"Fisher-Price Management" With Cory Doctorow  🔗 **Source:** ["Fisher-Price Management" With Cory Doctorow](https://share.snipd.com/episode/1cb16dbc-20bf-43f9-bb40-431d9d24f1d0) ## Highlights - **Fisher-Price Steering Wheel Illusion** - Executives often occupy a symbolic driver's seat while real work continues without them. - Cory Doctorow likens that illusion to sitting in the backseat with a Fisher-Price steering wheel, and AI can connect that toy wheel directly to the drivetrain. Transcript: Cory Doctorow Yeah, I'm indebted here to Riley Quinn from Trash Future, who is a font of excellent analogies. So, you know, if you are the boss, you fancy that you're in the driver's seat, but you know, if you don't show up for work, everything will more or less be fine. But if all the people who report to you don't show up for work, nothing is going to happen. And so maybe you're not in the driver's seat. Maybe you're in the backseat with a Fisher-Price steering wheel. But AI and the AI sales pitch is, what if we wired that Fisher-Price steering wheel directly into the car's drivetrain? So you have an idea and you ask a chatbot and it shits out the product. And there is no like ego shattering confrontation with someone who knows how to do something, who explains to you that your idea is like impossible or illegal or stupid, or you're all Going to get it, go to jail, or it's going to murder people. ([Time 0:04:48](https://share.snipd.com/snip/3f373500-0cf5-43fa-844d-79b5fa8ccb25)) - **AI As A Solace For Insecure Bosses** - AI comfort lets bosses avoid confrontations with skilled workers by producing seemingly plausible outputs on demand. - Doctorow recounts Hollywood writers using a chatbot like giving bad notes, which then spits out a script without someone to push back on feasibility. Transcript: Cory Doctorow And, you know, my first inkling of this, because the rest of the time when I'm not in London, I spend my time in L.A. We live in Burbank right near the studios. And I'm a member of one of the Hollywood guilds. I'm a writer in the Animators Guild. And so during the writer's strike, I went out on those picket lines and I was on the Warner picket line one day. A writer told me, you know, you give a chatbot notes or you give a chatbot prompts the same way you give shitty notes to a writer's room, right? Make me E.T., but it's about a dog and put a car chase and a love interest in it, which admittedly is just air bud. But if you say that to a writer's room, either tacitly or depending on how bullshit they're feeling explicitly, they will say, why don't you fuck off and make a spreadsheet? People here are making a movie. And that is a thing you assuredly do not know how to do. Go away. And what the chatbot does is it just shits out the script. And the fact that that script is almost certainly not producible in any way might not be as important from the perspective of the sales call and maybe from the perspective of the people Investing in the company that makes the chatbot as the fact that bosses are just like infinitely horny for this psychological comfort of engaging in the solipsistic fantasy that other People don't exist. ([Time 0:05:44](https://share.snipd.com/snip/3d9ccf83-d363-4161-bc3a-a36e792a72f1)) - **How Anti-Circumvention Killed Useful Hardware Markets** - Anti-circumvention rules became law through industry lobbying and trade leverage, not obvious public benefit. - Doctorow explains how DVD de-regionalization and printer ink markets were killed after countries accepted anti-circumvention in trade deals. Transcript: Cory Doctorow Right. Which is why, you know, there's not a company in Malaysia today that will make you like the universal printer unlocking kit that makes it just disabled, for every printer disables their Ink checking routine so that you can use generic ink or refilled cartridges with them, which would be a giant business. I mean, ink is like $10,000 a gallon. And so, you know, like you could just supply that to everyone. You know, 25, 30 years ago, Malaysia led the world to making de-regionalized DVD players. But once it signed up to its US free trade agreement, it promised to criminalize people who reverse engineer DVD players and de-regionalize them. And so that became the, that, that killed that, that export sector. It used to be like Malaysian DVD players were the world's, you know, most sought after DVD players because they were all region zero. Nick Suresh You know, growing up there, it didn't even occur to me that that was special. I, didn't even know regionalized DVD players existed. I just assumed that. Interesting. I've learned something. Cory Doctorow Well, anti-circumvention is a way to effectively invent a law because there's no copyright law that says if you go to Australia and bring home a Blinky Bill DVD and try and play it in Florida, That like you're not allowed to watch it, right? Like you bought the DVD, it's yours, it's your property. But, you know, the sector wants to segment the market and DVDs make it hard to do. You know, in the age of CRTs, it was easy because you had conflicting video standards, Palin and TSC, DVDs that just goes away, they're multi-output, digital, whatever. And so this business model becomes a law, right? The way anti-circumvention works is once it's illegal to reverse engineer and modify something, then if you make something that enshrines a business model, reverse engineering And modifying the device to frustrate your commercial desires becomes a crime. Jay Freeman calls it felony contempt of business model. And so this is why we've seen the metastasis of these what are called access controls, DRM, not just in DVD players, but printers, ventilators, cars, tractors, in smart speakers, phones. Obviously, Apple's single largest line item on the revenue sheet is the $100 billion a year they get from processing transactions for the app store because you have to use their 30% Payment processor for every dollar you collect in their app store. And it's just illegal to reverse engineer the phone and make a jailbreaking dongle you can buy in the checkout aisle at Walmart or Spar that jailbreaks your phone and installs a regional App store that charges 3% to the same software vendors and service providers. And so Apple gets to preserve this $100 million market. ([Time 0:16:47](https://share.snipd.com/snip/0bb120cb-31ba-42cf-b210-f2dae26d3d49)) - **Digital Sovereignty Beats National AI First** - Digital sovereignty risk is more immediate than national AI stacks because cloud control can be weaponized politically. - Doctorow warns that shutting off Office365 or bricking phones/tractors effectively kicks nations off the modern internet infrastructure. Transcript: Cory Doctorow Australia is sitting there, along with every other country in the world, having a digital sovereignty crisis. And the digital sovereignty crisis is fucking real. You know, Trump like shut off Microsoft Office 365 access for foreign officials that made him angry. Right. The International Criminal Court, he bricked it for swearing out an arrest warrant for genocide against Netanyahu. The high court that sentenced Jair Bolsonaro, bricked. Right. You lose your M365, you lose the email address you use to log in everything else because that's your Outlook address. Yeah. You lose all your working files, all your cloud documents, email archives, calendars, address books, everything. You're effectively like kicked off the fucking internet. And he's doing this to foreign governments. He could get up tomorrow and say like, I want Greenland. I'm not going to put tanks in nuke. I'm just going to shut down Denmark. Nick Suresh I was thinking about this this morning because I was chatting with Ed Zitron, who I don't know if you've spoken. Yeah, I'm guessing you have. I love it. So I was chatting with Ed and he said something he revealed some numbers to me and he went, you know, maybe this spells doom for Microsoft. Cory Doctorow And I had to think about that article. Yeah. Yeah. Nick Suresh Yeah. But I was also thinking about it. And I went, but Microsoft can just like 5x their prices and everyone has to, I mean, at least for a couple of years, just suck it up. Right. Like, what are you going to do? Cory Doctorow And it's not just office productivity software, although that's very important. It's also hardware. So you remember when those John Deere tractors were stolen from Ukraine and ended up in Chechnya by like Putin's thugs, John Deere kill switched them remotely. They can brick any tractor in the world. So again, Trump doesn't even have to shut down all the ministries and companies in Denmark. He can just brick all the tractors in Denmark. So there's this urgent need to get your data out of these silos and to jailbreak the firmware locks on these devices and also to create like compatibility layers between conflicting Services, right? If you're the EU, like the EU has built its own blue sky server, but like maybe they'll pouch some people off of blue sky. They're not going to get people who are stuck on Twitter and Instagram because they're stuck there because like they have a rare disease and that's where the people have the same rare Disease as them are, or because like they moved from another country and that's where their family is, or, you know, it's where their customers are, or their audience or whatever. But like when Facebook started, they solved this by giving everyone a MySpace scraper that you gave your MySpace credentials to, and it scraped everything waiting for you on MySpace And put it in your Facebook feed so you could eat your cake and have it too. You could still talk to your friends on MySpace, but use Facebook to do it. It's very like prodigy AOL ass nonsense that you have to care which network someone's on to talk to them, right? So you can make that app, right? All of that stuff is possible if you get rid of anti-circumvention law. So AI, when you talk about digital sovereignty, it's almost always an argument about how we build AI in our country. I was just on a panel in Toronto where five of us, four of them were like, how do we build national AI? And I was like, guys, if Trump orders Dario and Sam to switch off all your chatbots tomorrow, not one firm fails, not one ministry shuts down. ([Time 0:19:58](https://share.snipd.com/snip/af1b4be3-5692-4201-a725-b178a0d530d7)) - **Firing Skilled Workers For Chatbots Creates Fragile Voids** - Firms may fire skilled workers and replace them with chatbots that can't actually perform, creating long-term capability loss. - Doctorow fears told staff will retrain away and be unavailable when chatbots fail or are switched off. Transcript: Cory Doctorow I'll tell you what I'm afraid of though, right? Back to this idea that the AI can't do your job, but your boss is infinitely horny for replacing you with a chatbot. I think we're going to take people who do real jobs. We're going to fire them. We're going to replace them with chatbots that can't do their job. And so they'll do a half-ass shitty job of doing their job. But the people we fire are going to become discouraged. They're going to retrain. They're going to exit their job market. They won't be reassemblable when the chatbots all get switched off. And then we shut off the chatbots. And then we just have a void where like, you know, take functions we should never give to a chatbots, but we're doing anyway, like deciding whose children get taken away by the state, Right? Taking children away from parents who are abusive is a regrettably important thing that the state needs to do. It's a significant function of the state. And if we fire all the people who know how to do that job, which is a hard job that already the state doesn't do well, and we replace them with chatbots, and then the chatbots go away, we still Have this pressing need to protect children who are in abusive households. And the people who know how to do it are not available anymore. And the chatbot that was supposed to do it has been switched off because the data center went bust. And now what? ([Time 0:24:33](https://share.snipd.com/snip/e0985266-7492-4018-bfe1-008e38faa6dd)) - **Process Knowledge Keeps Complex Systems Alive** - Process knowledge is distributed, tacit, and non‑commodifiable, so firms undervalue it until it breaks production. - Doctorow gives factory and chip-fab examples where informal contacts and judgment saved lines and created new product opportunities. Transcript: Cory Doctorow And then there's the kind of the process knowledge, right, which I think is very systematically undervalued. And again, this ties into AI. So I'm sure you're familiar with this idea of process knowledge as opposed to IP, right? That's sometimes called tacit knowledge. But I think tacit knowledge is mostly an individual matter, whereas process knowledge tends to be spread out across individuals within divisions, across divisions within a firm, Across firms within a sector, both vertically and horizontally. And that process knowledge, it's like, if you can write it down and put it in a filing cabinet, it's not process knowledge, it's IP, which is to say that if it's a thing you can train a chatbot On, it's not process knowledge. And if it's process knowledge, you can't train a chatbot on it because it's all tacit. It doesn't exist in a form that can be fed to AI. And process knowledge is systematically undervalued by markets and firms because it's not commodifiable. Right. So it never shows up on a balance sheet. So important that we have the work to rule, which is when workers show up for work and only do what the IP says and don't employ any process knowledge. And then firms just shut down, right? If all you do is follow procedures, where you wrote in your piece that firms are very bad at writing down the procedures, in part because there's parts of the procedures that cannot be Embodied. They're like super contingent. They're what the Germans call fingerspitzengefil, the fingertip feeling, you know, and it's literally things like I apprentice. So I know with the pace at which feedstock goes in this machine based on a whole bunch of feedback that's coming from the machine. I can hear it. I can feel it. But I also know because I have process knowledge that when the machine jams and the person who can fix it is off, the person who trained me who retired last year, will do it for a hundred Pounds. And here's his mobile number. And like, that's the difference between the line shutting down and not shutting down. And my favorite example of recent process knowledge is there was someone at a chip fab where a whole ton of chips failed QA. And they only ran at like 50% of their nominal clock speed, but they knew they had a downstream customer where there was a faction within the firm that had been briefing for making lower End computers. And they called them up and they said, what if I could give you a ton of chips for like cheap? And they said, yeah, I think I can sell this new product line, which is called the MacBook Neo. And it's the most successful product Apple has launched in 10 years. Right. And it's so successful, they used up all the shitty chips and they're now making shitty chips for new MacBook Neos. Right. Like that's, that stuff is just that when you vaporize that, it doesn't come back. Right. It has to be rebuilt through like hard yards where you have to make like the machine has to break down over and over again. And you have to call in repair people and have like days long shutdowns of the factory while you wait for the machine to get fixed before you figure out the finger spitzing the finger to Feeling of how to load the machine. And before you get the phone number of the person who fixed it the first time, but then quit. But who will come in and do it for 100 bucks. ([Time 0:26:22](https://share.snipd.com/snip/e92a0680-8b5c-4e55-a6f9-5c922a1bfbb9)) - **Consolidation Turns Market Competition Into Internal Politics** - Large, consolidated firms turn competition inward, making internal politics the main career battleground. - Doctorow explains steep hierarchical pay jumps create 'dead man's shoes' dynamics where executives jockey and rat‑fuck to secure the next big initiative. Transcript: Cory Doctorow I also want to suggest to you that consolidation in a sector produces the circumstance where within a firm, the major competitor for any executive is another executive within that Firm. So it creates a kind of autoimmune disorder, right? Where like if you think about a Fortune 100 where you've got like directors, VPs, EVPs, SVPs, or SVPs, EVPs, presidents, CEO. That's a typical studio structure. Other companies have different ranks. But there's something like an order of magnitude difference in lifetime earning expectation between each of those tiers. And also, it's a very steep pyramid. And so it's dead man's shoes, right? You don't move up unless someone gets fired or promoted or dies. And so if there's a new thing, right, that might be the next thing, right? Like if AI might be the next thing for your firm, if you think about like the tech sector successors, right, where like Microsoft successor was the guy who figured out cloud for them. Apple successor was the guy who figured out logistics and Chinese contract manufacturing for them. And Google successor was also a cloud guy. Right. Their patronage network moves one step in one order of magnitude up the lifetime earning expectation on this steeply canted pyramid, which means that when there is anything that might Be that thing, everyone starts to, on the one hand, jockey to become the person who runs that, but quietly, because if the other execs think that you're going to make the move, then then Their patronage network all start to rat fuck all of your vassals in order to destroy them, in order to weaken your power base, because they don't know who should do this. They know it shouldn't be you, because if it's you, it's not their boss. ([Time 0:41:15](https://share.snipd.com/snip/a3a5d07f-f64e-438d-bae2-886a08eb9b7a)) - **Quantity Metrics Make AI Outputs Misleading** - Goodhart's Law shows metrics-led AI adoption can inflate meaningless outputs like PR counts. - Doctorow notes managers chase numbers (PRs, token spend) and junior coders boost superficial fixes without the discernment to spot subtle AI errors. Transcript: Cory Doctorow Yeah, that's a good question. It's a very number go up ish kind of phenomenon where like, you know, there's Goodhart's law. Yeah. That, you know, targets, metrics become targets, right? Like if you don't care about the quality of the PR, you can make a lot more PRs. And, you know, we saw this early with a lot of GitHub projects and other like, like get hosted projects, not just on GitHub, but GitLab and whatever, where people got really excited about AI, not just execs, but just like junior coders. And we're like, there's a project with 10 outstanding bugs. I'm going to fix all 10 of them with my chat bot. And I'm going to be the most bug fixing son of a bitch on this little corner of open source land. And I think it's, I don't know. Back to not caring how things break, but how they work. If you can just satisfy a unit test or whatever, if you can build, and you don't really have the discernment to distinguish between... I guess that's it. It's the discernment to distinguish between a good bug fix and a bad bug fix. It's especially pernicious if you think other people aren't real and what they do is nonsense. And that, you know, not as hard as what you do, because then you look at it and you're like, oh, well, that was easy. I guess that person is as much of an idiot as I thought they were because you don't understand what they do. But also, like you just lack discernment. I actually think this is a deep problem with the AI story about AI displacing labor. Like you, I'm less dismissive than Ed about whether AI is useful. But I think that AI is primarily useful to skilled practitioners who make thoughtful judgments about when and how to integrate it into their workflow because they have the discernment To distinguish good AI output from bad AI output. And they have the self-knowledge to know the pace at which they can make those fine distinctions because oftentimes, because AI is doing statistical extrapolation from things that Work, the things that break that AI makes are often as statistically close to a thing that would work as it's possible to be without actually working. And so it's really, really hard to spot those bugs. ([Time 0:56:49](https://share.snipd.com/snip/7115b098-b997-4249-a5c9-a21fa3a795ad)) - **AI Helps Teachers But Can't Fix Regimentation** - AI can improve teaching quality by auto-generating tests and answer keys, but it won't fix rigid standardized systems. - Doctorow describes his daughter's US high school where tests were single-use and teachers lacked flexibility; AI could only be a band-aid. Transcript: Cory Doctorow She went to a big, dumb American high school. It's actually the school they based Glee on. It is a very American-American high school. You know, they have a homecoming football game. And totally outside of my experience, I went to a weird alternative school in Toronto where the only sports team we had was an improv team. But she she did the math, right? She did like higher math and she was never allowed to take home her math quizzes or tests or exams because there was only one test. Because they have to teach certain concepts in a certain order because it ends up with standardized tests with SATs and AP classes and so on. So the teachers are extremely constrained in what they can teach. And so writing something that is this pernickety, a test that's this pernickety, it's a lot of work. So they have one and they don't want the answers leaking out. So you write your quiz, but you don't get to see, like you can review it in class, but you don't get to take it home to review the mistakes you've made. This is fucking batshit, right? My dad was a math teacher. Like you cannot teach math this way. This is like teaching target shooting by making you shoot and then like blindfolding you before you see where the bullets go in. You know, like it's right. It's nuts. So I could totally see a teacher vibe coding some Python that spits out tests and answer keys for every exam that go to those specific concepts and like use, you know, different variables Or numbers or whatever. It'd be very easy to write a unit test that like validates it. I could see them actually just like making a prompt that produces again, like you'd want them to actually work backwards and make sure that the questions are real and that they solve The way that they think they're doing. But that's the thing that if you have discernment, you can just look at it and you can tell, right? You just work the problem. So that would be a thing that would significantly improve the quality of the education. That teacher couldn't teach more students. ([Time 1:04:40](https://share.snipd.com/snip/ccbf917f-c46d-4a0d-abe1-a4a0df44f123)) - **Use Benign Neglect To Let Teams Self Organize** - Give teams autonomy and avoid heavy-handed microcontrol; benign neglect often produces high-performing groups. - Nick Suresh and Cory describe successful teams that thrived because directors 'left them alone' to self-organize. Transcript: Nick Suresh Like not many people fly out to Australia from there and get a highly paid software engineering job and all that stuff. And I've been writing about this in the blog before, but I was partially saved from that because I actually went six years without an education because I got put into a local school which Taught exactly the way you're talking about. And because I didn't speak Malay, my parents didn't teach me. The teacher just left me alone for six years with books. Oh, wow. And like that is grossly irresponsible. Right. They should not have done that. But Malaysian teachers, it's not like Australia or the US where they have a master's degree in education. Right. It's frequently like just some guy who left a village. Right. You know, two years. And they went, this is slightly more than working at McDonald's. You know, go crazy with it. There's 50 students per class. Wow. And it turned out just being totally uneducated was better. Benign neglect. Yes. Benign neglect is how, not just in education, right? I think most managerial control is so, even competently done, is just like inherently harmful in like because of the modern paradigms around it almost every effective team i've seen Even in software has been created under the condition of benign neglect it's usually just a guy who's like well i'm meant to be the director here but i don't think anyone will notice if I go fishing every day so i'm gonna go fishing and then just with you know, the right sort of alchemy of people who are motivated in the right way and the right juniors, they just sort of Become really high performance. Even, I probably shouldn't name a specific company, but there was a company which I really looked up to and a lot of our advisors come from it. And I was talking to a staff engineer there and I went, how did you all get so good at this? You know, what was your brilliant management technique? What did the directors do? Right. And he went, ah, benign neglect. That's actually the first time I heard that phrase. I just went, ah, they just left us alone. And some of us became really functional. And that was that, right. We just spread it internally. ([Time 1:08:43](https://share.snipd.com/snip/9f31b7f1-6ced-4e1d-9912-a49ef816ff04)) - **Don't Buy AI Now Buy The Steelcase Chairs Later** - Bubbles inflate hype and produce durable, cheap salvage after crashes, so wait for the downturn before heavy investment. - Doctorow recalls buying Steelcase Leap chairs for $25 after the dotcom crash as a metaphor for buying AI infrastructure at 10 cents on the dollar later. Transcript: Cory Doctorow I was in San Francisco during the dotcom bubble, and it was quite instructive. So I had always wanted a steel case leap chair, these very good ergonomic chairs. I know. But there was no way I was going to spend fifteen hundred dollars on a chair. And I really coveted them because a lot of the people I knew who worked in dot-coms had them. So like after the bubble burst, I was walking down the street in the mission and I saw a dot-com CEO and it was like those photos from 1929 of millionaires on Wall Street selling apples Off the bumper of their rolls. Except it was a dot-com CEO who lost his lease who had like a fleet of Steelcase Leap Chairs in the plastic wrap. And I bought six of them for $25 each off of them. And I use them as a dining room set for the next 10 years. And I feel like if you're an Australian policymaker or a firm or a university trying to figure out what your AI strategy is, and you're like, what we should do is buy AI now, you are buying Steel case leap chairs in the early spring. Well, early, I won't use seasons because of Australians, but in March of 2020 instead of September of 2020. Because we can see the crash on the horizon. There's going to be GPUs at 10 cents to the dollar. There's going to be data centers going begging. There's going going to be open source models that have plenty of room for optimizations. And there's going to be lots of skilled workers who know how to mobilize them. And like, don't buy this stuff now. Are you crazy? ([Time 1:17:10](https://share.snipd.com/snip/b1b71919-e3f7-4571-b999-1021bf42cc0b)) - **Use The Memex Method To Synthesize Ideas** - Build a durable personal knowledge system by blogging and revisiting notes rather than hoarding disorganized snippets. - Doctorow recommends his 'Memex Method' to synthesize ideas by iteratively improving public notes and connecting fragments into new work. Transcript: Cory Doctorow In the space, well, so I would say if you want to learn the method for being able to kind of combine information this way, read a little short essay I wrote called The Memex Method, as in Vannevar Bush's The Memex, which is about how blogging and revisiting your blog posts rather than taking notes for yourself is a very powerful way to mnemonically fix concepts in your Mind and then iteratively improve your understanding of them and combine them in novel ways that you kind of create a super saturated solution of fragments that kind of rattle around. And when they nucleate, they crystallize into synthetic ideas and stories for me and novels and so on, but also nonfiction and so on. That this is like, it's a hugely effective method. ([Time 1:19:19](https://share.snipd.com/snip/180a8751-2e7f-4944-ab85-524531eb5835))