What I Read in July 2026

Cory Doctorow's latest non-fiction release is an explanation of his views on the recent surge in so-called "AI" tools. This is a clear-sighted and well-reasoned view, I think. I don't entirely agree everywhere, but I think a lot of his points are hard to argue with.

He explains how tech bubbles work (it's all about keeping investors believing tech stocks will grow), then uses this to frame the current mass hysteria spreading from Silicon Valley. Doctorow also explains the concept of centaurs and reverse centaurs. Centaurs use tools to empower them, like a human head with a horse's body. Driving a car is centauric. Reverse centaurs are people forced to take action by a machine. Amazon Mechanical Turk workers are reverse centaurs.

There's a big focus on the use of "AI" in creative work. This is probably my favourite section. It changed my mind somewhat on the whole copyright question. Hitherto I've considered this a problem of "theft", I guess in the form of copyright infringement. Doctorow argues that training models is not infringement convincingly. Instead, the trouble is one of labour. Artists are not compensated for their work now because of monopoly power in creative industries. "AI" doesn't change this fundamentally, and increasing the power of copyright just gives more power to those currently abusing artists, instead of the new people abusing artists.

This also includes a lovely philosophical discussion on the nature of Art, and why "AI"'s attempts at making it feel so soulless. I really liked Doctorows perspective on this.

The book ends with the useful residues of bubbles. Doctorow believes this bubble has some useful residues in potentially cheap compute, and open weight models. I'm not sure on the compute side, I think I agree more with Ed Zitron – the amount of power required to run these GPUs makes them essentially useless for most of us.

Open weight models are here to stay though. I agree with Doctorow that they provide fundamental new capabilities for machines, and they can be used for good in some circumstances. I disagree with Doctorow that chatbots are a good use though. Perhaps this is just my own aversion to talking to people, but I've never wanted a chat interface for any task on a computer, beyond talking to my friends. I believe that the real uses of this technology are beyond the chatbot paradigm, taking the form of more traditional tools with powerful language/image/audio processing capabilities. For one thing, that should remove the urge to use them to avoid thinking.

Overall I highly recommend this book. Doctorow, as always, uses a clear voice to give sharp, incisive criticisms of technology. People who are at all affected by "AI" should read it.

This was a nice chaser to Unpaid, which I found coincidentally in this article by Cory Doctorow. It's a more in-depth discussion of the mechanics of Uber's disgusting practices of wage discrimination and algorithmic pricing.

As per usual, there's also the use of lobbying legislatures to preserve inherently immoral business models, particularly California's Proposition 22, which made the preposterous idea that gig workers are "independent contractors" permanent. It also stipulated that it can only be overturned by seven-eighths of legislators unanimously voting to do so.

• Why Reading Matters – Cal Newport
• The Dawn of Post-Literate Society – James Marriott
• The Elite College Students Who Can't Read Books – Rose Horowitch
• The Age of Reading is Over – Rose Horowitch

After reading the Cal Newport piece I fell down the links. The broad strokes are that reading is hugely in decline, and this might be a problem. (It certainly could be for the authors, who all write things as part of their livelihoods.) If you don't want to read all of these, but do want to read one, I'd recommend the Newport piece (short) or Horowitch's The Age of Reading is Over (long). I'm not sure exactly how I feel on all this.

On the one hand this has a whiff of moral panic about it. Anything The Atlantic writes about college students should probably be treated with a hefty dose of caution. (See the last 20 years of panicking about safe spaces and free speech on campus – I don't think it was the woke students we should have been worrying about when it comes to freedom of speech.) The Marriott piece also has a feeling of Western chauvinism to it. Complaints like this have accompanied most changes in reading – from novels to newspapers – as the pieces themselves point out. I'm also not convinced this is entirely new. I would have thought reading has decreased ever since the radio gave a different form of mass entertainment.

On the other hand, the articles are very convincing. The statistics Newport pulls out of the longer Horowitch piece speak for themselves. Reading long-form material is on the decline. I cannot imagine that this is a positive thing for our society. It feels tied to the anti-intellectualism that's been prominent in recent years, and the dwindling attention spans in the modern world. I certainly think there's cause for concern here.

The examples of even university students studying English-related fields being unable to read deeply especially concern me. One of the articles contains excerpts from students being asked to interpret a passage from Dickens' Bleak House, which is especially jarring. I'm sure that the article (and the study it's citing, which I decided I didn't want to read enough to spend $25 on) is cherry-picking the most surprising examples, but the statistical patterns on declining reading lists on courses is clear.

I was forced to reflect again at my own habits in recent years. Since university I've had a long struggle for my attention with my phone. My reading declined as I spent more time online, and the more digital time I spent the smaller my attention span got. It was a self-reinforcing cycle – the more I used the screens, the harder focusing got; and the harder focusing got the more it felt like all I could do was look at screens. Reading was feeling impossible.

Since then I've undergone a long and arduous process to try and disentangle myself. In dribs and drabs I have extricated myself from the algorithms, for the most part, but I still have work to do. I've long since left social media, but the attention thieves still had me – Reddit and YouTube were my hardest things to quit. It was not until recent years, when I rendered my phone an inert brick unless it's required for banking, that I finally started to take my life back.

I don't think smartphones are good for us. We cannot cope with being in constant reach of a box of boundless stimulation. It's exhausting, ti depletes our attention, and it robs us of both solitude and companionship. I think the world would be a happier place if more of us read books than scrolled distracting digital media. I know I'm happier the more I do.

An eye-opening account of Starmer's rise and fall. With friends like Labour Together and Starmer, who needs enemies? The continued ties between Labour Together and Burnham does not give me much hope for his premiership.

Christ, what drivel. The argument boils down to: businesses have no social obligations but to make money for shareholders, because shareholders are empowering executives to act as their agents. Shareholders, being alien creatures from the planet ShareHoldia and external to human society, have no responsibilities, so businesses have no accountability whatsoever. Of course businesses have social responsibilities – the shareholders, executives, and employees have them because they're all people living in societies!

Don't fear, Friedman suspects some shareholders might believe they have an ethical responsibility beyond sacrificing the poor on the altar of Mammon, and has an argument against it. These shareholders should be ignored, because other shareholders are greedy and care only about maximising profits regardless of the cost in human misery. Favouring the responsibility to the one set of shareholders rather than the other is definitely not a moral judgement, because morality is real difficult to choose between.

Friedman doesn't even believe that businesses have no social responsibilities. He caveats his doctrine of greed with "without breaking the law". That is a social responsibility, Milton. Those poor hard-done-by shareholders could afford another yacht or two if their executives would just press gang child slaves to save on labour costs, what makes that social responsibility any different to the rest?

Amusingly he includes "not committing fraud" and "not lying" as constraints on businesses too. These are even more like social responsibilities than the law, which perhaps could be argued is voted on by democratically elected governments. There's nothing to explain why this social responsibility is different, especially when it's so frequently broken by Milton-fans.

I cannot believe that this idea is so influential. Business leaders really were grasping at straws to justify greed, huh?

Someone shared a post which linked to this in work. It's a beautifully-written narrative about the use of LLMs in writing. I can't tell to which degree it's real, but I hope the end is, at least. Reading this perspective on their use in writing changed me a little.

The piece raises two concerns with the use of LLMs for improving writing. LLMs strip writing of its humanity, turning it into bland corporatised junk. Ultra Processed Prose. The use of LLMs also ruins the confidence of writers, who begin to see its output as "better". This summary does not do the writing justice – go read it yourself!

I've been thinking a lot about the use of LLMs recently – haven't we all? – especially in the context of software engineering. I've been trying to work out how I would use LLMs if all my ethical concerns about them were magically solved¹. There's only one use case I've found that I wouldn't object to, personally ². Review. Applying the LLM to work I created for critique.

Kingett has convinced me not to do this for prose. Even for technical writing. I think the quality of any writing comes from the voice that's writing it. That applies to literary tales or thorough technical documentation. Knuth is a stellar example – his playful enthusiasm rings through his writing. W. Richard Stevens is another. The world would be a poorer place if those voices were lost to blandness.

I'm also more hesitant to endorse their use for software. I still think such "semantic linting tools" would be useful in a professional context. The costs of software errors range from frustrating to fatal. An extra run through one of these tools to catch such errors strikes me as positive, albeit not exhaustive and noisy. Outside of work, though³, I no longer like the idea so much. The confidence crisis such critique could cause creates a clear cost. Maybe it would help people learn, but nurturing dependence would be a real downside.

My rambling aside, this moved me to tears. Go read it instead of this! Gosh, I wish I could write like that.

Truly an enlightening zine. I always get excited when a new one arrives. Standouts for me this time were Computiles, About Anamorphic Cryptography, the one about using playing cards to encode a private key, Empty Capabilities, First Steps in C, Linux in a Pixel Shader, and the one about a disgusting multi-user Android CVE where a user with a screen reader hears other users' messages.

Would recommend this to anyone with an interest in computers.

---

Other media I've consumed include software engineer and Zig creator Andrew Kelley's recent talk at Software Should Work. It handles our social responsibility as technology workers, and exhorts us not to fall into nihilism as we're coerced into creating software that coerces others. One quote stuck out to me in particular:

We are learning that software quality was maintained only through the will of software engineers combined with their accidentally strong negotiation position in the job market

The "only" might be a tad reductive, but it seems fair.

In the talk Kelley also shouted out Ninajirachi's wonderful album I Love My Computer. It was not new to me from the talk, but it is a great listen. Much of my listening this month has been to MASTER BOOT RECORD and KEYGEN CHURCH for the purposes of soundtracking a Pathfinder game.