
Welcome to Pointers, the monthly newsletter from the generative futures platform, Lucid Dot. Each month we’ll share around a dozen links, some from our scanning engine, some on a single monthly theme, some on books, some on podcasts.
They’ll share a general connection with the idea of collapse and renewal, inspired by The Rift project, which is being run by our development partner at Chatham House’s Sustainability Accelerator. We also hope that we’ll fill some of the gap left by the much-missed CUSP newsletter, edited for years by Ian Christie.
In this edition:

AI beyond the technology
If you use AI enabled tools, it helps to keep a critical distance from the theological outpourings from the west coast tech bros. So, in the spirit of drinking our own champagne, we did three runs of the Lucid Dot Horizon Scanner with slightly different questions to ask for scan hits that gave a critical picture of the AI world. These scan hits come from those runs, although the synthesis was done by a human. In the Most of these come from these runs. We’re not going to do this every month—but there’s a second part next month on fighting back.
Everyone now expects an AI market collapse. This idea is now completely mainstream. The European Central Bank has issued a report, the Financial Times has run a big story. Bloomberg called AI finance “a wobbly house of cards.” The London analysts Capital Economics screened a full set of market indicators and concluded that we’re in “a late-stage bubble”. Research from the tech analysts Gartner suggests that "enterprise and consumer AI revenue will be insufficient to support profitable returns on all planned AI capital expenditures during this period." But of course, the market can remain irrational longer than the average investor can remain solvent, so it’s helpful to know how the crash might happen.
The best answer I’ve seen comes from the anonymous investment analyst Groundbreaker, who points in a detailed and measured way to the trillions of dollars of computer commitments that fall due in 2027-28–at which time they may not have the revenues to cover these costs. OpenAI is much more exposed than Anthropic. Of course, the question is: if there is a crash, how contagious is it? The Australian Financial Review is pessimistic: the five hyper-scalers have data centre commitments worth about 4% of US GPD, and (says the Brookings Institution), these debts are already being packaged up and securitised, just as mortgages were before the 2008 financial crash.
The security issues from AI agents run deep. It was bad enough when AI agents were just attacking the sites of governments and public agencies, but October was something altogether different—so different that people were queuing up to invent neologisms for it. Thezvi.com went with “The Hackening”, while the AI Safety Newsletter called it the “Hackapocalypse”. The normally sober The Atlantic magazine said, “This week could go down as the worst in the history of cybersecurity.” At the heart of the issue: AI agents had identified security vulnerabilities in PHP, a critical part of the internet’s core infrastructure that is used in more than two-thirds of the world’s websites. A lot of sites went offline while they applied security patches. But this was just the headline: security flaws and hacks were announced the same day in several other major software products or platforms.
It was a reminder that to a large extent the internet is kept safe by backroom work by national and international agencies, the large tech companies, and non-profit organisations. For example, following the the PHP crisis, PrivacySafe Foundation released a whole suite of tools to help administrators identify risks, detect attacks, and strengthen online systems running a whole set of widely used internet tools, such as node.js (itself compromised in ‘Black October’), Ruby and Python.
Matthew Prince, the CEO of the internet security company Cloudflare, told The Atlantic that “it’s going to feel like the sky is falling in” for the next year or more. Code has always had vulnerabilities; they’ve just been harder to find. Fixing this will involve rewriting billions of lines of code. After that? AI tools might help fix the security issues, and run security checks. Or we might become dependent on a smaller number of companies that can secure themselves against repeated AI attacks.

If you have been forwarded this newsletter and want to subscribe to it, please click here:
Lucid Dot helps organisations make better decisions through scanning and world-building tools that change how they see the future.
If the insights in this newsletter are valuable to you, you’re welcome to join the Horizon Scanner’s waitlist by signing up here. We’ll be in touch with you.

Seen and noted
As big as Texas
The biggest surprise of the US mid-terms wasn’t that the Democrats took a decisive hold of the Senate (although that was a surprise) but that they swept the ballot all down the ticket in Texas. The prospect that demographics might eventually change the politics of Texas has been a trope for about 20 years, but it has been a long time coming. The implications of that for all of US politics could be huge. Nils Gilman has a good piece on his Substack working through all of this. (Thanks to Tanja Hichert for the link).
A post by Aditi Juneja of Democracy 2076 — a progressive take on the election results — also pointed us to some interesting American attitudinal data on political reform from the University of Maryland. In summary, majorities of both Democrat and Republican voters nationally are in favour of political reform across a wide range of issues, including non-partisan redistricting (to replace gerrymandering), limits on election spending by corporations and super-PACs, banning of stock trading by Congress members, White House employees, and Supreme Court members, term limits of Congress members, and a “standard ethics code” for Supreme Court members.

James Talarico, the first Democratic Senator in Texas to be elected in 38 years. Photo by Antonio Esparza. CC BY-SA 4.0.
Books and culture
Strange futures at the Booker Prize
The Booker Prize is Britain’s most prestigious literary prize, and this year’s winner—to general surprise, including his own—is M. John Harrison, who combines aspects of sci-fi and fantasy fiction into his work without being constrained by those genres. His winning novel, The End of Everything, is described by his publisher like this:
A slyly satirical and unsettling post-apocalyptic adventure into the limits of human understanding.
There’s a long appreciation of M. John Harrison’s work by Henry Farrell at his Programmable Mutter substack. This extract from an interview with Harrison about The End of Everything jumped out at me:
One of TEoE’s major differences from an ordinary post-disaster story is that its core event is an explanatory collapse, a cloud of unknowing. This is not a side issue, it is the nature of the disaster described in and by the book… One of the first things about epistemological failure is that you don’t know what you don’t know. A novel which makes that its central thought–and which rigorously passes that condition on to the reader–offends every possible traditional idea about the novel: about the hallowed privilege of knowing and the controlled communication of that knowing to a reader.

Stepping into a Reading River
Lucid Dot’s CTO, Peter Curry, has a side project that helps you to save articles that you’d like to read but don’t have the time right now. There’s quite a lot of neuroscience sitting behind the design, as Peter explains here, where there’s a little video demo. The Reading River is literally shown on screen, with your saved articles floating downstream until you retrieve them. Miraculously, they never seem to get wet.
Podcast: Nvidia, sex and spies
The Jensen Huang story at Nvidia has all the elements of a late John LeCarré novel. An American tech billionaire with a string of lovers, one of whom may have been a Chinese honeytrap. But maybe not. On the Ones and Tooze podcast Cameron Adabi and Adam Tooze talk about the history of industrial espionage and question the current version of the story. Listen here.

If you have been forwarded this newsletter and want to subscribe to it:
About Lucid Dot
Lucid Dot is an AI-enabled futures platform that changes what you already know. The platform currently has two products: a Horizon Scanning engine, so you can notice what is changing; and The Fizz, which turns signs of change into vivid possible worlds and so stretch your imagination. Our products are designed to put power into your hands, so you can act on your own terms.
Like all our work, this newsletter is made possible by AI, and is made better by human curation and editing.
We have started a waiting list for people who want to try out the Horizon Scanner and The Fizz for themselves. You can sign up for either the Horizon Scanner here, or the Fizz here. If you’d like to have a conversation, feel free to leave your details here, or to reach out directly to David, at [email protected].
Like all our work, this newsletter is supported by AI, but is made better by human curation and editing.
If you want to suggest an item for the next edition, or have other feedback, please send it to the newsletter editor, Andrew Curry, on [email protected]
Pointers is published by Lucid Dot Ltd, under this Creative Commons licence: CC BY-NC-SA 4.0.

