Hacker News: Legend2440

New comment by Legend2440 in "Amazon vs. Perplexity – U.S. Court of Appeals for the Ninth Circuit"

Legend2440 — Mon, 14 Sep 2026 22:10:03 +0000

Well, publishers did once try using this law to sue adblockers.

But they lost, using an adblocker isn't a violation of the CFAA. They also tried a legal theory under the DMCA's anti-circumvention provision and lost again.

New comment by Legend2440 in "Amazon vs. Perplexity – U.S. Court of Appeals for the Ninth Circuit"

Legend2440 — Mon, 14 Sep 2026 21:20:50 +0000

TL;DR Amazon is mad that Perplexity's agents can browse Amazon logged-in, with a username/password provided by the perplexity user.

Amazon argues this is against the CFAA because they do not authorize such use. They sued and got a preliminary injection. Perplexity appealed and got the injunction thrown out.

The case hasn't actually been to trial on the merits yet and is still undecided.

New comment by Legend2440 in "Patients Who Fight Health Insurance Denials Often Win"

Legend2440 — Mon, 14 Sep 2026 18:18:38 +0000

Seems broken, doesn't contain the entire article text.

New comment by Legend2440 in "Garry Tan wants US open-weight AI labs to 'distill' frontier models, too"

Legend2440 — Sun, 13 Sep 2026 19:28:09 +0000

That's not what I mean. If competitors can offer tokens 15x cheaper, the big labs must have high margins per token. (which they can use to amortize training costs)

>Sam Altman himself conceded that OpenAI is losing money on the $200 subscription.

They have since stopped offering the $200 subscription, probably for this reason.

Subscription margins are harder to judge because it depends on usage; token costs are a better comparison.

New comment by Legend2440 in "Garry Tan wants US open-weight AI labs to 'distill' frontier models, too"

Legend2440 — Sun, 13 Sep 2026 18:35:00 +0000

Unfortunately, the government doesn't want to ban data brokers because the government wants to buy from data brokers.

New comment by Legend2440 in "Garry Tan wants US open-weight AI labs to 'distill' frontier models, too"

Legend2440 — Sun, 13 Sep 2026 18:29:35 +0000

>inference is already being subsidized.

Inference is not being subsidized and in fact has pretty high margins.

Similar-sized open weight models on openrouter are 15x cheaper per token than the big labs. This should reflect the isolated cost of inference, since 3rd party hosts have no reason to subsidize and no training costs to amortize.

Only datacenter buildout costs are being subsidized.

New comment by Legend2440 in "Navier-Stokes Announcement"

Legend2440 — Sat, 12 Sep 2026 05:31:56 +0000

Their rules PDF says they won't accept any solution until at least two years after publication in a qualifying outlet. This allows time for the mathematical community to review and accept new results.

As the OpenAI proof hasn't been officially published yet, the clock hasn't started ticking.

New comment by Legend2440 in "Don't let anyone take away your big box of cables"

Legend2440 — Fri, 11 Sep 2026 01:59:21 +0000

>I know there's someone out there who could make use of them.

I wouldn't be so sure. I bet there's more cables out there than functioning equipment, especially for older standards like VGA or PS/2.

It's ok to throw out junk.

New comment by Legend2440 in "More questions about whether researchers can trust OpenAI with unpublished math"

Legend2440 — Fri, 11 Sep 2026 00:57:54 +0000

*to scoop a team working with Anthropic, their chief competitor.

Also, they didn't have the solution. They had a lesser problem no one cared about.

New comment by Legend2440 in "More questions about whether researchers can trust OpenAI with unpublished math"

Legend2440 — Thu, 10 Sep 2026 04:43:02 +0000

This is a really weak claim. The evidence they offer is just "someone somewhere says they had a discussion with AI about the topic at some point".

They don't even claim to have had a proof, only to have been working on it.

New comment by Legend2440 in "On the Navier–Stokes Millennium Prize Problem"

Legend2440 — Tue, 08 Sep 2026 17:33:05 +0000

Probably not. It's a millennium prize problem, a great many mathematicians have been working on it for a very long time.

New comment by Legend2440 in "On the Navier–Stokes Millennium Prize Problem"

Legend2440 — Tue, 08 Sep 2026 17:31:47 +0000

That other researcher was working on a smaller related problem.

He was also using LLMs to do it, so either way most of the credit goes to the LLM here.

New comment by Legend2440 in "On the Navier–Stokes Millennium Prize Problem"

Legend2440 — Tue, 08 Sep 2026 17:29:41 +0000

The prize is what, a million dollars?

OpenAI doesn't need a million dollars.

New comment by Legend2440 in "Trusting-Trust Attack against an Entire Linux Distribution"

Legend2440 — Mon, 07 Sep 2026 20:27:39 +0000

Nothing about this uses AI in any way. I'm so tired of people injecting it into every conversation.

New comment by Legend2440 in "How concerned should we be about Astra's recurrent architecture?"

Legend2440 — Fri, 04 Sep 2026 23:00:57 +0000

>the same number of weight should be able to hold many orders of magnitude more information by being reused.

No, not hold more information, perform longer computations.

E.g. if you want to solve sudokus, you will need more and more loops for larger grids. There is no shortcut.

New comment by Legend2440 in "Formalizing Fermat's Last Theorem"

Legend2440 — Fri, 04 Sep 2026 22:28:31 +0000

The point of Lean is that it can be mechanically verified by a proof checker.

New comment by Legend2440 in "Can AI design circuit boards yet?"

Legend2440 — Fri, 04 Sep 2026 21:15:43 +0000

Sounds like you're talking about the FPGA evolution experiments Adrian Thompson and others did in the 1990s: https://evolvablehardware.org/history.html

New comment by Legend2440 in "Google AI Mode shows same products 21.6% more expensive than traditional search"

Legend2440 — Fri, 04 Sep 2026 19:11:29 +0000

>So are companies paying more to have products featured in AI mode rather than just plain search?

No - in fact, the results listed in the article are not paid ads.

You only can pay to be in the 'sponsored results' section at the bottom of the AI mode results.

New comment by Legend2440 in "Google AI Mode shows same products 21.6% more expensive than traditional search"

Legend2440 — Fri, 04 Sep 2026 19:05:46 +0000

How exactly does this help them?

There is a lot of competition in the gift/trinket space. You could buy that birthday present from any number of sellers. If seller #183 jacks up prices, you will go buy from seller #765 instead.

New comment by Legend2440 in "Google AI Mode shows same products 21.6% more expensive than traditional search"

Legend2440 — Fri, 04 Sep 2026 19:02:59 +0000

Keep in mind that there is still a downwards force on prices from competition.

More information does not necessarily help AutoZone raise prices because (1. everybody already knows that people who buy car parts are likely trying to fix their car and (2. other sellers will undercut them and they will miss out on sales.