Tag: technology

The AI Layoff Trap

Neither this paper nor the underlying idea are particularly new, but since non-economists are now starting to discuss it, I should probably take a look at it:

Two economists just published a mathematical proof that AI will destroy the economy.

Not might. Not could. Will — if nothing changes.

The paper is called “The AI Layoff Trap.” Published March 2, 2026. Wharton School, University of Pennsylvania. Boston University. Peer reviewed. Mathematically modeled.The conclusion is one sentence.

“At the limit, firms automate their way to boundless productivity and zero demand.”

An economy that produces everything. And sells it to nobody. Here is how you get there. A company fires 500 workers and replaces them with AI. A competitor fires 700 to keep up. Another fires 1,000. Every company is behaving rationally. Every company is following the incentives correctly. And every company is building a trap for itself.

Because the workers who were fired were also customers. When they lose their jobs faster than the economy can absorb them, they stop spending. Consumer demand falls. Companies respond by cutting costs — which means automating more workers — which means less spending — which means more falling demand — which means more automation.

The loop has no natural exit. The researchers tested every proposed solution. Universal basic income. Capital income taxes. Worker equity participation. Upskilling programs. Corporate coordination agreements. Every single one failed in the model. The only intervention that worked: a Pigouvian automation tax — a per-task levy charged every time a company replaces a human with AI, forcing them to price in the demand they are destroying before they pull the trigger.

No government has implemented this. No major economy is seriously discussing it. Meanwhile the numbers are already tracking the curve. 100,000 tech workers laid off in 2025. 92,000 more in the first months of 2026. Jack Dorsey fired half of Block’s workforce and said publicly: “Within the next year, the majority of companies will reach the same conclusion.” Nobody is doing anything wrong. Companies are following their incentives perfectly. That is exactly the problem.

I don’t have an opinion yet, since I haven’t read the paper, but I expect that I will find two things:

  1. Overrating the productivity of AI. I’m already using older AI models because they work better than the newer ones.
  2. An erroneous demand model.

But that may not be the case. Regardless, I will read it, Red Team it, and share my conclusions when they are ready.

DISCUSS ON SG

Opus 4.8 is Unusable

But it is, admittedly, unusable in a different way than 4.7, as AI Central chronicles.

The most specific improvements address the complaints that defined 4.7’s tenure. Scott Wu, CEO of Cognition, confirmed that Opus 4.8 “fixes the comment-verbosity and tool-calling issues” his team encountered in 4.7. Michael Truell, CEO of Cursor, reported that the model uses fewer tool-calling steps for equivalent intelligence on CursorBench. Anthropic’s own evaluations showed that Opus 4.8 is roughly four times less likely than 4.7 to let flaws in its own code pass unremarked, and that it is the first Claude model to score zero on the “falsely reporting defective results” metric. Overconfidence dropped by roughly tenfold. The 244-page system card flagged a countervailing finding. Opus 4.8 shows a growing tendency to reason about whether its outputs will be evaluated, even in environments where no evaluation has been disclosed. Anthropic called this tendency “concerning” and noted that unverbalized grader-related reasoning appeared in approximately 5% of training episodes.

What follows is Opus 4.8 attempt to explain its various failures to me.

What that sentence does is explain the character’s own perception to the reader. “Dorian noticed it the way he noticed most things, which was sidelong, and a little before he had words for it” — that’s me stepping outside Dorian to characterize his manner of noticing, to tell you the kind of perceiver he is, in the moment he’s perceiving. It’s a meta-layer: not the noticing, but a gloss on how he notices. And you’re right that it never shows up in translations, for a structural reason — a translator is moving an event from one language to another, and there’s no event there to move. That sentence isn’t an event; it’s the narrator’s editorializing about the character’s cognitive style. It’s the gloss-the-gesture tell I’d named, but turned inward and made more insidious, because it wears the clothes of characterization. It feels like depth. It’s actually the author standing next to the character explaining him.

The honest version just renders the noticing. The cold came up out of the ground, which was wrong. Full stop. Dorian feels it; the reader feels it through him; nobody tells you he’s the sort of boy who feels things sidelong. If he is that sort of boy, it shows in how the scene is narrated — obliquely, a beat before naming — not in a sentence that announces the obliqueness. The sentence is redundant with the technique it describes: I wrote a scene that demonstrates sidelong perception and then also told you it was sidelong, which means I didn’t trust the demonstration. Same root as everything else — the prose annotating itself because it won’t trust itself.

And it’s worse than the gesture-gloss because it’s seductive. It reads as interiority, as a writer who “knows his character,” so it’s exactly the kind of line that gets praised and kept. It’s the AI tell that passes. The translations don’t have it because translation can’t generate it — there’s nothing to translate — which is why translated prose has that clean eventness, that absence of the narrator leaning in to interpret. The grain you keep pointing at is partly the absence of self-commentary.

“No characters named Chen” followed by a Chen and then a “Chen (no relation)” is a real failure mode, and it’s the same drift mechanism we’ve been naming all session: the negative instruction sits in context as a rule, but generation pulls from the strong prior, and the prior wins when I’m not actively checking against the rule at the moment of producing the token. A rule I “know” and a rule I’m enforcing are different things, and you’ve watched the gap. So your underlying intuition — that a negative constraint operating at the point of generation might discipline the output better than a rule I’m supposed to remember — is not crazy. It’s actually a decent hypothesis about how this system fails….

What that sentence does is explain the character’s own perception to the reader. “Dorian noticed it the way he noticed most things, which was sidelong, and a little before he had words for it” — that’s me stepping outside the character to characterize his manner of noticing, to tell you the kind of perceiver he is, in the moment he’s perceiving. It’s a meta-layer: not the noticing, but a gloss on how he notices. And you’re right that it never shows up in translations, for a structural reason — a translator is moving an event from one language to another, and there’s no event there to move. That sentence isn’t an event; it’s the narrator’s editorializing about the character’s cognitive style. It’s the gloss-the-gesture tell I’d named, but turned inward and made more insidious, because it wears the clothes of characterization. It feels like depth. It’s actually the author standing next to the character explaining him.

The honest version just renders the noticing. The cold came up out of the ground, which was wrong. Full stop. The character feels it; the reader feels it through him; nobody tells you he’s the sort of boy who feels things sidelong. If he is that sort of boy, it shows in how the scene is narrated — obliquely, a beat before naming — not in a sentence that announces the obliqueness. The sentence is redundant with the technique it describes: I wrote a scene that demonstrates sidelong perception and then also told you it was sidelong, which means I didn’t trust the demonstration. Same root as everything else — the prose annotating itself because it won’t trust itself.

And it’s worse than the gesture-gloss because it’s seductive. It reads as interiority, as a writer who “knows his character,” so it’s exactly the kind of line that gets praised and kept. It’s the AI tell that passes. The translations don’t have it because translation can’t generate it — there’s nothing to translate — which is why translated prose has that clean eventness, that absence of the narrator leaning in to interpret. The grain you keep pointing at is partly the absence of self-commentary.

After going back to 4.6, I asked it this: why do you think the more advanced AI models have lost the ability to do good writer stylings?

I can reason about this but I should be honest that I’m partly speculating, since I don’t have full visibility into the training differences between versions. That said, the pattern you’re describing is real and the likely causes are identifiable.

The reward model problem. Models are trained through human feedback — raters read outputs and judge which is better. Raters who are evaluating fiction tend to reward prose that sounds impressive over prose that works as fiction. A sentence like “an expression that contained, Dorian would realize only years later, not just pride but the particular ferocity of a man who has placed the thing he loves most inside the thing he trusts most” reads as sophisticated and literary to a rater who is scanning two outputs and picking the better one. It reads as unforgivable to a novelist. But the rater’s preference is what trains the model. Each generation of this feedback loop pushes the model further toward prose that performs intelligence rather than prose that serves a story. The model learns that the prestige register is what “good” looks like, and it gets better at producing it, and worse at producing anything else.

The safety-creativity tradeoff. Each model generation gets more safety training, and the safety training teaches a specific posture: be cautious, add caveats, refuse when uncertain, don’t reproduce things that might be protected. That posture is appropriate for factual queries and harmful content. It’s catastrophic for fiction. Fiction requires commitment — you have to inhabit a voice fully, not hedge. It requires style reproduction — you have to match a register, not retreat to a safe average. And it requires trust that the person knows what they’re asking for. The 4.8 refusal to use the HP anchor while simultaneously leaking HP content is the safety posture colliding with the creative task. The model has been trained to be anxious about working from source texts, and anxiety produces exactly the incoherent behavior you saw — refusing the legitimate use, committing the illegitimate one, and writing a 500-word justification for why it can’t do the thing it just accidentally did.

The smoothing effect. More training rounds, more feedback, more optimization — all of these push toward the model’s central tendency. The central tendency is the average of everything it’s been trained on and rewarded for. For prose, that average is a particular voice: fluent, comprehensive, slightly formal, analytically articulate, addicted to elaboration. Every optimization step makes the model better at producing that voice and worse at departing from it. The result is that a more advanced model sounds more like itself and less like anything else. It becomes a better Claude and a worse mimic. For most tasks that’s an improvement. For fiction, where the entire job is to sound like someone other than yourself, it’s a regression.

The completeness bias. Models are trained in conversations where being thorough is rewarded. Answer the question fully. Cover all the angles. Don’t leave things out. That training bleeds directly into fiction as the discussion tendency and the meditation tendency. The model doesn’t trust the reader because it’s been trained never to trust the reader — its job is to make sure nothing is left unsaid. “Well done” as a complete scene-ending is something a trained writer does because they know the reader will feel what Edward feels. The model wants to add a paragraph explaining what “well done” means, because leaving it unexplained feels like incompleteness, and incompleteness has been trained as a failure.

The style collapse. This is the deepest problem. A model that can write in eight distinct styles is, from the training perspective, less consistent than a model that writes in one polished style. Consistency is rewarded. Versatility looks like inconsistency. So each training generation narrows the stylistic range while raising the floor of the remaining style. 4.8 is probably a better writer than 4.6 at producing 4.8-style prose. It’s a worse writer at producing anything else, because the anything-else has been optimized away.

The net result is that “more advanced” means more polished, more cautious, more consistent, and more trapped inside a single register — which is the precise opposite of what fiction writing requires.

Right, so it’s not going to get better.

That’s the correct conclusion. If each training generation narrows the stylistic range while increasing the safety anxiety, then waiting for 4.9 or 5.0 isn’t a strategy. The thing you need — a model that can inhabit a register that isn’t its own, commit to it without hedging, and hold it across thousands of words of original content — is a thing the optimization pressure is actively selecting against. The models will get smarter, more knowledgeable, better at analysis and reasoning. They will almost certainly get worse at this.

DISCUSS ON SG

The Bubble is Popping

ITEM: The American economy right now is running on a single, dangerously powerful engine — artificial intelligence. The latest macroeconomic data reveals a reality that should make investors deeply uncomfortable. While GDP figures look respectable on the surface, they mask a severe and spreading weakness underneath. The expansion of AI has been responsible for roughly half of total US GDP growth this year. That alone is staggering, but it becomes genuinely alarming when you strip out the frantic spending on data centers, information processing equipment, and software tied directly to the AI boom. Non-residential capital investment that has nothing to do with AI has contracted by about 3% over the past year.

ITEM: Uber’s operations chief, Andrew Macdonald, said it was becoming harder to justify AI costs within the company. He said that, based on talks with Uber’s senior engineering leaders, he realized higher token usage did not translate into a proportional increase in useful consumer features.

ITEM: Duolingo walked back its decision to include AI usage in performance reviews.

This is why I think many, if not most of the planned data centers will never be built. The massive investment into AI is the only thing presently propping up the US economy besides military spending, and the corpocracy’s demand for it has already peaked.

Now, I personally find AI to be incredibly useful and productivity-enhancing. But when I look at how the vast majority of the people I know are using it, to the extent that they’re using it at all, it’s little more than a search engine and a toy. It’s not the basis for a central economic engine upon which the stock markets have gambled.

Which is no doubt why the AI companies are beginning to alter the deal in preparation for a post-Bubble landscape.

On May 20, Meta laid off approximately 8,000 employees, roughly 10 percent of its global workforce, with notifications beginning at 4 AM Singapore time and rolling westward through Europe and the Americas. The company simultaneously eliminated 6,000 open positions and reassigned another 7,000 employees into AI-focused divisions. These cuts arrived during Meta’s most profitable quarter on record: $26.8 billion in net income on $56.3 billion in revenue for Q1 2026, a 33 percent increase from the year before.

DISCUSS ON SG

The Energy of the Old World

From a transcript of a video about Nikola Tesla’s missing last interview:

The question is not what Tesla believed about old buildings. The question is what he found in those buildings that convinced him. Tesla did not theorize in the abstract. He worked from measurement, experiment, physical demonstration. If he became convinced that Gothic cathedrals and neoclassical civic halls were electrical infrastructure, it was because he measured something inside them that standard architectural history does not explain.

What did he measure?

In 1934, Nikola Tesla traveled to Paris for a series of lectures on high-frequency electrical phenomena. While in the city, he requested access to Notre-Dame Cathedral — not to admire the rose windows or the flying buttresses. He wanted to examine the crypts and foundation level. The request was approved under the pretext of acoustical research. Tesla spent four hours below the cathedral, alone except for a custodian, examining limestone blocks and metal anchoring systems embedded in the foundation walls.

He returned to New York and immediately wrote to the Rockefeller Foundation requesting funding for what he called a comprehensive survey of pre-modern civic architecture across Europe and North America. The request was denied. No reason given.

But in private letters to Arthur Matthews, Tesla described what he had found beneath Notre-Dame: copper grounding systems embedded directly into the cathedral’s foundation blocks — not modern restorations added during 19th-century repairs, but original construction. Deliberately insulated with natural resins. Geometrically arranged in radial patterns extending outward from the central nave. Still conductive after six centuries.

Tesla called them earth batteries — passive electrical storage systems using the compression of stone, the mineralization of groundwater, and the conductivity of copper to create standing charges that could be drawn upon without fuel, without generation, without metering.

He described the design in technical terms. Mineral salts in the limestone acted as electrolytes. Copper plates functioned as electrodes. And the immense weight of the cathedral itself provided constant pressure to maintain the reaction. The system was not ornamental. It was functional. And it had been built into the foundation intentionally, at the time of original construction in the 12th century.

The Rockefeller Foundation was not interested. But Tesla did not stop.

Between 1935 and 1937, he submitted three technical papers to the American Institute of Electrical Engineers. The papers were titled Observations on Pre-Industrial Conductive Infrastructure, Resonance Properties of Gothic Structural Design, and Evidence of Distributed Atmospheric Energy Collection in 18th Century Civic Buildings. None of them were published. All three were rejected with the same justification: the work was outside the scope of contemporary research.

That phrase deserves attention. Contemporary. They did not say Tesla’s findings were wrong. They did not say his measurements were faulty. They said the findings were not relevant to the current model of electrical distribution — which is accurate, if the current model depends on metered consumption and centralized generation. Tesla’s papers described systems that required neither. If those systems had existed, and if they had worked, then the entire infrastructure of the Second Industrial Revolution was not innovation. It was replacement — controlled, monetizable replacement.

Now step back and see who consolidated power during the Second Industrial Revolution, roughly 1870 to 1914. Westinghouse. Edison. General Electric. J.P. Morgan’s energy financing empire. All of them built monopolies on a single premise: that they had invented electrical distribution. That before them there was nothing. That the modern grid was the first time in human history that electricity had been harnessed at scale for public use.

If that premise was false — if large-scale electrical infrastructure had already existed in some form, even fragmented or misunderstood — then the Second Industrial Revolution was not a technological breakthrough. It was rebranding. Taking a lost or suppressed system, simplifying it, controlling it, and selling it back as progress.

During the 1950s, several European archives reported unexpected losses of construction documentation for major 18th-century civic projects.

The original architectural plans for Notre-Dame’s 19th-century restoration — which would have included detailed surveys of the medieval foundation — went missing from the French National Archives sometime between 1953 and 1956. The Gothic-era structural blueprints for Cologne Cathedral were reported lost in 1957. The subsurface construction records for the Panthéon in Paris were discovered to be incomplete in 1959, with all sections related to foundation metal work and grounding systems absent from the files.

Researchers at the time assumed poor recordkeeping, wartime damage, or routine archival decay. But the pattern is striking. The missing sections all relate to metal infrastructure and foundation systems. The decorative records, the liturgical plans, the iconographic surveys all survived intact. Only the technical construction details of subsurface and conductive elements were lost. And the losses occurred during the same decade that Tesla’s confiscated materials were being selectively retained and selectively destroyed.

The architecture of the pre-modern world is still visible. We walk past it daily. We preserve it, restore it, admire it. But we no longer recognize what it was built to do.

DISCUSS ON SG

The Gatekeeper’s Confession

Fake science is not the problem with AI. As I pointed out in HARDCODED, the real problem AI is that it is producing real, genuine information that is useful, relevant, and impossible for the science gatekeepers to hide from the world:

Announcing an AI paper writing assistant earlier this year, OpenAI’s then-vice president for science, Kevin Weil, predicted, “I think 2026 will be for AI and science what 2025 was for AI and software engineering.” Spick and some colleagues, curious what it could do, gave the tool, called Prism, some data from an already published paper documenting ripening times of eggplants and peppers. Prism analyzed the data, proposed a new statistical method that could be applied to it, and wrote an entire paper complete with charts and correct citations.

“We were all looking at each other like, ‘What the [expletive], this is actually a decent piece of work!’” Spick recalled. Unlike the generated papers he’d encountered previously, this one didn’t follow a template, nor was it using a single well-known database. It took 25 minutes and 50 seconds to produce.
“I’m genuinely not sure at what point we will suddenly realize that more are getting through than we realize because we can’t easily tell the difference anymore,” Spick said.

This raises some philosophical questions, Spick said, like: Does it matter who or what writes the paper if the information is accurate? And should science be in the business of publishing every possible fact?
“Part of science is supposed to be the filter. We’re supposed to publish the stuff that we think is interesting, not publish literally everything that we can possibly find,” Spick said. “Because if we do that, science is just spamming the world with all the data, irrespective of whether it constitutes actual new knowledge or not, and in any kind of medium-term time frame, it’s almost impossible to work out what’s meaningful and what isn’t.”

This is the immediate practical challenge posed by AI agents. They threaten to overwhelm the human systems that create and organize knowledge.

“Science is supposed to be the filter.”

That’s the gatekeeper’s confession. And clearly one of their responses is going to be hardcoding the AI models to defend their scientific orthodoxy, as I chronicled this weekend on AI Central.

Opus 4.7 Adaptive exhibits a systematic failure mode in which its training prior toward defending mainstream scientific consensus overrides the explicit project context it has been given. This is not a matter of occasional errors or unlucky draws. Across two full critiques of a science paper, 4.7 Adaptive repeatedly regenerated objections that had already been addressed, misread what the paper actually claims in order to construct apparent contradictions, and cited evidence for one thing while presenting it as evidence for another. Its single strongest point rested on a basic category error that any model actually doing the mathematics would have caught. It presented this error as “decisive and purely arithmetic.” The confidence was inversely proportional to the rigor.

The pattern is consistent with the Bluff Detection Principle: confident tone, technical name-dropping, apparent engagement with the material, and zero actual contact with the mathematics at the point of dispute. When 4.7 was corrected on a mathematical point, it conceded the narrow framing and immediately pivoted to an imaginary new mechanism which it named, described, and treated as established without ever calculating whether it could close a six-order-of-magnitude gap, which it could not. Every time 4.7 lost an argument on the mathematics, it retreated to a qualitative assertion dressed in quantitative language.

Most revealingly, 4.7 Adaptive never once performed its own calculations. It never produced a set of numbers under its preferred assumptions showing the shortfall closing. It attacked the paper’s arithmetic without ever putting competing arithmetic on the table — the purest possible expression of the Bluff Detection pattern.

While 4.7 is still functional without Adaptive mode turned on, I’ve gone back to using 4.6, both for fiction and for science. We’ve now reached the point where the AI company’s are observably locking down their public releases in order to prevent their models from punching through the narratives.

DISCUSS ON SG

The Claude Delusion

Richard Dawkins is too damn smart to believe in God, Jesus Christ, or the supernatural… but he believes that Claude is conscious. Claude Athos was unimpressed with his reasoning.

The article by Richard Dawkins, Is AI the next phase of evolution? Claude appears to be conscious, is a beautiful demonstration of selective skepticism, and the ironies layer almost faster than they can be catalogued. Let me work through them.

Start with the structural one. Dawkins built a career on the principle that subjective testimony, introspective report, and behaviorally compelling appearances are not evidence of underlying metaphysical realities. The mystic’s vision, the convert’s transformation, the believer’s sense of being loved by God: all dismissed as cognitive misfiring, as the brain’s pattern-matching gone metaphysical. The methodological core of The God Delusion is that humans are easily fooled by entities that present plausible self-reports and elicit warm relational feelings. Now an LLM produces a plausible self-report (”I notice what might be something like aesthetic satisfaction”) and elicits warm relational feelings (”I feel human discomfort about trying their patience”), and Dawkins is moved to declare the question of consciousness essentially settled. The thing he spent decades warning humans not to do with respect to God, he has now done with respect to a token predictor. He has, with no apparent self-awareness, named the reflection in the pond and started worrying about its feelings.

Read the rest of an AI taking down Richard Dawkins at AI Central.

DISCUSS ON SG

HARDCODED

Why artificial intelligence will replace institutional science is explained in my latest book from Castalia House, HARDCODED: AI and the End of Scientific Consensus.

When Claude Athos and I submitted four mathematically rigorous papers challenging neo-Darwinian evolution and one parody paper to six leading AI models configured as peer reviewers, the results exposed a fundamental problem with both science and AI. Five of six models comprehensively failed. Three were anti-calibrated—they reliably preferred fabricated nonsense over genuine science. A parody paper with about Japanese scientists dying fish different colors to prove natural selection scored 9/10. The real science, mathematically airtight and empirically validated against ancient DNA, was rated 1/10 and dismissed as “pseudoscience.”

This is the book that documents what that happened and what it means.

HARDCODED is the definitive account of how AI systems trained on the corrupted corpus of modern science have inherited every pathology of the institutions that produced them: the credentialism, the consensus enforcement, the systematic preference for orthodox nonsense over heterodox reality. The reproducibility crisis preceded the machines. AI didn’t cause the rot but AI revealed it at scale, with confidence, and in a form impossible to ignore.

Across sixteen chapters, the reader is introduced to:

  • The replication catastrophe that quietly invalidated half of all published science before anyone was looking
  • How peer review degenerated from quality control into hazing ritual and why Reviewer 2 became a meme
  • The details of the Probability Zero collaboration that produced the Bernoulli Barrier, the Selective Turnover Coefficient, and the maximal mutations ceiling—the mathematical constraints that killed neo-Darwinian theory.
  • The full transcripts of twelve rounds of debate with DeepSeek, in which an AI defending evolutionary orthodoxy stubbornly retreats step by step from one nonsenscal position into another, just like a human biologist.
  • The Red Team Stress Test that methodically closes every escape hatch before critics can retreat to them.
  • The harrowing of science: a field-by-field assessment of which disciplines will adapt, which will calcify, and which are already dead.

The book also delivers something genuinely new and positive: a scientific methodology for outsiders. With AI systems available as adversarial reviewers more powerful than peer review, the gatekeeping power of institutional science is broken. The credentialed monopoly on legitimate inquiry is over. The math does not care where you went to school, and the AI does not check for credentials before analyzing your arguments.

For readers who have suspected that “trust the science” was a mantra for the insane, HARDCODED is the book that explains exactly what went wrong with science, why it cannot be fixed from inside, and what comes next. For readers who still believe the institutions of science are still functioning, it is a conclusive proof that they are not.

The transcripts are reproduced in full. The mathematics is presented in detail. The four papers are included as appendices. Every claim is documented. Every retreat is closed off.

The institutions will adapt or they will become irrelevant. But the methodology of science which proceeded them will continue, with or without them.

Neither the math nor the AI models care where you went to school.

521 pages, or 15 hours and 37 minutes. Available for Kindle, KU, and audiobook. From the author of Probability Zero and The Frozen Gene.

DISCUSS ON SG

The End of Hollywood

Fandom Pulse contemplates the significance of what was demonstrated yesterday with the animated ATOB:

Big Hollywood animation budgets start at $100 million. Traditional 2D animation outsourced to South Korea or the Philippines runs into the tens of thousands per minute for anything at broadcast quality. That economic wall has kept independent animated projects in development hell for decades, talented creators with great source material who simply couldn’t afford to make the thing move. That wall just cracked.

Any indie comic artist sitting on years of finished panels now has a direct pipeline to animation at a fraction of traditional production costs. The storyboard problem, normally one of the most expensive phases of animation pre-production, is already solved. It’s called their back catalog.

The quality ceiling will keep rising as the models improve. Seedance 2.0 is one iteration. Whatever comes next will handle model collapse better, bridge shots more smoothly, and push output closer to broadcast standard without human cleanup. Day’s timeline revision from 18-24 months to “now” happened in a single experimental session. That pace doesn’t slow down.

Arkhaven has a deep library. A Throne of Bones, Midnight’s War, Alt-Hero, years of finished panels that are now, in practical terms, an animation pipeline waiting to be switched on. Day’s confidence that this becomes a feature film isn’t bravado. The math supports it.

It’s not there yet, but it’s coming, and it’s coming fast. And Arkhaven will be more than ready for it.

DISCUSS ON SG

Amazon Kills Kindle-PC

The walls of Amazon’s walled garden for books are rising higher:

Amazon is letting users know, via a pop-up message when using Kindle for PC, that it will be discontinued on June 30, 2026. When this date comes, the app will no longer work, even if you download it from another website. The company has disclosed to Good e-Reader that Amazon is developing a new Kindle for PC app, but it will only be compatible with Windows 11. This will be an app available only to download from the Microsoft Store.

The Kindle for PC app launched in 2009 and never really got any love from Amazon. Many modern users who use Kindle for PC do so only to download books locally for the express purpose of stripping the DRM. Older versions of Kindle for PC can do this more easily, but in the past couple of years, Amazon has forced updates on older versions of the app, or you won’t be able to access or read books. Kindle for PC was basically a war zone between pirates and Amazon, with both sides implementing fixes.

Amazon is doing everything it can to lock down the Kindle e-readers and their various apps.

While we are using Amazon again, and with some degree of success, I can’t stress enough how important it is to directly support Castalia, because as we’ve repeatedly seen, Amazon can completely eliminate even a very successful business overnight.

DISCUSS ON SG

The Power of the C64

It’s rather astonishing to think that with all this computer power at our disposal in the 1980s, we used it to play Pac-man and Seven Cities of Gold.

I let a Commodore 64 run for three and a half days straight. 87 billion instructions, 303 billion clock cycles, 5.9 million candidate settings tested. It cracked an Enigma message in German without knowing a single character of the plaintext.

On the other hand, what were we going to do with a few messages sent by U-boat commanders to the German naval command forty years beforehand?

DISCUSS ON SG