We just don't make photos like this anymore. This is the X-15A-2 with external tanks on Balls 8 on October 3, 1967

We need a new word for token lust: the insatiable desire to use more tokens

Tried building a rewriter like this with an ensemble of models stitched together with different cultural personas, tool calls to search engines, and insertion of words from the Actor's Thesaurus. I got to ~23% on Pangram with a lot of work. 100% is very impressive.

In retrospect, I should have bought a lot more $CRM. Software isn't dead!

Hey @postmarkapp are you having an outage again? Really frustrating...

Well, hopefully I timed that right. $AAOI

did u know Codex has a separate rate limit pool for GPT-5.3-Codex-Spark?

this bad boy can fit so many tmuxes in it

Pro tip - alias `rm -rf` to `trash` on MacOS (using trash-cli by @sindresorhus) so that when Opus decides to nuke your files you can recover them

Denton Cooley reportedly answered in the affirmative when asked if he considered himself to be the best heart surgeon in the world during a trial. "Don't you think that's being rather immodest?” the lawyer replied. "Perhaps," Cooley responded. "But remember I'm under oath."

I'm sure everyone is doing this, but here it is in Lean: github.com/sebmellen/jacobia…

INCOMPREHENSIBLY AMAZING DEFENSE from England!!!! WOW! 🏴󠁧󠁢󠁥󠁮󠁧󠁿

America is somehow very good at consistently finding ‘the one thing’ that really matters and excelling in that particular thing at the correct time (roughly in order — republican democracy, territorial acquisition, industrialization, capital markets, military supremacy, cultural revolution, high finance, semiconductors, internet, software, space, AI, etc.) The downside of this is that it is very difficult to excel in multiple things at once. Attention follows what succeeds and what does not succeed, or is second-best, is left relatively forgotten and uncared for. Whether that be people/companies/ideas/etc.

whether crypto or AI, it's all tokenomics

Gemini 3.5 Flash High x Fast in @antigravity is insanely good. WOW! I think this is the fastest model I've used in a large codebase, with generally better accuracy than GPT 5.5 fast.

Even God had to rest on the seventh day. You think you’re not working hard enough? Have some mercy on yourself.

Porco Rosso edit I made a while back (turn on sound!)

just bought hantavir.us — what do I put up?

Do you think any cheesy Apple board members said "It's time to Ternus around"