fsloth

This article from Popular Mechanics from 1959 -almost 70 years ago - sounds eerily similar as the pitch decks of so many "AI" CAD startups.

Some problems appear almost as industrial invariants.

I’m not sure we are at the top. Swarmed lower tier models appear to scale up in competence. So something like a Fable swarm running at 100x the speed might provide surprises.

But ”top of the curve reached” feels like the likelier scenario.

I don’t think average user _needs_ to solve frontier challenges. ”Call to Jane”, ”turn on the lights” and ”what’s the weather this afternoon” is more like it I would guess.

Ofc if the model has some critical bugs that’s another matter.

Weird - I would have expected Disney of all streamers to respect IP better given how dependent they are on IP.

To me "explosive" means it comes unannounced and evacuates bowels within few seconds, ovepowering any sphincter resistance.

If that's your normal poop maybe seek medical advice.

Whoa the acceleration from the start surprised me. It’s huge and it just … zipped off the pad like a bullet (ok not real bullet but still).

Hmm I never could get the DWARF debug information to hook up properly, must give this a try next time I need to return to the topic.

If you don't mind - what's your specific setup when debugging Sokol, I might start from replicating that workflow?

" it's cross-platform"

Nothing in real time graphics _actually_ is cross platform. You are always programming a vendor hardware.

"Cross platform" really has no value in graphics programming, honestly, when discussing the lowest level programmable api.

You always need to explicitly say which platforms you target, and then test on those platforms.

The main questions should be "on which platform will my users run it" and "can I debug and test this".

So if you are starting out - pick out the easiest api for you. Don't go looking for universality as a value on it's owm (if you want universal rendering, choose CPU rendering. if you want high perf - choose your GPU platform(s) explicitly).

"The "universal" nature of WebGPU with the "web" focus feels a lot like the graphics API equivalent of WASM"

Having shipped non-trivial WASM code in two projects That's not a positive thing. WASM is stuck in zone between living and the dead - just good enough (barely) to merit creating production code when you need the perf in browser but not good enough to register as industrial quality tool as it's not possible really to debug a WASM application.

If you really, really need the perf in your browser app, WASM may be the only option, but it's not really a good one.

TBH to me she reads as high performing idealist, not naive. Different things, may give similar signals.

"insufferable and very naive."

IMHO she is just not from US. US culture _is_ totally foreign to anyone not from there.

Read the part where she describes her interactions with New Zealand prime minister and you will get it. That part to me feels natural (I'm from Finland). Everything else in the book reads like absurdist black comedy to me and not how I would expect people to behave normally.

If you loved Catch-22 you will love ”Careless people”! I warmly recommend a read. Not to support Sarah necessarily - just because the book is great! It’s not really a spill-the-tea as much absurdist escapade in corporate america.

Ok the details on Zuckerberg and ”Lean In” Sandberg _are_ bizarre. Which makes it a super fun read.

”Web compilers are dime-a-dozen”

Oh didn’t know that! What is the best for ”compiles to wasm in browser” featureset?

I came to the comments exactly for this ("wait I thought we 'knew' this already").

I'm so happy we have HN with likeminded people and no noise.

Hiit training can use any type of exercise that allows quick change of pace.

The routine in Tabata’s original HIIT work was just cycling.

Crossfit afaik is a more complex affair than just cycling.

So is the risk of injury due to specific movement or the interval quality of the movement?

"hiit also causes a lot of damage in a lot of ways"

Oh! I didn't know about this. Are there any references you could quote?

There is no bad publicity! I wonder if OpenAI explicilty asked for this.

I could read this as black comedy.

"Take him to a psychologist" was standard playbook in soviet block to silence dissent. Everyone knows of the gulag archipelago but just declaring someone insane was part of the toolkit.

No, read the post. This is directly copying another product. "Vibe coding" which may or may not have happened is not the point here.

If you need this stuff in production I recommend getting a good computer graphics textbook.

These cheat sheets are convenient _if you already know what you are doing_ and _are confident the cheat sheet uses same presentation as your problem domain_.

That said I'm not sure if there is a nice book that would be both exact and practical at the same time.

"The boss" is not who says what is good enough. Ultimately it's the customer. In many industries it seems good enough is not very good.

Then there are industries where the customer complains if code is slow. They will actually hire expensive consultants to analyze and benchmark the code. And while the consultants likely are not more talented than inhouse staff, now you have both sides very interested at looking at the problem from engineering perspective.

In this case "good" includes performance.

On the topic of public engineering projects overall, I'm wondering are there better books on this topic than Robert Caro's "Power broker"?

Which is basically "Soul of a new machine" for municipalities with all the political mess this implies. How do you get stuff done - and at what price.

I guess they can say "Made in China, designed by Apple in California" in the packaging but at least they still take pride in the design. With AI it sounds you are disavowing also the authorship of the design.

"Founding cannot be a commodity. If it is, you have no moat or point, meaning you instantly collapse again, because you are an interchangeable commodity."

IMHO you still need to find the product and PMF

There are bunch of books startup world recommends which sort of all start from the principle of product, users, traction.

This is sort of scaffolding around that. It's not entirely insane to try to formalize this process - there already are books that do this (Bill Aulet, Disciplined entrepreneurship).

"nor does it make sense for society to have people founding businesses at a scale"

Maybe not at scale of moving lawns but I'm pretty sure the world is full of nichces that still lack specific software offering or where options of software offerings are limited.

This is like "Uber for logging" or "time reservation system for cat dentists" level of "take existing product category and apply to a domain you know".

So not every cat dentist needs to found a cat dentist time reservation app but I'm sure there are niches withing niches with business opportunities awaiting.

This! The only way to get to a stable system at least with c/c++ source, where you can hunt bugs, is to have a fairly large unit test coverage. When something fails - add that as test case; run ctest - pray that this is discoverable with tests.

So wasm is a really strange compilation target for systems programming languages.

I mean there _are_ ways to debug it in a browser but they sort of suck.

"has already replaced several software engineers. "

I think this is incorrect, unless your friends employed software engineers, or would have employed software engineers.

"I think most of us are happy to believe that most companies simply have bad leadership"

I don't believe this for a second.

I don't know if this makes me the minority or not though!

It trivializes forces influencing large public companies.

Yes - there can be good and bad leadership - and bad leadership is just bad.

But good leadership can be totally helpless in a public company.

Not recognizing this is a huge gap.

The other forces that incluence a company:

- Board

- Shareholders (via board)

- Banks. Lots of companies have loans. The banks generally have the companies by the balls and can dictate many things when things go south - either due to leaderhip or just prevailing market winds

As an example of shareholder/board direct and rapid influence: an activist purchases shares. Installs board member. Causes rapid change in one or more aspects of business structure or strategy to support _their_ portfolio strategy (of course aligned with interests of other shareholders).

etc.

I wish it was so as well but

"When McKinsey Comes To Town" by Bogdanich and Forsythe

documents exactly standard tactics such as these in e.g. insurance.

"the crazy part to me is how blatant the executives of bricks and minifigs are in saying go ahead and try to sue us, we’ll drag this out "

To my experience this is a common strategy in disputes when the corporate party has people who operate as uncivilized brutes. I think it's part of the McKinseyfication of companies - profits at all cost - and here's the playbook.

My personal experience is from private parking control. Rather than be professional about my reclamation, their first response was "only criminals dispute these and we win all the court cases".

So I think trying to be imposing and villanous to scare the other non-corporate party to back off is a common global corporate playbook in situations in matters where companies enter contractual complex space with individuals.

Well there are already offerings like githits (https://news.ycombinator.com/item?id=46105112) that sort of promise optimize bang-per-buck of inference