
00:28:52.400 fentanyl strip. I have like
00:28:54.320 lain on a couch and had an
00:28:56.280 anesthetist and a cardiologist standing
00:28:58.340 by while they administered a lot
00:29:00.500 of ketamine to me. It was very
00:29:02.220 interesting and cool, but also
00:29:04.000 the quality of this delusion
00:29:06.120 is the ketamine delusion.
00:29:07.920 In the same way that
00:29:10.280 cocaine feels very 80s coded
00:29:12.300 and the 80s feel very cocaine
00:29:14.180 coded, I think 2020s
00:29:16.520 are very ketamine coded
00:29:18.220 and ketamine is very 2020 coded.
00:29:20.640 Cocaine had been around since I think the
00:29:22.160 20s and so is Kat right since
00:29:24.200 the 30s but it didn't
00:29:25.820 find it's like social
00:29:28.020 milieu until this
00:29:30.140 moment. Well now you're making
00:29:32.220 me want to bring on an expert on ketamine
00:29:34.300 maybe that'll be a guest for a future
00:29:36.340 episode. I thought you were going to say now I'm making you want to do
00:29:38.320 some coke. No.
00:29:43.320 No of all
00:29:44.220 of the drugs out there cocaine might
00:29:46.340 be the least interesting to me because
00:29:48.400 I have enough trouble with caffeine like I
00:29:50.360 don't like caffeine. It makes me sort of jittery and nervous. And so the idea of something that
00:29:54.240 would stimulate me further? Absolutely not. Well, more caffeine for me then. Yes, that's what
00:30:01.360 everyone tells me. Unfortunately, we still live in capitalism, and that means we need to talk
00:30:07.360 about our partners and sponsors. But fortunately, our partners and sponsors are great. And if you
00:30:12.520 want to support Dreaming Against the Machine directly, you can join us on Patreon, where $5
00:30:17.240 a month gives you the chance to ask questions of upcoming guests and a dedicated stream of cat
00:30:22.700 pictures. We love our patrons, and we also love being a part of the Multitude Collective at
00:30:28.560 MultitudeShows.com. Multitude is a podcast company made up of passionate people creating shows you
00:30:33.840 can count on. Shows like Tiny Matters. Science shapes every facet of our lives, but so much of
00:30:39.960 its influence is overlooked or buried in the past. Tiny Matters is an award-winning science podcast
00:30:45.120 about tiny things, from molecules to microbes, that have a big and often surprising impact on
00:30:50.440 society, past and present. From deadly diseases to forensic toxicology to the search for
00:30:56.280 extraterrestrial life, hosts and former scientists Sam Jones and Deboki Chakravarty embrace the awe
00:31:02.900 and messiness of science and its significance, asking questions like, why is wildlife thriving
00:31:08.740 at the Chernobyl nuclear power plant? Will humans ever live on Mars? And why is smallpox the only
00:31:15.100 human infectious disease we've eradicated. Tiny Matters is created by the American Chemical 0.96
00:31:20.200 Society, a nonprofit scientific organization based in Washington, D.C., and is produced by
00:31:25.540 Multitude. New episodes every Wednesday, wherever you listen to podcasts. And here's a word from
00:31:30.940 our sponsors. One of this week's sponsors is Shaker and Spoon, a monthly cocktail and mocktail
00:31:36.120 subscription. Full disclosure, they sent me a free box this month, and I'm excited to try it
00:31:40.880 on a night off soon. Each box includes three original recipes from award-winning bartenders
00:31:46.200 with enough ingredients to make four of each of those drinks for at least 12 drinks in total.
00:31:51.380 And each box includes everything you need to make the drink except for the alcohol.
00:31:55.740 And they're built around one spirit per box, so you really just need one bottle of something that
00:32:00.600 you probably already have on hand if you drink. And if you don't drink, every box includes zero
00:32:05.980 proof options for every recipe, and some of the drinks are designed to be completely spirit-free
00:32:10.820 from the start. The boxes come with recipe cards that include guidance on mixing techniques and
00:32:16.720 glossary of terms that teach you how to shake, stir, build, and garnish these unusual cocktails
00:32:22.340 with eclectic ingredients. And they also include everything else that you need to try out these
00:32:26.920 drinks like custom syrups, bitters, extracts, garnishes, and aromatics. And so you can try
00:32:33.560 the drinks without going off and buying full bottles of niche ingredients that you'll never
00:32:37.340 use again. It's a great idea for date nights or hosting a party or bringing a gift to someone
00:32:43.920 else's party. And you can always skip or swap boxes or even just buy individual kits or ingredients
00:32:50.560 without subscribing. So you can try this stuff out at their website, shakerandspoon.com.
00:32:56.500 Use code DATM for 10% off your purchase.
00:33:00.560 so how do we fight this bubble like yes we go after the story but but what else can we do i
00:33:09.940 mean like you and i we go after this story we we do that in in our work uh what can listeners of
00:33:16.520 this podcast do to try to fight this bubble well it really depends because it has these different
00:33:21.840 facets right so like in some ways it's just the normal stuff like what do you do about uh the fact
00:33:28.900 that local councils are being bribed to use eminent domain and to ignore environmental review
00:33:35.940 rules and to ignore consultation rules about water and energy usage to build data centers.
00:33:42.360 I mean, you restore those things. Now, how do you do that? Well, you get involved in local politics,
00:33:46.360 you become part of a polity. Local politics, easier to get involved with than you think.
00:33:50.300 I'm not saying you have to elect a senator here, right? Like, have you ever looked at a mom for
00:33:54.720 liberty. Like, have you ever heard one of them speak? They're not sending their best. They're
00:33:58.560 not sending their brightest. And yet they were able to take over a lot of poorly regarded local
00:34:03.380 offices and do a lot of damage. So just, you know, you got to get involved in your local politics
00:34:08.580 to really make that work. Now, as to the other questions, like what a creative workers do,
00:34:16.180 we've for 40 years now been expanding copyright. We made it last longer. We made it cover more
00:34:22.940 kinds of works. We made it cover more uses of those works. We created a much more punitive
00:34:28.120 statutory damages regime and a much more simple basis for achieving those statutory damages,
00:34:34.100 extracting them. We did manage to make the media industry much larger and more profitable in that
00:34:39.960 time, but we did not make artists richer. In fact, artists are poorer both as a share of the money
00:34:46.560 and in real terms than they were when we started. And that's because they're bargaining into a
00:34:51.540 market with like five publishers and four studios and three labels and two companies that do the
00:34:55.620 apps and one ebook audiobook company and giving us more to bargain with is giving us more to bargain
00:35:00.380 away and so like that you know it's like giving your bully kid extra lunch money there's just like
00:35:06.020 not an amount of lunch money you give them that gets them fed and so instead of that we can look
00:35:12.260 to the actual successes that creative workers have had in resisting ai and the most significant
00:35:17.860 success that any group of creative workers has had is the screenwriters. And the screenwriters,
00:35:23.000 now they're a militant union. They'd had a really important strike against the talent agencies that
00:35:28.140 were rolled up by private equity that helped them build institutional capacity. But what they also
00:35:35.260 have going for them is they and the other Hollywood guilds are the only workers in America allowed to
00:35:40.500 do something called sectoral bargaining. So sectoral bargaining is when a group of workers
00:35:45.000 are allowed to bargain with all the employers in a sector.
00:35:47.280 Like, if we had a sectoral bargaining for fast food,
00:35:49.820 everyone who worked at a fast food restaurant
00:35:51.580 would have a contract with every fast food owner,
00:35:54.620 franchisee in the country.
00:35:56.720 Lots of sectors are organized this way outside of America.
00:36:00.140 But in America, sectoral bargaining was outlawed in 1947
00:36:03.080 with the Taft-Hartley Act.
00:36:05.120 And the Hollywood unions were carved out, right?
00:36:08.540 So the reason, like, illustrators can't do
00:36:11.500 what the Hollywood writers did
00:36:13.300 is because they don't have sectoral bargaining
00:36:15.420 with everyone who hires commercial illustrators.
00:36:17.640 I'm just, I'm really angry right now
00:36:19.180 as a freelance journalist.
00:36:20.720 I want sectoral bargaining with, you know,
00:36:24.280 like where I can make common cause
00:36:25.800 with my fellow freelancers and staff writers.
00:36:29.020 A hundred percent.
00:36:30.140 And here's the kicker.
00:36:31.560 If you're out there fighting for more copyright,
00:36:33.760 you're on the same side as your boss 0.99
00:36:35.280 and no other work in America gives a shit. 0.99
00:36:37.640 And if you're out there for sectoral bargaining, 0.99
00:36:40.440 your boss is furious with you
00:36:42.420 and every other worker in America
00:36:44.000 wants the same thing as you.
00:36:45.620 Right?
00:36:45.940 So if we're going to like conjure up a new law
00:36:47.840 and we haven't gone into this,
00:36:49.080 but like as someone who's worked on copyright
00:36:51.000 for 25 years,
00:36:52.200 just anyone who tells you that it's obvious
00:36:54.060 that training a model is an infringement of copyright 0.99
00:36:56.360 is just shitting you.
00:36:57.860 That's not to say that we couldn't make 0.98
00:36:59.040 a new copyright law
00:37:00.000 that makes training a model unambiguously unlawful.
00:37:02.660 But you know, all these settlements,
00:37:04.300 like the settlement from Anthropic, 0.96
00:37:06.220 there was just a green light to go forward
00:37:08.280 against Facebook for downloading a bunch of porn.
00:37:10.260 these are all for pirating ebooks and movies they're not for training on them right like
00:37:15.440 counting the elements in a pirate ebook is not illegal for the same reason that if you go to the
00:37:21.300 flea market and count the number of beats per minute on a pirate cd you buy there that publishing
00:37:26.980 the count is not illegal even if buying the pirated cd is and we do not want to make identifying and
00:37:33.640 publishing facts about copyrighted works an activity that requires a license yeah major 0.67
00:37:38.240 proprietors of copyright are chuds like the Ellison's, right? Do we really want facts about 0.99
00:37:44.460 their copyrighted works to be something that you can only learn and publish if you have their
00:37:50.220 permission? This is nuts. And so, yeah, okay, maybe you think, oh, all right, well, I think I could
00:37:55.980 wordsmith a copyright law that would let us identify facts about the Ellison's copyright
00:38:00.520 output, but not let training happen. Then you still have to fall back on the fact that your boss
00:38:06.540 at the media company really would like to fire you
00:38:09.780 and replace you with a model.
00:38:11.280 And whenever these lawsuits drop,
00:38:13.680 the companies that are suing,
00:38:15.240 they publish press releases saying,
00:38:16.920 we're bitterly disappointed
00:38:18.000 that they didn't license this catalog from us.
00:38:21.540 Right?
00:38:21.720 They don't say we're bitterly disappointed
00:38:23.000 that you're trying to like replace workers with AI.
00:38:25.440 They're like, no, no, no.
00:38:26.180 We just want to be in charge.
00:38:28.520 Yeah, no, this is, well, okay.
00:38:30.340 Now you're reminding me of something that you said
00:38:32.440 back on that panel that I was talking about,
00:38:34.360 that, you know, you know that you've made some sort of mistake if you were fighting on the same
00:38:39.240 side as the bosses instead of the other workers? Well, at the very least, if you're going to do
00:38:43.840 some Bayesian reasoning here, right, and you're going to go like, I don't really understand these
00:38:48.080 the ins and outs of these issues. I'm not a computer scientist. I'm not a copyright lawyer.
00:38:52.220 I'm not a labor lawyer. But like broadly, when I'm on the same side as my boss and on the opposite
00:38:58.960 side of all the other workers and the other position puts me on the same side as all the
00:39:03.200 other workers, but the opposite side of my boss, I am probably going down the wrong path here.
00:39:08.300 And then, you know, I can explain to you as someone who knows a little about machine learning
00:39:13.280 and a lot about copyright law, why that is, but even if you knew nothing about it, it's just,
00:39:17.800 you know, like if I'm going to eat from this buffet and everyone who eats from it ends up
00:39:21.180 spending the night puking and shitting their guts out, and this buffet is full of people talking
00:39:25.320 about what a great meal they've had, and you don't know anything about microbiology, you can still
00:39:30.200 make a good judgment about which buffet to eat at uh yeah that's a good analogy
00:39:37.580 okay so in addition to being a incredible non-fiction author and uh and and essayist
00:39:48.680 blogger and whatnot as i said at the beginning you also write science fiction i do and i think
00:39:56.200 that i mean you know this from reading my book i think that a lot of these problems that exist
00:40:02.980 with the tech billionaires ideas about what the future could look like or the ideas that they're
00:40:07.160 trying to sell us on uh these guys are not particularly original their ideas came from
00:40:12.700 science fiction or at least bad readings of science fiction sure and so i i have to ask you
00:40:19.360 you know what do you think is going on there and and how do you feel about it as a science fiction
00:40:25.860 all that so that i think there is a always been a strain of science fiction fan and science fiction
00:40:31.180 writer who thought that they were seeing farther uh and they were predicting the future and i think
00:40:38.080 the more clear-eyed ones have always said no no these are parables that use the future to analyze
00:40:43.160 the present but i don't think that's unusual i think a lot of us get high on our own supply
00:40:47.620 someone told me this wasn't true so fact check me if i'm wrong here in the comments but there's a
00:40:54.220 a writer called el doctor who's not related to me who is dead now and he wrote a book called
00:40:59.820 a collection of essays called the creationists and the title essay is about the origin of
00:41:06.020 the genesis myth which the hebrews stole from the babylonians and the babylonians according
00:41:12.540 to doctor oh no relation the reason they thought it was true right you know in the beginning blah 0.70
00:41:18.400 blah blah yeah was that they just couldn't believe that an idea that cool originated in
00:41:23.340 their imagination, so they assumed it must be a revelation. They were just like, anything this
00:41:28.140 cool must be divine. That is, I think, a common place to go to. I think there is just a fine line
00:41:39.380 between being very excited about an idea you've had and assuming that if it's exciting, it must
00:41:44.460 be true or doable. And one of the wonders of being a science fiction writer is that you can get
00:41:50.940 excited about an idea and turned it into a cracking narrative without having to try and
00:41:55.200 make it happen, which, you know, is the place where a lot of these things founder, where you
00:42:00.500 have an idea and you start doing it and eventually you get trapped in the, you know, the last 10%
00:42:06.000 that takes 99% of the work, you know, and it just turns out that there's a reason that this thing
00:42:11.280 that seems obvious to you never happened. And it doesn't mean it won't ever happen, but it does
00:42:15.580 mean that like it's it's not necessarily within our grasp and one of the things about science
00:42:22.400 fiction is that it is a genre that is very closely connected to fantasy yeah and not just because the
00:42:30.060 same readers and writers congregate around and the same publishers and editors published it
00:42:33.780 but also because something about the trappings of science fiction where you create the illusion
00:42:41.260 of internal consistency and where you have enough things that feel true that the reader will go
00:42:47.440 along with you with thought experiments of things that can't ever be true that allows you to sort
00:42:53.020 of compose an internally consistent magic system and have it feel real but also allows you to have
00:42:57.820 a rocket ship that has a bunch of things that feel like they're scientifically plausible and
00:43:03.100 then you announce that it travels faster than the speed of light and the reader goes along with you
00:43:06.960 Yeah. Obviously that's fantasy. Yeah. Right. You know, as is the time machine, as is everything
00:43:13.480 else. Um, and you know, there was a movement within science fiction called the mundane movement
00:43:18.460 to write science fiction that didn't violate any known laws of physics. It was an interesting
00:43:23.260 constraint and some writers did some good work with it. Jeff Ryman particularly, but it never
00:43:28.080 caught on for a really good reason. Right. Cause the fantasy element is really important in science
00:43:33.420 fiction it's it's it's where your macguffins come from it's it's how you're able to do those thought
00:43:39.400 experiments you know and at its best one of the moves that science fiction makes is it um reifies
00:43:47.060 some aspect of our technology into something of global scale and significance in order to magnify
00:43:54.060 its capabilities and its limitations and its consequences and it's a bit like when you go to
00:43:58.500 the doctor and the doctor swabs the back of your throat and then she like rubs it on a petri dish
00:44:03.180 puts it in a medium over the weekend and then goes back on Monday and looks at it under a
00:44:07.640 microscope. And the thing that makes that diagnostically useful is not that it's an
00:44:11.860 accurate representation of your body, it's that it's a usefully inaccurate representation of your
00:44:16.020 body. We've reified one fact about your body, the gunk at the back of your throat. And we are
00:44:21.460 leaving aside everything else in order to find something out about you. But no one would mistake
00:44:25.520 the gunk for an accurate representation of your body. It is because it's inaccurate that it's
00:44:32.080 useful diagnostically and there's a lot of good science fiction that does this yeah um but it
00:44:37.820 leads people who misread it or who have a naive reading of it or who are naive in their writing
00:44:43.120 of it who fall prey to the babylonian delusion that anything this cool must be true uh it
00:44:49.100 misleads people into thinking that this simplified model can be can be uh actually applied used to
00:44:58.820 predict the future. Harry Seldon is real, you know, all of those, all of those things.
00:45:07.200 Harry Seldon is, is interesting. Harry Seldon, I guess, okay, for listeners who don't know,
00:45:11.420 Harry Seldon is sort of the central figure of the foundation novels by Isaac Asimov, which are
00:45:18.520 really important in the history of science fiction. And he was in those novels, the founder
00:45:28.220 of a discipline called psycho history, which allows you to predict the future scientifically
00:45:35.980 and to thousands of years into the deep future.
00:45:40.400 Yes, exactly.
00:45:41.820 It's why Paul Krugman became an economist very famously.
00:45:44.500 He keeps he says, I became an economist because I wanted to be Harry Seldon.
00:45:47.600 Yeah, no, I mean, and he has actually, if you go back and look at Krugman's early career,
00:45:52.980 he has a paper about the uh economics of interstellar trade uh uh taking special
00:45:59.740 relativity into account which is um a fun paper uh if if you're uh the right kind of nerd for it
00:46:07.020 which he is and apparently i am too um but um not that he thinks that interstellar trade is going
00:46:13.740 to happen it was just a right thought experiment he's just carrying he's just carrying the thought
00:46:19.000 experiment out yeah see what it would be like exactly yeah because like okay how do you deal
00:46:23.860 with uh the economics of trade when clocks don't run at the same speed on ships and whatnot because
00:46:31.120 they're going near the speed of light stuff like that um but yeah carl schrader did an amazing book
00:46:37.060 based on this i'm trying to remember what it was called uh but it's a it's based on the idea of
00:46:42.180 suspended animation and you have planets based on how much resource a planet has which is to say
00:46:47.980 how long it takes the robots to accumulate
00:46:50.100 enough to keep you alive. You slow
00:46:52.260 yourself down. So if you're on a planet
00:46:54.120 where to get a day's worth of resources
00:46:56.480 it takes a hundred years,
00:46:58.300 you sleep for a hundred years and then you wake up
00:47:00.180 for a day and then you sleep for a hundred years and you wake up for
00:47:02.180 a day. And there are other civilizations
00:47:04.080 where it's a week and other civilizations where it's a
00:47:06.120 month and other civilizations where it's a
00:47:08.160 day. And
00:47:09.700 he has
00:47:12.240 all of these in a set of interstellar
00:47:14.020 things. It's amazing.
00:47:16.400 I've never heard of these books.
00:47:17.980 Um, that, God, I, hang on, I'm going to look it up.
00:47:20.640 Okay.
00:47:21.320 Well, while you're looking it up, one of the striking things to me about those foundation
00:47:24.880 books, they go after a kind of technological determinism, right?
00:47:28.540 Harry Seldon.
00:47:29.240 It's called lockstep.
00:47:30.300 Okay, good.
00:47:31.260 Sorry.
00:47:31.620 Yeah.
00:47:31.920 Harry Seldon's sort of vision of the future is that, you know, technology is the one thing
00:47:38.060 that really matters.
00:47:39.520 And so, you know, and science, technology and science and everything else is just, you
00:47:45.280 know, messy details that that don't ultimately matter for the grand scope of human history.
00:47:50.760 And so this is like the idea behind these novels is he's doing, you know, Asimov was doing the
00:47:57.200 decline and fall of the Roman Empire in space. And and so Selden's idea was to take this science
00:48:04.660 of psychohistory and use it to take this interstellar dark age that was supposed to happen
00:48:10.760 between the first and second galactic empires
00:48:13.740 and shorten it by a factor of 15 or 20, something like that.
00:48:17.920 Yeah, that's the point, is it's not purely deterministic
00:48:20.580 because what he discovers is that there's a way to intervene in it.
00:48:24.700 You can't head it off, but you can change the way that it pulls out.
00:48:29.740 So, I mean, there's a sense in which that's a very humanistic message
00:48:32.800 that kind of jives with, say, the work of Ada Palmer,
00:48:35.300 who really has done amazing work to talk about how
00:48:40.760 There's a there's a certain amount that comes to you through the great forces of history and that produces an envelope within which human agency can change outcomes.
00:48:51.240 And the great forces of history are really just a word for the human agency of a previous era in history that set in motion certain forces.
00:48:59.160 But, yeah, it's it's it's it's not quite fatalistic.
00:49:04.340 Yeah. Right. He's not like like Kurt Vonnegut eventually gets into this place where he's just like, you know,
00:49:09.400 the first billiard ball in the universe, knock the second billiard ball, and then everything
00:49:14.100 has been determined since then. And nothing matters. We're all just like a cosmic Newtonian
00:49:20.900 dance. And even the fact that you're offended by that was set in motion by the billiard
00:49:27.220 balls, you know. Mark Twain, too. Mark Twain thought this.
00:49:31.320 Well, the funny thing to me, though, is, you know, Selden's utopia is sort of a utopia
00:49:37.720 of scientists and engineers.
00:49:40.300 And so it's sort of easy to see
00:49:42.740 where a tech billionaire might look at it
00:49:47.040 and say, oh, yeah, I want to be Harry Seldon.
00:49:49.080 On the other hand, when you say,
00:49:51.440 oh, well, you know, this is a set of sci-fi novels
00:49:53.940 that's centered around a person
00:49:56.640 who founded a scientific approach to history,
00:50:02.420 you know, the scientific approach to history,
00:50:03.920 that's a phrase that was already around.
00:50:07.720 And it's a line that belongs to someone else.
00:50:10.360 A guy with a big beard.
00:50:11.240 Yeah, exactly.
00:50:11.980 A guy with a big beard came up with that one.
00:50:13.600 Yeah.
00:50:15.100 Yeah.
00:50:16.200 You know, I think that what's interesting about those Asimov novels, psychoanalytically, in respect of the people we see today, is it supposes that the social sciences can be turned into a kind of physics.
00:50:29.320 Yeah, exactly.
00:50:30.340 And that qualitative questions can be eliminated from the social sciences, that you can quantize qualitative aspects of humanity with no loss of fidelity, and then stick them in models and perform mathematical operations on them and create high fidelity predictions of what's going to happen.
00:50:47.220 And of course, that's just not true. I mean, no, like, and Asimov was writing in the age of Claude Shannon, who's saying things like, you know, you're going to lose information when you, when you perform that quantizing step.
00:50:58.960 And yet, you know, this is where the hand-waving is, right?
00:51:03.120 And, you know, hand-waving is an art among science fiction writers.
00:51:08.740 Heinlein, but lots of other writers, have written AI stories that you can see the hand-waving turning into an investor story today.
00:51:15.340 Because the hand-waving is, we took computers and we did computer with them.
00:51:19.320 But we added more computers and we did more computer with them.
00:51:21.940 And then one day, they woke up.
00:51:24.360 Right.
00:51:25.100 Right?
00:51:25.500 That's just Sam Altman, right?
00:51:28.020 Sam Altman is doing the first chapter of The Moon is a Harsh Mistress.
00:51:33.180 Right.
00:51:33.520 But he's raising hundreds of billions of dollars off of it.
00:51:39.080 God.
00:51:40.740 And then Elon Musk wants to go to the moon and throw rocks at us.
00:51:45.200 Yeah.
00:51:46.200 Also, The Moon is a Harsh Mistress.
00:51:48.400 I know.
00:51:48.980 Yeah.
00:51:49.260 Yeah.
00:51:50.040 A little Heinlein is a dangerous thing.
00:51:52.200 A lot of Heinlein is an even more dangerous thing.
00:51:54.140 I was about to say, yeah.
00:51:55.560 Heinlein's not...
00:51:56.600 Let us all thank ourselves that they're confining themselves to early Heinlein and not the stories he wrote when he had the benign tumor that was restricting the flow of blood to his brain. 0.99
00:52:06.400 And he told his editor to fuck off and started writing 900-page novels about traveling back in time and having sex with his mother. 0.96
00:52:14.900 I think Charlie Strauss would be the person to bring on to talk about that. 0.99
00:52:19.340 He's definitely the guy to talk to about that.
00:52:21.380 Yeah.
00:52:22.280 Oh, God.
00:52:23.300 Anyway.
00:52:26.600 Well, we, uh, we've gone pretty far afield, but, um, I mean, it is, it is interesting to me though
00:52:35.680 that, you know, again, it's not an original idea, right? The idea that you can take the social
00:52:40.560 sciences and, and, and sand off the things that make them complicated and difficult and turn them
00:52:45.480 into physics. I mean, speaking as a physicist and also as someone reasonably conversant in the
00:52:50.240 history of science, no, you can't do that, but that's not an original idea, right? You know,
00:52:54.680 people have wanted and tried to do that for many, many years, and they have failed because it
00:53:00.120 doesn't work, right? Well, look, the Nobel Prize in economics was created because economists were
00:53:07.780 tired of being not taken seriously as a quantitative science, and so they found a guy called Nobel to
00:53:13.060 give them a prize. It is now folded into the wider Nobel infrastructure, but it's not a Nobel Prize.
00:53:21.460 Yeah, it's the not a Nobel Prize in economics.
00:53:24.120 Yeah, no, it's the Nobel Memorial Prize.
00:53:27.280 Yeah, yeah, it's the like butthurt consolation prize
00:53:31.340 for people who want to be taken seriously as physicists
00:53:34.080 when really what they're doing is social science.
00:53:36.140 Yeah, no, I mean, the thing that's crazy about that to me as a physicist
00:53:40.520 is like the reason why physics admits like closed form mathematical solutions
00:53:47.820 to some, not all, of the problems of interest in the field
00:53:52.340 is precisely because physics is going after
00:53:55.980 just about the simplest questions
00:53:57.860 that you can ask about the natural world, right?
00:54:01.020 If you start asking questions about systems
00:54:03.920 that are more complicated than a few particles
00:54:07.700 or a highly idealized system of non-living things
00:54:13.100 or the universe itself at scale so large
00:54:16.760 that you can just forget about anything
00:54:18.780 that is smaller than a galaxy cluster
00:54:21.140 and any forces other than gravity.
00:54:23.340 Like, unless you're dealing in one of those domains,
00:54:27.020 the world is just too complicated
00:54:29.300 to admit of those sorts of solutions.
00:54:32.280 Well, unless you're one of those farmers
00:54:33.740 who breeds perfectly spherical cows
00:54:35.360 of uniform density on your frictionless surface.
00:54:37.560 Exactly, and they emit milk isotropically
00:54:39.980 in all directions, yeah.
00:54:41.460 Yeah.
00:54:44.120 No, I mean, but that's exactly the thing, right?
00:54:46.760 I mean, the social sciences have to be complex in other ways, in ways that physics is less complicated, because otherwise they're not going to be able to study social phenomena.
00:54:59.440 But that makes them messy and brings us back to the fact that hell is other people.
00:55:04.980 Hell is other people.
00:55:05.560 You know, the funny thing is, you're right when you said earlier that, like, the annoying thing about other people is that they don't understand that you're right and they're wrong.
00:55:16.000 But I think that the tech billionaires, I think that they see it in an even more sort of nakedly sociopathic way.
00:55:25.600 And that the thing that's awful about other people is that they need things.
00:55:30.880 Sure.
00:55:31.820 Yeah.
00:55:32.020 But that's a way in which they're wrong.
00:55:33.340 Right.
00:55:33.740 Right.
00:55:33.960 Like, you know, Jeff Bezos has very strong ideas about whether you need to pee.
00:55:39.280 And the answer is you don't.
00:55:40.600 Right.
00:55:41.000 And yet you keep insisting on peeing.
00:55:43.320 Yeah.
00:55:43.540 And like the people who are like, oh, we got to do quality of life stuff in San Francisco because people poop everywhere.
00:55:50.960 And the thing that we can't do to resolve that is give them toilets.
00:55:54.200 We can do anything except put toilets in San Francisco to stop people from pooping everywhere.
00:55:58.880 I promise you it's not going to work, right?
00:56:01.620 There's just like there's no amount of telling people they're not allowed to poop that will stop people from pooping.
00:56:07.360 I read that important textbook as a child.
00:56:10.340 Everybody poops.
00:56:11.160 uh and uh i've i've absorbed its message you're reminding me of the the the sort of administrator
00:56:18.900 brain that happens when you decide that when when somebody comes to you with the complaint that for
00:56:26.780 example there's too many snack wrappers accumulating in the computer lab and the solution is not to put
00:56:34.340 uh trash cans into the computer lab because people aren't supposed to be eating in the computer lab
00:56:38.560 because there are computers in there.
00:56:39.780 So the solution is to put a sign on the computer lab
00:56:42.060 saying no food and drink allowed in the computer lab.
00:56:45.200 Well, and it's the same thing that says,
00:56:46.740 well, okay, people who have a house
00:56:49.480 find it easier to get off drugs,
00:56:50.940 but they shouldn't be taking drugs,
00:56:52.620 so they can't have a house until they're clean and sober.
00:56:55.300 Oh, God. 1.00
00:56:56.240 Yeah, that's sort of sanctimonious crap. 0.99
00:56:59.020 Well, it's the same thing all around. 0.99
00:57:00.440 It's the difference between harm reduction
00:57:01.820 and honestly, a kind of consequentialism
00:57:05.480 and utilitarianism, which these guys say they embrace.
00:57:08.080 Right.
00:57:08.560 Right. I mean, I'll call it idealism. It's not idealism in the sense of hewing to any ideal that I like, but it hews to an ideal that says that, you know, you should all do what I tell you. And if you're not smart enough to do that.
00:57:22.320 And maybe this is a good place to close things out because this is something that I, well, I guess I mentioned it toward the end of my book, but I think that in a way, these guys are too idealistic in the sense that they are married to an idea about how the world is regardless of all evidence to the contrary, right?
00:57:45.660 Peter Thiel thinks that he's not going to die or that he doesn't have to die.
00:57:50.500 Unless Greta Thunberg kills him.
00:57:52.320 That's exactly. Oh God. Um, but yeah, I mean, I just, and, and yet they see themselves as the 0.98
00:58:04.160 hard nosed practical men out in the world. Yeah. It's, it's true. Like that is the posture,
00:58:10.160 right? It's, um, facts don't care about your feelings. Right. Exactly. Right. I I'm the,
00:58:15.020 I'm an empiricist. You're squishy. You can't admit that you are just a bag of predictable
00:58:22.420 outputs in response to inputs. I am BF Skinner. You are in my box. And back to sort of
00:58:31.600 billionarism as being closely related to solipsism, I think broadly, you know,
00:58:37.000 Pax seeing like a state, anything that requires you to deal with people as populations rather
00:58:41.960 than as individuals induces you to stop thinking about people as real and start thinking about them
00:58:47.360 as statistical phenomena and not as things that have qualitative internal experience.
00:58:53.080 I agree. And that's a downer. And we have to find a way to bring these people back onto a level
00:59:03.760 where we can deal with them. You know, the best politicians, the ones that I have lots of respect
00:59:09.560 for. Yeah. Don't seem to lose sight of that. You know, Mom Donnie, I know he's only been in office
00:59:14.140 for a little while, but Mom Donnie seems very centered on the individual experiences of his
00:59:18.580 constituents. Yeah. In a way that is very different from, say, how Rob Ford, who was the
00:59:24.080 mayor of Toronto, was infamous for, like, he was the guy, if you, if you called, so he had a talk
00:59:29.220 radio show where he and his brothers, now the premier of Ontario, would do, you know, right-wing
00:59:32.980 talk radio. But you could ring in and you could say, I have a pothole. And he would show up and
00:59:37.200 fill it right which is like that's another way of thinking about other people is real but it's very
00:59:42.540 different from saying like i want to do policies that make people better off yeah because i listen
00:59:50.060 to them all i listen to where the potholes are coming from and what's stopping the potholes
00:59:54.680 from coming to existence and i listen to the pothole filling workers and what stops them from
00:59:59.320 filling the potholes and i and i asked experts how to improve pothole filling and he's filled
01:00:03.940 hundred thousand potholes during his tenure right mom donnie and that's more than rob ford ever
01:00:09.580 managed so there is some way to view people as real without becoming one of these pathological
01:00:14.840 weirdos like rob ford probably involves also not smoking crack yeah or doing too much ketamine
01:00:22.820 yeah we're doing too much ketamine just enough ketamine yes exactly don't do too much drugs and
01:00:28.480 stay in school, kids. That's
01:00:30.260 the message of this podcast.
01:00:32.320 Right. Good. Okay, well, Corey,
01:00:34.680 it has been a delight having you
01:00:36.540 on the show, and let's
01:00:38.520 talk more soon. All right. Nice chatting.
01:00:41.440 Thanks again to this week's
01:00:42.680 guest, Corey Doctorow. Next week
01:00:44.660 we have Katie Mack,
01:00:46.600 and we're going to be talking
01:00:47.780 about quite a bit farther in the future
01:00:50.660 than we usually do on here. We'll be
01:00:52.700 discussing the end of the universe.
01:00:55.620 To submit
01:00:56.420 questions for future guests and to suggest other guests, and to see more pictures of Babka,
01:01:03.040 join the conversation on Patreon. You can also find us on YouTube, on Instagram at DATMPod,
01:01:09.580 on the web, and on Blue Sky at DreamingAgainstTheMachine.com, or just find us wherever
01:01:15.020 you get your podcasts. Dreaming Against the Machine is a proud member of Multitude Productions.
01:01:20.560 Our executive producer is Nick Carissimi.
01:01:23.820 Our associate producer is Rosie Thomas.
01:01:26.300 Our theme music is by Jared Emerson Johnson.
01:01:29.220 Our show logo is by Nick James.
01:01:31.780 And our fearless leader is Babka, the greatest cat in the observable universe.
01:01:36.780 I'm Adam Becker, and I'll see you next week.