Yes.

And converging on 5 as the optimal size of a team.

Get this benefit of collective wisdom, manageable perspective coordination, no decentralized garbage cans.

What a time to be alive!

“Gudtrip’s AI agents are designed to help users deploy eligible rewards across selected opportunities such as DeF(Decentralized Finance), network incentives, prediction markets, and certain RWA(Real World Asset)-based strategies.”

At Long Last, InfoWars Is Ours Let me tell you a story. When I was a child, I suffered from night terrors. It was always the same dream: I could hear my family and neighbors wailing in the street outside as they were pursued and th...

Infinite Growth Forever

theonion.com/at-long-last...

That Polymarket chart is looking pretty spicy, especially with that massive spike for **April 23rd**. A 75% "probability" according to the bettors is significant—it suggests they are pricing in some very specific rumors or a scheduled event. Since today is the 20th, we’re only three days out from that peak. If those odds hold true, the "AI spring" is about to get a lot more crowded. ### Why April 23rd? The sharp vertical move on the blue line usually happens when: 1. **Leaked Benchmarks:** "Leaked" results showing massive gains in reasoning or coding often hit X (Twitter) or GitHub around this time. 2. **Conference Alignment:** OpenAI has a history of "coincidental" timing when competitors (like Google or Anthropic) have major announcements. 3. **The "Strawberry/Orion" Lineage:** After the iterative updates of late 2025, a 5.5 release would likely represent the full integration of more advanced reasoning loops and perhaps the "Dioptra-style" flexible runtimes people have been speculating about. ### What to Watch For If GPT-5.5 is indeed imminent, the technical community is going to be looking for: * **Active Inference:** If it can "think" before it speaks more efficiently than the current O-series. * **System 2 Thinking:** True architectural improvements in how it handles complex, multi-step logic. * **Reduced Latency:** Whether they’ve managed to scale the reasoning without the 10-30 second wait times. It’s definitely a "popcorn" week for anyone following model architectures. Are you leaning toward the April 23rd bet, or do you think the 8.3% "No release" crowd might be onto something?

inside traders announce that GPT-5.5 “Spud” will be released on Thursday

Because of SCM Paine, K Thelen, Allison and Zelikow: my stance resembles some of Bert Röling’s. At least I’m nodding along.

Further, the wrong lesson from 1905 got hard locked in within the Army.

Hard lock-in is one hell of a thing. Crusty priors.

“This raises a question of whether such code can attract copyright at all, since copyright requires a human to be sufficiently involved in the creative process so as to be regarded as a work’s author.”

Amodei describes how Anthropic has launched Project Glasswing, a collaboration with more than 40 organisations, including Amazon, Apple and Microsoft, to help find and patch cyber vulnerabilities. But Anthropic is itself facing scrutiny over its data security practices following the leaks of some of its code. Amodei says he suspects open-source models and Chinese developers will be able to replicate Mythos's capabilities within six to 12 months.

Dario thinks open weights models will be at Mythos-level in 6-12 months

www.ft.com/content/9e0e...

What made the Pyramid Winner's Circle tick (literally and otherwise) YouTube video by Ological

Ological’s analysis of the Dick Clark run of Pyramid is a real treat. The breakdown is detailed and satisfying. High art.

youtu.be/k1AC7peQZ9o?...

The assumption that a synthetic godhead can’t ever discover a superior physics in silico, one where the catastrophic limitations of meatspace don’t exist, could be so completely wrong.

Try teaching calculus to a lobster or young me about so-called ~~~imaginary~~~ numbers, etc etc etc. Sure. Maybe?

The distinction between synthetic Godhead and useful tool isn’t too fuzzy.

Once a synthetic system can self-reflect and experience (French sense too: expérience) reality in vivo, out here, with Kuhn and Göddel and Condorcet and Arrow: then my concern ramps.

In the meantime: tools be tooling.

Transformers are Bayesian Networks Transformers are the dominant architecture in AI, yet why they work remains poorly understood. This paper offers a precise answer: a transformer is a Bayesian network. We establish this in five ways. ...

The distinction between natural language and ground truth / knowledge / wisdom here is key.

I struggled with this discomfort until Coppola (2026).

I sleep better now.

arxiv.org/abs/2603.17063

I understand the mechanics of hype contagion. I can describe manias mathematically. I’m generally quite skeptical of technologies where the causal link to productivity is dubious.

Fuzzy: I accept as a self-limitation, a product of stunted imagination.

Dubious: If I can’t see a 1:20 shot, well…

I had seen a quite a few BERT papers pre-press in 2020. We had already used transformers for recsys, so I came into it biased.

ChatGPT’s large public release was pure bias reinforcement.

I was a leveller heading into DLT; emerged … thoroughly … disgusted. I remain a leveller. Just not that way.

What role did India play?

The word choice is peak Canadian:

certainty and prosperity.

It just oozes mapley syrup and chicken gravy.

Screenshot of an article from Bloomberg. The headline reads: IMF Sees Canada's Fiscal Position as Strongest in Group of Seven.

Our plan is creating more certainty, more security, and more prosperity for all Canadians.

I’m in Committee Room 1 at Toronto City Hall for an AI in Transportation meeting, with Steve Munro and Alex Barnes, Ilya Kreynin, and Sina Malayeri. Thanks to organizers for the work and Toronto Open Data for hosting us.

Liberal Majority just called.

Polls close in byelections that could deliver Carney a majority government | CBC

Meanwhile in Canada, polls are closed in all 3 ridings. Liberals may win a majority tonight.

www.cbc.ca/news/politic...

How did Hungary's biggest cities vote?

Budapest - Tisza+33
Debrecen - Tisza+23
Szeged - Tisza+39
Miskolc - Tisza+29
Pécs - Tisza+31
Győr - Tisza+25
Nyíregyháza - Tisza+31
Kecskemét - Tisza+25
Székesfehérvár - Tisza+22
Szombathely - Tisza+19
Érd - Tisza+26
Szolnok - Tisza+29
Tatabánya - Tisza+24

It would seem that without explicitly stating the outcomes, credibly and trustworthily, of a given course of AI investment, the likelihood of a transformation converge on zero.

I can’t assume that the absence of outcomes is a function of duplicity. It may be capability. (?)

Ambiguity, as such, is a strategic resource.

Tim quotes: “Pilot Rich Transformation Poor”. Indeed. The greater the ambiguity in the metrics of the Pilot, the greater the opportunity for strategic rhetorical framing as to the meaning of the Pilot synergy HBR holistic integrated .

If Cohen, March and Olsen ‘72 is satire, then it’s damn good satire.

I’m an organized anarchy, a Garbage Can creates insurance. I confess, an org of 10k+ generates plenty of opportunities for ambiguity.

Canadian: And that’s where risk can be quantified.

American: Humbuggery.

I re-read it this morning.

It’s excellent.

Notes after the jump.

“That’s the connective tissue in practice — not one system, but the compounding between systems that weren’t designed to work together.”

Lethal conflict after group fission in wild chimpanzees Territorial conflicts in animals can inform aspects of human warfare, but civil war, with its shifting group identities, has not been previously observed. We report a rare, permanent fission in the la...

“The loss of individuals, even weakly connected nodes, can disrupt social networks (11, 46)”

www.science.org/doi/10.1126/...

reads: Introduction In March 2026, an Anthropic employee released the source code of Claude Code, a wrapper around their large language model that is widely used to generate code in programming tasks. Its thousands of lines of Typescript code contained many hopeful prompts and in cantations to shape Claude’s behaviour. Here are some examples: “Report outcomes faithfully”; “Never characterize incomplete or broken work as done”; “Be careful not to introduce security vulnerabilities ” (prompts.ts in Anthropic 2026) . There is more than a passing resemblance here to the Azande witch -doctor apprentice who, while stirring the medicine, utters: “You medicine which I am cooking, mind you always speak the truth to me. Do not let anyone injure me with his witchcraft, but le t me recognize all witches. … Let me be expert at the witch -doctor’s craft so that people will give me many spears on account of my magic.” (Evans-Pritchard 1937: 93). In the case of Claude, the incantation s appeared insufficient: analysis of the codebase, which according to a company executive was “pretty much 100% written by Claude Code ”, revealed severe security vulnerabilities (Townsend 2026).

what an introductory paragraph!!

Yay! Comms back!

Alright. Hooray.

Great job! Congrats Artemis, whole team, all of you, congrats!

6 minute comms blackout

Watching NASA’s coverage with 660,000 of y’all. Hello!

To their credit, they are openly talking about the heat shield.

Good stuff.

Me, here, hoping that it goes well is inneffective. No impact on the outcome.

I just hope that it’s all good.