Hacker News: layer8

New comment by layer8 in "Netflix Prices Went Up Again – I Bought a DVD Player Instead"

layer8 — Thu, 09 Apr 2026 23:13:51 +0000

One thing to pay attention to is that Netflix deletes your watchlist if you “pause” for longer than ten months.

New comment by layer8 in "Netflix Prices Went Up Again – I Bought a DVD Player Instead"

layer8 — Thu, 09 Apr 2026 22:58:16 +0000

Yes, for older TV titles, the main reason to opt for Blu-ray is the better sound quality. Although DVD supports uncompressed audio (LPCM), that was rarely used outside Japan, and regular stereo audio typically used pretty mediocre compression.

When using subtitles, another reason is the higher-resolution fonts.

New comment by layer8 in "ChatGPT Pro now starts at $100/month"

layer8 — Thu, 09 Apr 2026 19:01:11 +0000

It states “From $100”. Standard pricing speak.

Scientists discover 'most chemically pristine' star yet found in the universe

layer8 — Thu, 09 Apr 2026 18:50:07 +0000

Article URL: https://news.uchicago.edu/story/scientists-discover-most-chemically-pristine-star-yet-found-universe

Comments URL: https://news.ycombinator.com/item?id=47707974

Points: 4

# Comments: 0

Satellites capture the volatile human–luminescence relationship

layer8 — Thu, 09 Apr 2026 18:41:59 +0000

Article URL: https://phys.org/news/2026-04-satellites-capture-volatile-humanluminescence-relationship.html

Comments URL: https://news.ycombinator.com/item?id=47707846

Points: 2

# Comments: 0

New comment by layer8 in "Claude mixes up who said what"

layer8 — Thu, 09 Apr 2026 18:08:08 +0000

First, the “optimizations” are not IEEE 754 compliant. So nondeterminism with floating-point operations is not an inherent property of using floating-point arithmetics, it’s a consequence of disregarding the standard by deliberately opting in to such nondeterminism.

Secondly, as I quoted the paper is explicitly making the point that there is a source of nondeterminism outside of the models and GPUs, hence ensuring that the floating-point arithmetics are deterministic doesn’t help.

New comment by layer8 in "Tree Calculus"

layer8 — Thu, 09 Apr 2026 17:56:56 +0000

> the application of E1 to E2 attaches E2 to the root of E1 on the right.

It’s completely unclear to me what this means. The literal meaning is obviously wrong, because attaching a tree to a root that already has two child nodes would result in a ternary node, but apparently all trees in tree calculus are binary.

New comment by layer8 in "The Pentagon Threatened Pope Leo XIV's Ambassador with the Avignon Papacy"

layer8 — Thu, 09 Apr 2026 17:43:43 +0000

They don’t disagree that they conduct diplomacy based on force. They disagree that they should instead promote dialogue and seek consensus among all parties.

New comment by layer8 in "The Pentagon Threatened Pope Leo XIV's Ambassador with the Avignon Papacy"

layer8 — Thu, 09 Apr 2026 17:41:03 +0000

I find it positively interesting that pope Leo’s outspokenness is apparently considered such a threat by the Pentagon.

New comment by layer8 in "The Pentagon Threatened Pope Leo XIV's Ambassador with the Avignon Papacy"

layer8 — Thu, 09 Apr 2026 17:32:29 +0000

It’s still much preferable to them being strong and competent in subduing everyone to their will.

New comment by layer8 in "Claude mixes up who said what"

layer8 — Thu, 09 Apr 2026 16:41:55 +0000

You still can’t predict whether the in-language responses will be correct or not.

As an analogy: If, for a compiler, you verify that its output is valid machine code, that doesn’t tell you whether the output machine code is faithful to the input source code. For example, you might want to have the assurance that if the input specifies a terminating program, then the output machine code represents a terminating program as well. For a compiler, you can guarantee that such properties are true by construction.

More generally, you can write your programs such that you can prove from their code that they satisfy properties you are interested in for all inputs.

With LLMs, however, you have no practical way to reason about relations between the properties of inputs and outputs.

New comment by layer8 in "Am I German or Autistic?"

layer8 — Thu, 09 Apr 2026 14:53:33 +0000

Your grammar is wrong.

New comment by layer8 in "Claude mixes up who said what"

layer8 — Thu, 09 Apr 2026 12:54:17 +0000

In this context, it means being able to deterministically predict properties of the output based on properties of the input. That is, you don’t treat each distinct input as a unicorn, but instead consider properties of the input, and you want to know useful properties of the output. With LLMs, you can only do that statistically at best, but not deterministically, in the sense of being able to know that whenever the input has property A then the output will always have property B.

New comment by layer8 in "Claude mixes up who said what and that's not OK"

layer8 — Thu, 09 Apr 2026 11:52:05 +0000

This has the potential to improve things a lot, though there would still be a failure mode when the user quotes the model or the model (e.g. in thinking) quotes the user.

New comment by layer8 in "Claude mixes up who said what"

layer8 — Thu, 09 Apr 2026 11:40:06 +0000

You still can’t deterministically guarantee anything about the output based on the input, other than repeatability for the exact same input.

New comment by layer8 in "Claude mixes up who said what"

layer8 — Thu, 09 Apr 2026 11:36:56 +0000

This has nothing to do with FP inaccuracies, and your link does confirm that:

“Although the use of multiple GPUs introduces some randomness (Nvidia, 2024), it can be eliminated by setting random seeds, so that AI models are deterministic given the same input. […] In order to support this line of reasoning, we ran Llama3-8b on our local GPUs without any optimizations, yielding deterministic results. This indicates that the models and GPUs themselves are not the only source of non-determinism.”

New comment by layer8 in "The AI Great Leap Forward"

layer8 — Wed, 08 Apr 2026 22:06:03 +0000

If they collapse, then because their value proposition doesn’t add up. It’s unclear why that should be different with their competitors then.

New comment by layer8 in "Union types in C# 15"

layer8 — Wed, 08 Apr 2026 18:41:13 +0000

> so our type is really "OneOrMoreOrNone"

If I understand correctly, it’s actually OneOrOneOrMoreOrNone. Because you have two different distinguishable representations of “one”.

The only reason to use this would be if you typically have exactly one, and you want to avoid the overhead of an enumeration in that typical case. In other words, AnyNumberButOftenJustOne.

New comment by layer8 in "Worldwide Semiconductor Revenue to Exceed $1.3T in 2026"

layer8 — Wed, 08 Apr 2026 17:02:02 +0000

But also: “Gartner estimates DRAM and NAND flash annual prices in 2026 will increase by 125% and 234%, respectively, and any meaningful pricing relief is not expected until late 2027.”

Worldwide Semiconductor Revenue to Exceed $1.3T in 2026

layer8 — Wed, 08 Apr 2026 17:02:02 +0000

Article URL: https://www.gartner.com/en/newsroom/press-releases/2026-04-08-gartner-forecasts-worldwide-semiconductor-revenue-to-exceed-us-dollars-one-point-3-trillion-in-2026

Comments URL: https://news.ycombinator.com/item?id=47692999

Points: 1

# Comments: 2