Internship etc. proposals miss that status quo job seeking is a parallel process for applicants. An internship model means each candidate can only “consider” one employer at a time. I think this is what Steve’s campfire proposal is trying to counter by making samples/internship outputs public?

I think the real solution is something like the Bar Exam. Apply the hard filter once, publicly, administered by a trusted neutral party. Anyone that gets through is assumed to be good enough, and interviews can assess only firm/team specifics

The real solution is stop proposing solutions like a bar exam. Software engineering evolves too quickly for anything but a work sample to work.

This is what Triplebyte tried to do, but unfortunately they failed in the market. Maybe they were ahead of their time and a similar concept would work now.

Yeah I wrote “by a third party” then thought of them and revised to “trusted neutral party”. Has to be someone with ~no financial interest in the candidates success

Per TFA Steve got hired by Amazon and then kicked off the acquisition of his former company.

Whatever clause in whatever contract is just the starting point for negotiations if things go south.

Option C is, yknow, pass law (and ratify amendments) s.t. the interpretive power of the court is less of a factor

Implicit in the human to human to AI conversational chain is the second person’s assumption that the first person didn’t think to ask AI.

The mere fact of asking another human a question (absent a strong pattern of behavior to the contrary) should be strong evidence the interlocutor wants a human answer! Sending an AI answer should have the same social valence as sending a lmgtfy link; appropriate for bad actors but a pretty insulting response to an earnest question.

I don't know if it counts as a 'pixel' font, but https://fsd.it/shop/fonts/pragmatapro/ has hand-drawn bitmaps for a huge swath of unicode (and hand-hinting for aliased rendering IIRC?)

It's not quite as overtly retro, but it's a great functional font, and a great art object besides (at least that's how I justified the price!_

I really want to like Pragmata Pro, but for such a price I'd like to see a couple more examples of the font used for programming! The website only has three tiny examples for Haskell, Agda, and APL (!). The tester from MonoLisa should be the benchmark here

DeepSeek claims to have trained on something like 2k H800, this is ~0.5k GH200 … it’s not nothing. Sure they’re not going to _serve_ it at scale, but that’s not the point?

Also the line between “finetuning a base model” and “man this is a real good initialization” gets pretty blurry at scale.

Altogether a pretty presumptuous take.

Because greater Los Angeles is the USA's (post-)WWII aerospace hub disguised as a megacity and cultural production center? All sorts of folks spent the 40s-00s (scientifically) blowing stuff up in the hills, and manufacturing the resulting products down in the basin and points south. Those businesses needed labor, which needed nearby housing, and here we are.

That's... not really a reasonable characterization of LA's urban growth patterns. To begin with, Hollywood quite clearly predates the aerospace buildout in the 40's and 50's. It was an oil production and refining hub before that, and an agricultural shipping center even before the dust bowl.

This particular neighborhood in Orange County certainly looks aerospacey, but I bet the Disney-centered service workers in Anaheim made up just as much of the population as the industrial folks.

Big cities are big for a bunch of reasons, basically. There are no simple answers at this scale.

This part of the OC is very defense heavy.

I suspect subscription limits are quite a bit higher than the equivalent tokens their dollar cost could purchase. I similarly feel like I can get a lot done with a $20/mo Claude Pro subscriptions, but also can easily spend $10-20/day at API pricing with similar usage.

Yep. I get $6k - $8k worth of tokens (at api rates) using the $200 max subscription.

I don't understand why people are using the API pricing instead of the Pro/Max subscriptions? What am I missing?

Enterprise customers don't get that option. But also if you want a fully custom harness, you also don't get that option.

Personally I prefer the API pricing because I feel like I'm not going to get rug pulled on my work. When it comes to personal stuff, I use the shit out of my sub, but it's not making me money.

I’ve made the same argument On Here. Paying the full price (should!) make you consider you usage, pick the right model, delegate to cheaper/local providers, …. It makes you use the models the way they’re going to be used after the subsidy ends.

Depends on what you're optimizing for. I'd hope that "after the subsidy ends", the "cheaper/local providers" will be at the level of at least current SOTA models. If not, then there's hardly a point using them anyway; if yes, then by sticking to subscription workflow you'll be learning the very workflow you'll be using "after subsidy ends".

Either way, I don't see much point of intentional austerity in times of extreme growth. There will be time for austerity once the growth ends.

Anthropic is forcing large enterprises onto api billing instead of subscriptions.

Because with Max subscriptions, you have to use the Claude Agent SDK, which is basically running Claude Code underneath. You don't get to use the chat/Messages APIs with personal subscriptions, for that you need the API pricing.

Terms of service prohibit subscriptions for employees of companies bigger than X people. I suppose they could all sign up as individuals and try to get away with it but presumably that would look pretty obvious with a tiny bit of analytics.

No 5-hourly or weekly limit.

Can verify that I've gotten about $400 worth of tokens from my $20 sub.

Now that sounds like a business I’d like to invest in! When’s that Anthropic IPO anyway?

It’s hilariously persistent in this, esp. for anything even slightly divergent from the beaten path. Discount everything the AI box says about emacs to zero.