Mark Wotton

new jynpost jyn.dev/the-terminal...

not really. gonna be a hit.

Hoi An really copped the brunt /cc @timwoolworth.bsky.social

i.imgflip.com/ab36o5.jpg

@arachno.capital your mortal enemy! bsky.app/profile/fain...

Feels a bit infinite-game-ish?

Especially tricky if they did distill from ChatGPT as seems to be the consensus

I think it was mostly system prompt tuning? I did look into a while ago, my information might be out of date, but I think it’d be hard to feed an LLM a fully sanitized version of the world

Seems like it. You get a commissar spin about the Uyghurs but it won’t discuss Tienanmen Square at all

you seem to be able to get away with it if it's really targeted, like DeepSeek's Tienanmen Square refusals, but you can't shift the whole worldview and expect coherency.

It’s going to be very funny to see grok’s coding scores tank after he lobotomises it.

Scoped vs unscoped keeps paying off in unexpected places like nursery-based structured concurrency

S tier

I meant the movie but both, I guess

It is a fantasy, after all.

Heavy Vegetable are incredible, listen to them.

just failed at napping, a feat routinely managed by newborns minutes after birth

Necromancy but make it Python. “I’m not dead yet!”

I think there’s some necessary hypocrisy to any kind of attempt at self improvement

Seven Charts Showing How Canada/Mexico Tariffs Would Harm the US Auto Industry (and American Car Buyers) Particular industries and individuals would disproportionately suffer from Trump’s proposed 25 percent tariffs on Canada and Mexico. In this blog post, we focus on the US automotive industry and provi...

Fair. I had to laugh a bit at this www.cato.org/blog/seven-c... - how often does the Cato Institute have to tell a sitting Republican president he doesn’t understand basic economics?

Hard to blame Canada for responding to clear provocation though

21. A pyramid has 54 edges, how many face(s) does this pyramid have?

starting to think they’re smuggling some Lovecraftian geometry into my kid’s maths comp

More that I had some implicit idea of how much complexity, cleverness, moving pieces would be required to get the results we are seeing, and I was just wrong - scaling the obvious thing and adding the obvious buffs like chain of thought seems to have worked way better than expected

Definitely feels like some kind of complexity conservation law is being broken

I think a lot of people still have their heads in the sand, telling themselves stories about how counting the number of rs in words is a meaningful datapoint.

if false then absurdity

ah - and some numbers are going to be split points more often than others. yeah, that's annoying.

Kris why are you asking for extra homework