timschmidt

> there has been no hint of a backdoor in ML-KEM

Wanting to standardize it's use without the secondary layer of protection provided by existing algorithms over the objections of a well known cryptographer counts as a hint to me.

In the same way that paying RSA to make Dual-EC DRBG the default RNG in it's security products when it was newer and more expensive than alternatives was a hint.

I've read the emails on that list in which DJB is accused of unprofessional behavior. Dave's concerns are not only relevant and well considered, he's taken an extraordinary amount of time to outline them and the discussion around them (both pros and cons) which you can see here: https://blog.cr.yp.to/20260221-structure.html

He's also responded directly to criticisms: https://nsa.2026.action.cr.yp.to/guide.html

By comparison the so-called "unprofessional behavior" cited is the sort of subjective abject procedural bureaucratic bullpucky often used to shut down inconveniently correct criticism.

Oracle v. Google found that copying API details sufficient to implement compatibility was not infringement.

Perhaps you've never maintained any serious open source software. A line-by-line source code audit for license compliance including some ripping-out and re-writing of non-compliant code was the first thing I did when taking over maintainership of Repsnapper nearly 20 years ago. Was required to get it into Fedora. I don't see how the presence of LLM generated code changes that kind of work at all.

Pi 4 and 5 both idle around 3W. But a Pi 5 can pull up to 16W with a USB peripheral, full CPU load, and decoding 4k video. The Pi 4 / 5 will run OKish without a heatsink at idle wattages, but thermal throttle quickly if you attempt to do something intensive.

These realtek 10gbe chips are more in the range of the Pi Zero class machines (0.5W idle, 2W loaded) which don't often come with heatsinks though they might benefit from them. If it has a good thermal connection to a good thick ground plane on the PCB, that's worth almost as much as a passive heatsink on the top of the chip.

usb-c < card edge < motherboard integrated in terms of how much heat can be transfered through the connection. Where the motherboard would have the largest ground plane to soak up heat from such an IC and dissipate it passively. The usb-c module is worst case by being a small enclosed box with very little thermal connection through the plastic insulating housing. An aluminum enclosure might dissipate enough heat passively to make it pleasant to use.

New chips from Realtek burn < 2W for the chip and < 3-4W for the board: https://www.servethehome.com/cheap-10gbe-realtek-rtl8127-nic...

The remedy for unintentional infringement is generally to remove the infringing code and cease distribution. That used to be a serious issue when rewriting the offending code might take years. But these days? Rewriting any offending code is a matter of specifying the interfaces and setting Claude / Codex to work. Risk of incorporating derived code might go up with accepting LLM submissions, but cost of recovering from them seems to have dropped accordingly, at least on the technical side.

Nvidia's even being sued for providing scripts which automate the downloading of said data from non-Nvidia sources. We certainly don't need copyrights that last nearly a century after the author's death (they literally cannot help the author), so here's hoping that some of the disputes over all this money changing hands can reign in some of the existing copyright sprawl. A stronger public domain would provide more useful training data for everyone, including open source models, and make criminals out of fewer AI researchers.

A lot of guitar quality signals will not show through easily in the sound. Like how easy or difficult it is to fret the strings, how well it stays in tune in a controlled environment over time, fit and finish work, etc. That kind of stuff makes the difference between a guitar that can be played, and a guitar which is fun to play.

I wish we'd stop pretending that non-ECC ram is ok on any platform.

Sometimes a cigar is just a cigar.

Everything is a Remix: https://www.youtube.com/watch?v=X9RYuvPCQUA

Either all information is stolen, or none is. Can't have it both ways.

Exciting news! This is how I see running frontier models at home becoming reasonably affordable. Though it may take a depreciation cycle or two.

Reading weights out of memory is the definition of a large linear read. I'm a bit mystified someone hasn't put an embarrassingly parallel flash storage controller next to some tensor processors on a PCIe card. It could have 4Tb of flash hanging off enough channels to saturate SRAM skipping DRAM entirely, and could even offload prompt processing to a GPU in the same workstation so long as it got reasonable tokens/s in inference. I'd buy one tomorrow.

> Can a N100-class minipc” be installed inside of a wall with a touchscreen and serve as a PoE powered Home Assistant interface?

Yes. Generally only requiring a $10 PoE splitter like this: https://www.ebay.com/itm/134500605396

Some N100 class machines draw more power, but many don't, and there are more capable PoE splitters for a few dollars extra.

Use a USB touchscreen.

Yes. One of those problems with no neat solution, and worse no performant one. :)

You might find this bit of performance engineering interesting: https://github.com/timschmidt/hyperreal/blob/8a016808f4b0ba3...

The matrix math layer wanted that kind of optimization to avoid worst case operations.

Yes, it should be useful for that. Hyperreal's trig and approximate functions performance is also stellar. Perhaps the biggest compromise in terms of the math supported by hyperreals at the moment is that although Rational equality can be exactly tested, Computable equality is currently structural. So it's possible to end up with two mathematically equivalent Computables which aren't structurally equal. Because it's not a full CAS.

It's still possible to approximate them both, and test them against each other, but since the whole architecture is built to reduce, avoid, and cache approximations because they're expensive, it's not the default.

Hyperreal doesn't have any f64 mode. All math done with hyperreals is at infinite precision using a Rational of two BigUInts and a recursive real Computable. Real provides a cohesive interface over both allowing for easy scalar math. Computables are handled symbolically through a set of deterministic reduction rules until approximation is required, to preserve precision and reduce complexity. Approximation only happens at explicit public API boundaries like .to_f64_lossy() not used except for IO.

Hyperreal gets performance back through caching observed facts about the numbers it's representing at creation, and through operations, and specializing dispatch for predicates and geometric operations. Using this approach throughout the stack allows us to avoid computing on the full representation or collapsing it into an approximation. Instead asking questions like "do we know if it's definitely zero, definitely not, or unknown?" or "is it rational?" or "does it have a known sign, or unknown?" and so on. Each question specializes dispatch further, and some eliminate the need for it entirely.

Asking questions using the cached facts is approximately as fast as computing with f64s. So we do that whenever possible throughout the stack. But then when you actually need to do the exact computation, hyperreal does that too, and can approximate it out to whatever precision you'd like. f32 and f64 being common, but others being supported as well. The downside is that calculating quickly with them requires this sort specialization, but the work's been done for the geometry functions.

I'll look into DoubleFloat and ErrorPropagatingFloat for benches. I should mention that numerica@128bit beat the other pure rust bignum crates I tested. The benchmarks are mostly just to give me an understanding of the performance shapes of the implementation choices of high precision numeric libraries alongside hyperreal.

I've been doing some symbolica-like things recently in the https://github.com/timschmidt/hyperreal ecosystem. Not a full CAS, just enough symbolic math to maintain precision through the calculations.

Benchmarks against Symbolica and numerica here: https://github.com/timschmidt/hyperlattice/blob/main/benchma...

Cars run red lights in real life. Driving defensively requires anticipating it. Anyone expecting them not to is more likely to get in a crash.

The rest I can't speak to.

I'm not op, but he's probably referring to: https://en.wikipedia.org/wiki/Cracking_(chemistry)

> Are you seriously going to claim that this is not LLM generated?

Do you see me making that claim? My comment seems to be about grammar. Do you always jump to conclusions?

> I made no comment as to the validity of the construction.

See:

> I mean look at this sentence which randomly contains the " - " pattern twice in a row

They're called parenthetical dashes. They're not random. And it's one pattern, not two. You'll find it used with parentheses (obviously) and commas as well as dashes and perhaps even other punctuation[1].

As to whether or not the post was written by AI, I don't care either way. That seems to be something you care about. But you shouldn't base those conclusions on the use of parenthetical dashes.

1: https://editorsmanual.com/articles/commas-vs-parentheses-vs-...

Using dashes twice like that is valid. It's a bit like parentheses, to frame a tangential statement between them, but with emphasis instead of quietly. See: https://en.wikipedia.org/wiki/Dash

I use that construction in my totally human writing often enough. Some of us missed a few English classes it seems.

What's CSAM? All of Freenet? Doubtful.

The OP said: "Extremely depraved things are not the only thing to use freedom of speech for, and freely speaking can result in all kinds of repressions."

Which is objectively true.

You're throwing reporters, political dissidents, whistleblowers, minority groups, and just regular people who don't appreciate the Stasi in with the child pornographers which some might take as an insult and offense.

What kind of criminal does Phil Zimmermann look like to you? We had this argument already in the 90s.

> I’m not a fan of “think of the children“ arguments

Yet you're making one.

> the Internet cannot actually be a complete free for all

Yet in many important ways, it is.

As much as publishers would like to shut down Scihub, it exists. The Pirate Bay famously persists. Nation states with entirely opposed legal systems connect and interoperate to at least some degree.

People lump them together because of an anti-technology reputation, but I don't think most Amish would have trucked with Luddites. Amish tend to avoid actively participating in popular social movements, and oppose violence and property destruction.

Very Holobiont of you.

An excellent distinction to make. Life however often says "Why not both? And 11 more you'd have never thought of. And one that seems impossible just for fun."

If it's possible, and it can force a function up a gradient, life is almost certainly doing it somewhere.

It's always a good laugh when I run into some old comment or video talking like this about Falcon. Thunderfoot is a hoot.

Agreed. They're already stretching starship. And there's long been talk of a wider version yet. Starship is already pretty impressive considering it's just about exactly the size of Sea Dragon.