HN user

ohdeardear

-22 karma
Posts0
Comments15
View on HN
No posts found.

It's not just strange, but completely regarded, _assuming_ no more information, but since it's JPM, of course they know more, which probably make it a lot less regarded.

I would not be happy with this trade if I had any $JPM.

Their web-application is worth $200,000. The software of the training infrastructure is worth perhaps $2M; the inference infrastructure is worth perhaps $500,000. Their hardware is worth nothing in 3 years. Their B2B relations are worth perhaps $5M. The "data" they have is worth nothing in 5 years, because a sufficiently smart model will be able to learn without human feedback.

They have no moat. So, how do you get to $4B?

I think the models are wrong way too often for relatively simple queries, so unless they give a secret prompt like "be wrong a lot in the free version" to users, it's basically worthless.

I never sold hardware, so there is that. I am rather anal about correctness and virtualization products are really complicated and as such almost nobody gets them to work reliably (if there is just one known bug, it's trash, IMO).

It's not curious at all; your workload was probably easy. The same program can crash constantly on one piece of _correct_ hardware, while working fine on a different piece of hardware.

I was torturing the hardware and KVM wasn't designed to do that, until about two years ago.

I hate virtualization with a passion, because it's just more crap that doesn't work written by people that shouldn't touch computers.

In theory, it's great. In practice, if you need to get "support" from someone else, it's not so great anymore, as all these companies have been discovering.

I would use VM technology if whoever wrote it would provide me with a contract saying that if anyone were to find just one program that would crash their VM (while not crashing a real machine) or miscompute, that I would get a billion dollars.

To answer your question: I was smart enough to never use it in the first place.

Without proper press access how is there any real accountability?

No.

Real accountability is that the people can torture their leaders when they fail, but that just doesn't happen anymore.

Imagine that this was just a big rock and Trump was sitting on top of the rock like with a group of apes. Also, let's assume that Trump had set fire on the entire banana supply. Do you think the apes would not have picked a different leader immediately?

Rational people would understand that if you make people lose billions that there should be consequences, but someone the world population is more stupid than a bunch of apes.

The US is preparing for war and disinformation is a tool in such a war. The Pentagon just wants the same level of control as the other authoritarian regimes.

It kind of shows that "democracy" was never real to begin with.

Certainly with Sora 2 level of technology they can just claim whoever they don't like has blown up a federal building while they were asleep. It's not like you can have an alibi when you sleep and everyone sleeps. In a way, this AI nightmare necessitates protocols for protection against false accusations like only being able to open the door exiting a house when three other witnesses are present. There are cryptographic solutions like verified cameras, but almost nobody has those now, not even news crews publish signed videos.

Journalism in war time has no real meaning, because if it's secret information the journalist basically becomes an adversary and at some point it becomes cheaper to kill them. Those waging wars have historically always been corrupt. So, that leaves repeating whatever the Pentagon wants you to know.

Also, calling the US a democracy is like putting lipstick on a pig and saying it's a hot babe.

Perhaps Switzerland still has a democracy, but most kind of suck in various ways (and most importantly, don't do anything to improve their democracies). In a real democracy, there would be continuous improvement with better checks and balances. At some point you start to wonder whether democracy just exists to give people the illusion that their voice means anything. Also, in a real democracy there would be equal opportunity and advertising budget for all political parties. That's just not the case in many democracies.

So, blatantly obvious autocracies are probably worse than our current "democracies", but let's not pretend democracy is a thing right now.

Wireguard FPGA 9 months ago

CompCert would be a good example, but everything I have done professionally is also stable; with exclusively people like me, bug tracking systems would not need to exist.

I also have made some software that is proven (meaning from a small 500 line proof kernel) to be correct relative to a trivial implementation (and yes, full correctness is difficult to achieve).

Wireguard FPGA 9 months ago

What exactly is the application you are thinking of? I also do IoT, but "low-power" can mean different things for different people.

Wireguard FPGA 9 months ago

Unless you physically build the FPGA, you still have a black box, but you just shifted the problem (now, I am not saying that this is a bad thing, since if you run Linux on Intel, it's still proprietary and people still run Linux).

Wireguard FPGA 9 months ago

I do not agree he was a great programmer. All of his programs are trivial from a computer science perspective.

In fact, you can quite easily check this by trying to let an LLM generate a program like grep. It can do that. Now, there also exist programs for which LLMs can't generate code, because it's too complex.

Wireguard FPGA 9 months ago

man -T grep | grep 'Free Soft\|Thom'

  (Cop)108 348 Q(yright 1998-2000, 2002, 2005-2023 Free Softw)-.1 E(are F)
Sure, he wrote _a_ version of grep, and probably the first, but who cares? "The" (sure, you might run some bsd grep) current version of grep certainly doesn't.