Hacker News new | past | comments | ask | show | jobs | submit | dgroshev's comments login

Yes! Since any function returning a float might need to return NaN, NaNs are represented as a special kind of float, with most bits unused.

Cool. Is that why NaN isn't equal to anything? Not even when comparing a NaN variable to itself? (I assume it's because you can't have some NaN values be equal, and others not)

$2t is the price tag of the entire program [1], from its conception to the jets retiring decades from now (note that programs tend to get extended, so the retirement year is probably too pessimistic). For comparison, the US is still producing F-16s, with the first F-16 produced in 1976.

F-35s has a much lower crash rate than F-16s during their first 20 years in service [2] and just recently passed 1 million flight hours [3]. The program has its problems, but it resulted in an incredibly capable fighter plane. Practically every US ally that has access to the F-35 run their evaluations and concluded that the F-35 is the best option (eg [4], quote: "F-35A offers highest overall benefit at lowest cost by far").

[1]: https://www.gao.gov/blog/f-35-will-now-exceed-2-trillion-mil...

[2]: https://www.aerotime.aero/articles/are-new-fighter-jets-more...

[3]: https://theaviationist.com/2025/03/04/f-35-one-million-fligh...

[4]: https://www.admin.ch/gov/en/start/documentation/media-releas...


> As Ukraine shows, peer conflicts are won by other means, not stealth air superiority

I don't think you can conclude that when neither of the belligerents has the capability. As Gulf War shows, training and capabilities (including stealth) do enable SEAD/DEAD to an extent that unlocks air superiority.


Perhaps you're not aware that this happened? https://en.wikipedia.org/wiki/Iran%E2%80%93U.S._RQ-170_incid...

USA has air superiority only against 3rd world countries, and even then, history shows that air superiority has never won any war.


I don't understand what that link is supposed to prove. A single F-117A getting shot down in Yugoslavia due to complacency and chance doesn't negate Nighthawks bombing SAM defended areas with impunity. Iraq had a strong and integrated AD network for the time.

Air superiority alone doesn't, but it's a massive force multiplier.


I sent a link about Iran, you comment about Jugoslavia.

You don't have to read my link, but you can also skip on making completely unrelated comments if you don't feel like doing the reading.


I read it and I know about that case. However, as I said I'm struggling to see your point.

I guessed you meant that that one case proves something about air superiority or Iran having an advantage over USAF, so I responded with a historical parallel.


I talk about peer conflict, which Gulf war wasn't. Old soviet tech, poorly trained soldiers with very low morale doesn't make them anyhow a peer to US army of that era. It was just a variant of that shooting goat herders, defenseless even against Apache choppers who have 0 stealth and fly low & slow.


It's not impossible, it's just hard to be better than libtorrent, otherwise a new client only makes sense as an exercise.

Arvid Norberg is hardly a "beginner developer", yet look how hard it is to make a performant implementation of uTP: https://github.com/arvidn/libtorrent/issues/3542


Few more challenges:

- seeding a million active torrents (this is entirely possible and needed but drives a lot of architecture)

- a UI for a million active torrents

- be nice to trackers (keep-alive, batched reports; again, architecture gets tricky, iirc libtorrent can't do that still)

- correct pipelining

- congestion control on uTP

- I/O scheduling/batching

- testing all of the above

It is quite a lot.


Looks like we have same experience with torrents :-)


Yeah. I was working on [0] for a while (I might get back to it when I have more time) and implemented a chunk of the protocol with an eye for a more scalable libtorrent replacement, and it's not entirely trivial. Even things like the piece picker get trickier at scale.

[0]: https://dgroshev.com/blog/torrentdyne-encryption/


From experience, regulation as an explanation for EU startup competitiveness is overused so much it's almost meaningless. Can you point out specific laws that you consider existential for startups?

What I find matters way, way more is two factors:

- Concentration of capital. The US has an ecosystem of wealthy people that want to put their money somewhere. This is good for startups, but can also backfire as we can see in the news.

- Unified market. EU is not a single market, it's several dozen markets with different regulations, different languages, and different cultures. You can't sell the same B2C product with the same marketing in Germany, Spain, and Sweden as easily as you can in California, Ohio, and Texas.



First, your last point answers your first question: a non-unified market is an implicit result of too many regulations. Harmonizing them would create a more unified market. The US is efficient because it is more homogenous. That efficiency is one of the things that leads to capital formation.

So, I think you have causation backwards. Capital formation doesn't really happen because it's too difficult to build and grow things in Europe.

Look at tech in Silicon Valley - all that capital formation is years worth of growth and reinvestment.

Look at oil & gas Texas - again, all that capital comes from years of growth and reinvestment.

And what you learn in silicon valley you can generally apply to starting a company in Austin Texas. What would happen if Mercedes wanted to move it's company (HQ and all) to Spain? How much would it have to relearn from a regulatory perspective?


This is nonsense though. What does "better" mean in this case? A judge is not a black box with an input (the case) and an output (the verdict), the entire point of having a judge is to have empathy, conscience, and personal responsibility built into the system.

It's a blind spot that too many people have because we take those qualities for granted. LLMs unbundle them, so we need to start recognising the inherent value of humans, fast. I wrote a few words about it here: https://dgroshev.com/blog/feel-bad/


In civil law countries verdicts should not depend on an individual judge's feefees.


They should depend on individual case specifics, which is exactly what PAC learning is bad at.


A judgement is by necessity subjective. Have a look at some sentencing remarks, for example this, point 19 forward: https://www.judiciary.uk/wp-content/uploads/2021/09/Wayne-Co...

Someone has to make a call. The weight of the call rests on the person's life experience, their understanding of the context and the cost to the society, their empathy to both the defendant and the accused, and their conscience. Treating it as a black box exercise misses the point completely.


That's not about nationality though. That PR is about (re)enabling OpenTofu to work more smoothly with Russian SaaSes, which are either already sanctioned or are likely to be sanctioned.

Everything is political, being "apolitical" is a political choice. You can't escape politics.



Previously (includes informed critique of the paper): https://news.ycombinator.com/item?id=42495703



Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: