The biggest things I've struggled with are models having taste. For spec writing I was having a lot of issues with them making statements that were interpretable in a superposition of ways, eg "we'll do XYZ with entities that support and need it" when there's 3 possible entities and the model hand-waved at exactly the wrong tokens.

I added AGENTS.md guidance + memories to be unambiguous (with short but good examples) and "no coined shorthand". Now I'm getting a marked increase in specificity, but it's places that don't matter (claude explaining existing code to itself). I'm having difficulty controlling the spew of new text, but feature writing/research still gets fuzzy and lazy around the difficult underspecified aspects of the problem/feature.

Do I just keep dumping examples into reviewer subagent context and have them rewrite and simplify the research subagent spew?

I repeatedly have "Risks and gotchas" sections have a whole paragraph dedicated to things that don't matter, and then a single sentence bullet point that's actually a huge problem when I dig into it.

Do I have a generic hook to make subagents to review bullet points and size them commensurate with impact? If I tell them to "have taste", do they have taste?

I'm just trying to get good code done. I hate these things.

In your workflow, who implements the review feedback - the review subagent or the code-writing-subagent?

Do have a baseline styleguide (like Google's Go style guide) for the review subagents, or is it entirely the subjective things and specific corrections? I remember 6 months ago it seemed like piling general "good taste" code advice into AGENTS.md was considered bad.

Do you move between harnesses or have you gone all in on claude? I've bounced between claude/codex/omp, maybe to my detriment.

To be clear, SotA models/harnesses are really good at making things that work one-shot, and their code golf and debugging game is insane.

When I go to implement, even with a good spec, I end up with code that's 80% of the way there in a fractal manner. The modular decomposition is 80% of the way to good code. The function decomposition is 80% of the way there. The computation structure and variable naming within functions is 80% of the way there. I can walk the AI through it and address each level of issues, but it's tedious as hell, and not clearly faster than doing it myself in some cases.

This is with omp/claude/codex, with Fable 5.1/Opus 5.5/6 Astra. The Chinese models do better at staying coherent, but they're a little less smart IME. I've tried many permutations of "fable planning, opus sub-agent implementation"; "pass a branch back and forth between opus and astra, as reviewers and feedback implementers". I haven't tried the "software factory with an architect and 2 juniors" thing. I haven't tried AGENTS.md beyond the /i-have-adhd, iso-24495 (lol at the Opus 5 induced PTSD), "No cleft constructions or Latinate absolutes" and general project orientation. Model character seems to change too fast to make agents worth it, and I see stuff about skills and excessive AGENTS reducing model capabilities.

Am I holding it wrong?

You might as well say "the obsession with sex has always held humanity back". Maybe, but it's complicated...

He's a paleontologist and the talk starts with some pretty interesting statistical analysis aiming to estimate T rex clutch sizes (66 eggs) and hatchling weights (1.5kg) as well as T rex parental care. It ends with looking at evolutionary history trends in increasing parental investment in radically different animals.

He talks a little slow, so a little playback speed bump won't go amiss.

He's a paleontologist and the talk starts with some pretty interesting statistical analysis aiming to estimate T rex clutch sizes (66 eggs) and hatchling weights (1.5kg) as well as T rex parental care. It ends with looking at evolutionary history trends in increasing parental investment in radically different animals.

He talks a little slow, so a little playback speed bump won't go amiss.

If employees can be persuaded to move themselves + systems to a board controlled chat instance, that's an angle, but Mullenweg has stronger cards if he's liked by employees.

Edit: I take that back, [1] this guy got prison time. Do Chamath next.

[1] https://www.justice.gov/usao-sdny/pr/former-ceo-special-purp...

I suspect the nitpicking helps engagement metrics and it doesn't harm RLVF outcomes, so there's just not good signal against it.

Since they're clearly a warhammer 40k fan, a recent discovery of mine for mini painting: artist gouache or artist watercolor are awesome for pin-lining and recess shading the way oil paint or enamels are often used for those tasks, but only water is needed instead of white spirit. And they're indefinitely "editable" with a damp brush / qtip. The only downside is that you need to varnish to seal the watercolor before painting over it. A black and a burnt umber alone will get you really far. Make sure the gouache or watercolor use a gum arabic binder.

Demo of the technique from 3:30 - 5:20 (strong Italian accent warning, closed captions help me): https://www.youtube.com/watch?v=7vHDq3NJj4E

In a cable leaked by WikiLeaks in 2010, Amos Yadlin, former general of the Israeli Air Force, said in 2007 that Israel would be "happy" if Hamas take over Gaza and regarded it as a positive step, in order to make Gaza to be treated as a hostile state.

The details of money flow are complicated, but there are credible rumors of Israel funding Hamas via Quatar. There's a chance Hamas would have lost leadership to the Palestinian Authority had Israel not supported it during that era.

https://www.jpost.com/israel-hamas-war/article-799792

Former Mossad Chief admits to government funding for Hamas via Qatar, calls policy a mistake

An observation from a year ago: people on AI text RPG subreddits saying they had a lot of difficulty with Gemini role playing ambiguous characters and that almost always the characters would betray them, or misread the human role player's motives as negative. Someone pointed out this paper [1] that showed significant differences between the different LLMs strategic behavior playing iterated prisoner's dilemma where Gemini had exactly that behavior, and speculatively that difference was emerging in RPG character behavior.

I wish they'd used bigger models (they used gemini-2.5-flash, gpt-4o-mini, claude-3-haiku-20240307). From the abstract: Our results show that LLMs are highly competitive, consistently surviving and sometimes even proliferating in these complex ecosystems. Furthermore, they exhibit distinctive and persistent "strategic fingerprints": Google's Gemini models proved strategically ruthless, exploiting cooperative opponents and retaliating against defectors, while OpenAI's models remained highly cooperative, a trait that proved catastrophic in hostile environments. ... Later, we see that Anthropic’s Claude is more cooperative still, but nonetheless outperforms OpenAI head-to-head

Obviously Opus 5 is wildly different than Haiku 3 but I'd expect Opus 5's fundamental suspicion of user intent, and anti-sycophancy via necessarily finding something to nitpick, is still present in your styleguided output.

[1] https://arxiv.org/abs/2507.02618

It doesn't clearly state that it was part of the room block at Benn's initial booking, and it implies that the booking was not associated with Flock when the initial booking was made.

My use of it comes from early 2010s era paleo-food community observations about hunter gatherers having less crowded jaws than agriculturalist and modern peoples, hypothesized to be due to both micronutrient intake and also eating tougher foods. Cramped jaws and facial structures lead to both malocclusion and also nasal breathing issues, and mouth breathing also reinforces those pathological facial changes. Granted, this stuff is from sources that I've heard wildly varying opinions on (Weston A Price; Mike Mew). Jaw and facial structure continues to change into adulthood. I needed extensive orthodontics in my youth and have struggled with sleep apnea in adulthood. This broad set of hypotheses made sense to me, and chewing mastic gum was the easiest way to get a full dose of "chew tougher food" in my life. I try to snack on raw carrots too.

Speaking to attractiveness, there's an uncomfortable photo of identical twins in one of the Weston Price books, one given orthodontic treatment that widened their facial structure and one was not. The one with the widened facial structure was more attractive. The difference wasn't gigantic, but it wasn't small either.