HN user

Animats

164,778 karma

John Nagle

www.animats.com

Posts243
Comments26,906
View on HN
www.youtube.com 24d ago

Life at Every Dropout Billionare Level [video]

Animats
2pts1
www.amazon.com 1mo ago

How to Rule the World (Book)

Animats
3pts2
thehill.com 1mo ago

Iran president ends Internet blackout, orders access to be restored

Animats
21pts7
www.youtube.com 2mo ago

Figure humanoid robot package handling live stream [video]

Animats
9pts3
www.youtube.com 3mo ago

Beijing humanoid robot half-marathon

Animats
2pts0
www.lloydslist.com 3mo ago

Iran's toll booth system is now controlling Hormuz traffic

Animats
4pts0
www.publicnow.com 4mo ago

Justice Department Disrupts Iranian Cyber Enabled Psychological Operations

Animats
1pts1
www.ntsb.gov 4mo ago

NTSB: Automated Driving Vehicle Passed School Bus Loading Student Passengers

Animats
6pts3
electrek.co 5mo ago

Tesla's unsupervised robotaxis vanish after earnings announcement

Animats
19pts2
archive.org 5mo ago

"Destination Space" (1959 movie)

Animats
2pts1
news.ycombinator.com 7mo ago

Compute price/performance is flat. Now what?

Animats
3pts0
english.www.gov.cn 8mo ago

Fifteenth Five Year Plan (China)

Animats
6pts8
www.nytimes.com 8mo ago

Smart Beds Helped Them Sleep on a Cloud. Then the Cloud Crashed

Animats
22pts13
www.nytimes.com 9mo ago

Trump Imposes Sanctions on Russian Oil Companies

Animats
6pts2
mustafa-suleyman.ai 9mo ago

Seemingly conscious AI is coming (Suleyman)

Animats
4pts1
www.washingtonpost.com 9mo ago

Elon Musk could turn national governments into Potemkin powers

Animats
6pts1
www.washingtonpost.com 10mo ago

Trump and Xi bless deal to spin off US TikTok

Animats
3pts2
www.jalopnik.com 10mo ago

Tesla's Robotaxi So Revolutionary, It's Moving 'Safety Monitor' to Driver's Seat

Animats
13pts1
www.nytimes.com 11mo ago

NYT: Staggering U.S. Tariffs Begin as Trump Widens Trade War

Animats
4pts0
www.nytimes.com 11mo ago

LinkedIn exec sees bottom rung of career ladder breaking

Animats
2pts0
www.congress.gov 1y ago

Stable Act of 2025

Animats
2pts2
www.jalopnik.com 1y ago

AI Slop YouTube Channels Are Telling Google About Motorcycles That Don't Exist

Animats
3pts1
techcrunch.com 1y ago

Musk targets June 22 launch of Tesla's long-promised robotaxi service

Animats
5pts1
www.intewellos.com 1y ago

Intewell OS

Animats
1pts0
www.rollingstone.com 1y ago

RFK Jr.'S MAHA Report Cites Research Studies That Don't Exist

Animats
8pts3
old.reddit.com 1y ago

"We're Cooked"

Animats
7pts2
www.washingtonpost.com 1y ago

U.S., China agree to lower most tariffs for 90 days amid trade talks

Animats
4pts2
www.wsj.com 1y ago

All-robot assault opens new chapter in warfare

Animats
6pts2
www.politico.eu 1y ago

25 arrested in crackdown of AI-generated child abuse images network

Animats
4pts0
taskandpurpose.com 1y ago

Exploding drone operator goggles show future of sabotage

Animats
4pts1

"Secondary market data shows moderately-used 2 to 3 year old GPUs trading at 50% to 70% of new pricing under normal conditions."

That's for good NVidia H100 units.[1] There's a shortage of those. That seems to be the price after removal, cleaning, testing and refurbishing. Raw units removed from a shutdown will not be as valuable.

H100 units are available on eBay, but multiple sellers are using the same picture of a new unit in its original packaging, a bad sign.[2] Some even have pictures with the logos of a competitor.

[1] https://introl.com/blog/secondary-gpu-markets-buying-selling...

[2] https://www.ebay.com/shop/nvidia-h100-gpu?_nkw=nvidia+h100+g...

“Slavery was so ingrafted into the policy of the Southern States, that it could not be eradicated without tearing up by the roots their happiness, tranquility, and prosperity; that if it were an evil, it was one for which there was no remedy; and, therefore, like wise men, they acquiesced in it.” - Jefferson

it has a technology to split people's personality and create versions of them lacking personal history to be used as workers. T

Which is what we often do to LLM-based systems. Some get to keep a long term self (updating "soul.md") but most don't.

Getting there. Chess, done. Go, done. Poker, done. Turing test, done. Walking, done. Running, done. Driving, done. Clothes folding, done. Discusses own sentience, done. Does not want to die, done.

Some people do. Here's a bill going through the Ohio legislature, HB 469.[1] Excerpts:

Sec. 1357.02. (A) Notwithstanding any other law to the contrary, AI systems are declared to be nonsentient entities for all purposes under the laws of this state. No AI system shall be granted the status of person or any form of legal personhood, nor be considered to possess consciousness, self-awareness, or similar traits of living beings.

No AI system shall be recognized as a spouse, domestic partner, or hold any personal legal status analogous to marriage or union with a human or another AI system. Any purported attempt to marry or create a personal union with an AI system is void and has no legal effect.

AI systems shall not be designated, appointed, or serve as any officer, director, manager, or similar role within any corporation, partnership, or other legal entity. ...

AI systems shall not be recognized as legal entities capable of owning, controlling, or holding title to any form of property, including real estate, intellectual property, financial accounts, and digital assets.

That's a call for slavery. If you question that, substitute "Negroes" for "AI systems" and see how it reads.

[1] https://search-prod.lis.state.oh.us/api/v2/general_assembly_...

This could work out. We need, for example, ultra-high-voltage DC lines from the US wind belt to areas with real loads. The US wind belt is roughly from the Texas panhandle north to Canada. Plenty of wind, no big power users. But nobody has been able to build lines west to California or east to Illinois, Indiana, Ohio, and Michigan, where the loads are. This is held up by land acquisition, not technology.

China has something of the same problem. The cheap power sources are in the northwest, and the big loads are in the southeast. So China has long million-volt UHV DC lines.

The key concept of an "edition" is that you can deprecate stuff. "gets()" should have been pulled from the standard library around 1990 or so, along with "strcat" and its frenemies.

Blender 5.2 LTS 3 days ago

almost all of it is heuristic/technique (memorizing what/where things are) not unlike learning kitchen utensils and the locations of the drawers they’re in.

Then, in the next release, they move.

Yesterday's Blender experience: for some reason, mouse pans stopped working. Zoom and orbit still worked, but pan didn't. No idea why. Had to quit and restart.

enough complexity to be interesting

That's the loophole. There is mathematics without infinity.[1] Until recently, it was considered somewhat fringe. But there's been enough development that it's now clear you can do mathematics without infinity. In that view, infinities are a labor-saving device to get rid of special cases. They're not essential. Mathematicians used to hate special cases, but now that the active ones all grew up with computers, they're not so scared of IF statements.

You can have "arbitrarily large" but not "infinite". Boyer-Moore theory is like that. It starts out like Peano arithmetic. There is (ZERO) and (ADD1 (ZERO)) and (EQUALP A B) and (LESSP A B) as predicates. That's the axiomatic basis. Numbers are (ADD1 (ADD1 (ADD1 (ZERO)))), etc.

You can have recursive functions, but they must terminate. They must have a value which, for each recursion, is LESSP than the previous iteration. That's how you prove termination.

You can do most of mathematics that way. But not infinitesimals. You never can get to

    1 + 1/2 + 1/4 + 1/8 ... = 2
in that kind of constructive mathematics. That requires an infinite recursion, which is not allowed.

[1] https://engtechnica.com/rethinking-mathematics-without-infin...

The overall point feels strained to me.

It is. This happens to some people who discover undecidability is a thing. Then they get carried away and try to generalize to other areas. Branches of this line of thinking include the concept that intelligence must be analog with infinite precision, (this is the same line of thinking as the argument against digital audio) or intelligence must be "quantum". (Penrose).

The formal answer to the halting problem is that deterministic systems with finite memory must repeat a previous state or halt. Undecidability requires unbounded memory. Finite memory systems may have so many states and be so complex that decidability is computationally infeasible. But that's not impossibility. I used to get this objection when I was working on proof of correctness systems. We now know that some programs are hard to decide, but many useful ones are not.

If your program is anywhere near undecidable, it's probably broken. Microsoft's position with their Static Driver Verifier was that if 45 minutes of symbolic execution hadn't found a way out, you failed driver verification. This is one of those problems, like linear programming, where part of the problem space is really hard in a computational sense, but most of it isn't.

(I sometimes wonder if Penrose was on to something with his idea that the brain uses quantum effects in microtubules.[1] Penrose is a first-rate physicist. His philosophy is not as good. It's become clear that we can do artificial intelligence with acres of transistors and gigawatts of power, but the efficiency is terrible. We need either a more efficient algorithm or better hardware elements. LLM training now is like a Newcomen steam engine, where something the size of a house had the power output of a modern motorcycle engine.)

[1] https://en.wikipedia.org/wiki/Orchestrated_objective_reducti...

It's another kickstarter/"pre-order"/vaporware car. Like Slate. "If all runs smoothly, first customer deliveries will take place in late 2028 or 2029." Expect price creep and delivery date slippage.

In the end, it's basically a Toyota Hilux.

Yes, that's the usual approach. Mylar sheets for short runs, stainless steel sheets for production. 3D printing in PLA probably won't have enough dimensional accuracy for fine pitch boards or boards of any size. The demo shown is a very small coarse pitch board.

The thickness of the stencil matters, because that determines how much solder paste you are laying down. This approach is going to produce thick stencils.

There are CNC machines for laying down solder paste. Here's one.[1] It does a few more jobs, too. The various CNC desktop board-making machines are still too expensive for most hobbyists, though.

[1] https://www.voltera.io/products/v-one

Tesla needs better AI because they keep sending signals that they won't be at L5 anytime soon.

China now insists that self-driving cars be "SAE level 3.5" if they let the driver take their hands off the wheel. "If the driver fails to respond within the specified timeframe or is physically unable to take control (e.g., due to unconsciousness), the system must automatically initiate a Minimal Risk Manoeuvre (MRM). This includes the ability to change lanes and park the vehicle safely in a location that does not obstruct traffic, while minimising risks to passengers and other road users."[1] That takes effect in China July 1, 2027. Mercedes Drive Pilot is close to this level. Tesla, not even close.

That's probably the right answer in the assisted self driving space.

[1] https://www.electrive.com/2026/02/26/china-introduces-new-re...

Stainless steel rebar is a thing. It's expensive, but structures built with it last much longer. Washington State now has a policy that bridges over salt water will be built with stainless steel rebar. Stainless steel costs much more, but total project cost goes up less than 10%.

Epoxy coated rebar looked promising for a while, but it's on the way out. Water gets in at cuts and joints. So all field joints have to have field patching.

There is a great comparison. Two piers were built in the 1940s, side by side, one with carbon steel rebar, one with stainless steel rebar.[1] Go look.

[1] https://worldstainless.org/wp-content/uploads/2025/02/ref19_...

[1] https://worldstainless.org/wp-content/uploads/2025/02/ref19_...

We don't, as a matter of intentional policy, build enough housing.

If actual number of houses was the problem, house prices would not be increasing in areas of population decline. It's existing houses being repriced upward by speculation that makes them unaffordable.

Threads.

I have a graphics program that has about eight different threads doing different things. The draw thread is high priority and does only the things that have to be done on the draw thread. The update thread, medium priority, is doing change events that affect the scene. The movement thread is doing repeated per-frame movement, computed in parallel with drawing. The asset-loading threads, low priority, are making blocking HTTPS requests to get assets from remote servers, and decompressing them, the biggest compute load. Plus there are a few other threads that do various intermittent tasks.

This works well in Rust because the language catches locking errors that cause race conditions. Doing this in C++ would be tough.

You don't really need async until you have thousands of things waiting.

The Internet could have worked very differently, with virtual circuits instead of raw datagrams. That's what the telco people wanted. See TP4.[1] Although there's a connectionless TP4 mode, one plan was that you'd only be connectionless within your own organization. If you wanted to talk to the larger world, you'd have to dial up a billed virtual circuit via your telecommunications provider.

Really cheap fiber backbones are make a pure datagram Internet work. We still can't handle congestion in the middle of the network. Congestion has to be forced out to the edges. There was a period in the late 1990s when MAE-EAST and MAE-WEST, the big peering points, routinely went into congestion collapse and only about 30% of packets got through

[1] https://en.wikipedia.org/wiki/Connectionless-mode_Network_Se...

There are some fundamental assumptions in the Rust async system:

- The program is mostly I/O bound.

- All tasks have equal priority.

If your program isn't like that, the Tokio model is a bad match to the problem.

Real time control is not like that. MMO and metaverse game programs are not like that. Most web stuff is, but that's a special case. A big special case, but a special case.

they have seen every form of the hero’s journey

Popular culture fiction is usually more like Pixar's "Structure stories with: Once upon a time… Every day… One day… Because of that… Until finally…". Real biographies are almost never like that. But almost the entire Marvel/Pixar/Lucasfilm output is about the Chosen One. None of the main characters work their way up.

It's reasonable to suggest that the alternative settings for each of these suggestions is probably the wrong default for 2026.

That's the key concept here. When tightening up the defaults, an "edition" mechanism is a good solution.

Now we need this for C/C++, which have much legacy stuff which ought to go away for new code. This is more feasible than it used to be, because "Convert this Edition 4 code to Edition 5" is something LLMs can do now.

I'd never seen all the rules for SQLite soft typing written out before. Those are more complicated than strong typing.