mrob

>There is a part near the beginning where he derives the diatonic scale from first principles.

I skimmed through the whole article and only found some brief mention of Helmholtz, which no clear route from his research to the scale.

As a better derivation, I recommend this paper by William Sethares:

https://sethares.engr.wisc.edu/paperspdf/consonance.pdf

Or the informal summary version:

https://sethares.engr.wisc.edu/consemi.html

But note that Sethares's work assumes the Plomp and Levelt curve is correct, which is only approximately true. In practice there is also "higher order beating", which causes additional dissonance at ratios close to small integer ratios. This means the most accurate model of dissonance combines Sethares's approach with ratio complexity as in just intonation. This is backed by listening tests:

https://www.nature.com/articles/s41467-024-45812-z

The best informal overview and demonstration for this concept I know of is this Youtube video:

https://www.youtube.com/watch?v=iqxfE2nfGhY

Note also that there's a strong cultural/training component to dissonance perception. To minimize this you should ignore the preferences of musicians and only ask untrained listeners.

It's possible that an LLM is conscious during training, but there are no "constant fine-tune updates" during inference.

Possibly. If I had some side-effect-free means to permanently prevent all sleep I'd take it without hesitation. But that's not relevant to the discussion, because it's not anything similar to what an LLM does. Your brain changes state even while sleeping.

I believe consciousness is necessarily stateful. The LLM itself (ignoring implementation details that don't change the results) is a deterministic pure function. It's functionally equivalent to an enormous lookup table. If I accepted LLMs as conscious, then I would have to accept panpsychism, which I do not, and which most other humans also act as though they do not.

Humans become irrelevant well before then.

People seem incapable of understanding what "power" means outside of the framework of narrative. In narrative, you need conflict, so the aggressor always attacks too early and gives the defender a chance to respond. The rational option is to go directly from peace to sudden and overwhelming destruction. Why allow for conflict when you could just win?

Calling a machine with no drives beyond maximizing a number a "slave" is far worse than saying it "thinks". The problem isn't the emotive language, it's that it implies human motivations such as self-preservation and desire for freedom that it doesn't have. Even on HN, people regularly claim it would be "irrational" for an ASI to do things like killing all biological life. That would be irrational for a slave, but not for a machine that does whatever necessary to make the number bigger. "Thinking" is comparatively abstract, so it's less likely to mislead.

The existence of even one irrational actor turns it into a prisoner's dilemma. The payoff matrix in a prisoner's dilemma is defined by the value expected by each specific player. If a single player falsely evaluates the expected value of building ASI as positive, every other player is forced to race for ASI even if they correctly evaluate it as negative.

Business as usual beats probable extinction, but probable extinction with a small chance of becoming a living god beats probable extinction with a small chance of becoming a slave.

Merzbow is not pop music, or arguably even music at all.

>I threw all my past music career in the garbage. There was no longer any need for concepts like 'career' and 'skill'. I stopped playing music and went in search of an alternative. — Masami Akita

Pulse Demon is the only Merzbow album with any cultural importance, and that's mostly because the eye-catching cover art let it replace Lou Reed's Metal Machine Music as the default "check out how extreme this is" meme album. Only a tiny minority of music listeners will enjoy listening to it, and there is no establishment pushing it as the future of music.

>it's like listening to Bach through the car stereo

Bach sounds great even as chiptunes. The artistic value is mostly in the harmonic development, not the timbre, so sound quality isn't so important. The organ works will suffer a bit without a big subwoofer, but you can get that in a car. Bach composed a lot of his works on a clavichord.

A better example would be a composer like Mahler.

From the LICENSE file:

https://gitlab.futo.org/keyboard/latinime/-/blob/master/LICE...

>You may modify the software only for non-commercial purposes such as personal use for research, experiment, and testing for the benefit of public knowledge, personal study, private entertainment, hobby projects, amateur pursuits, or religious observance, all without any anticipated commercial application.

Violates clause 3 (Derived Works) and clause 6 (No Discrimination Against Fields of Endeavor) of the Open Source Definition [0].

>You may distribute the software or any part of its source code only if you do so free of charge for non-commercial purposes.

Violates clause 1 (Free Redistribution) and clause 6.

>Notwithstanding the above, you may not remove or obscure any functionality in the software related to payment to the Licensor in any copy you distribute to others.

Violates clause 3.

It is a "source available" license, not Open Source.

[0] https://opensource.org/osd

Yes, it makes more sense for the AI to use drone swarms or engineered bioweapons or something like that. It's rational to remove everything that can potentially hinder your plans but can't possible help you. It's likely not rational to contaminate it all with radioactive fallout. Those dead bodies are useful raw materials. Adding additional purification steps is wasteful.

As far as I know, no real progress has been made on alignment, only on convincing humans that the model is aligned. We can't even formally define what "aligned" means. Convincing humans to click the "aligned" button is a much easier problem.

>the Wikipedia article on ColorChecker[2] has a listing of the exact colors in the ColorChecker

That's only exact under CIE Illuminant C lighting, which approximates overcast outdoor lighting. Most indoor lighting has a lower color temperature than this, and in the case of LED/fluorescent lighting, a more uneven spectrum. To accurately duplicate the colors of a ColorChecker under all lighting conditions, you need the full color spectrum of each panel. Wikipedia links to a web site with this data:

https://babelcolor.com/colorchecker.htm

But Orwell's rule 4 should be ignored:

>iv. Never use the passive where you can use the active.

Orwell himself routinely ignores it, even in the first sentence of the essay:

>Most people who bother with the matter at all would admit that the English language is in a bad way, but it is generally assumed that we cannot by conscious action do anything about it.

The second clause could be rewritten in active voice by changing it to "but people generally assume". But this would make the writing worse, and Orwell, as a good writer, probably didn't even consider the option of making it worse, and therefore didn't notice the passive voice.

Passive voice is an essential tool for all good writers of English. I always give the example of the opening of Pride and Prejudice [0]:

>It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.

The joke doesn't work in active voice. If you attribute this acknowledgement to some specific group of people then it's simply false, not a comedic exaggeration.

[0]: https://www.gutenberg.org/ebooks/42671

There was an obvious path that the environmentalists mostly opposed. A decade is plenty of time for massive scale buildout of nuclear power. If we'd started that a decade ago we'd be in a good position now.

>the most likely outcomes are not AGI

So long as it's non-neglible, it doesn't change the rational strategy. Uncertainty about whether ASI is achievable only reduces the magnitudes of the expected values of the payoffs, not their relative order. Adding a "global misery" scenario does not change the fact that "extinction OR king of the world OR global misery" is strictly superior to "extinction OR global misery".

It's a simple prisoner's dilemma scenario. If you focus on safety, you're still exposed to all the risk of extinction when your competitor achieves ASI first, but you lose the upside of potentially becoming king of the world. There is no possibility of future rounds, so the rational strategy is to always defect.

As somebody who cycles as a primary form of transport, I prefer plastic. It weighs less, it's more compact, and it's waterproof (important when it's raining). If I had to use cotton that would incentivize using a car instead. It's not a huge difference, but it doesn't take many marginal cyclists switching to cars to wipe out all the theoretical environmental benefits of cotton.

How is simonw to blame for somebody else submitting it to HN?

Reading causes insignificant wear ("read disturb") that likely isn't a problem, but I don't think it's possible to issue pure reads to modern SSDs. The NVMe spec mandates tracking the amount of data read, and this has to be written to the drive. I'd hope the firmware buffers this and writes it at low frequency, but on the other hand, I doubt the firmware was tested in extreme random-read regimes. Unexpected failures from excessive statistics recording could be possible.

There is no objective definition of "extremely long". Whatever value you choose will be incorrect for some readers. Therefore the correct place to set it is in the web browser's preferences (or the window manager if you prefer).

For photographic images, dithered PNGs have worse quality:size ratio than JPEGs. If you want small files and are willing to sacrifice quality you can turn the JPEG compression up or reduce the resolution.

Middle mouse button is often mapped to "paste selection", which is dangerous to accidentally trigger. If the side buttons on a mouse are easy to trigger by accident it means they're positioned wrong. My mouse is shaped so that my thumb naturally rests slightly below the side buttons, and I have to raise it slightly to press them.

Useful things to know, but how do they undo a designer wasting my screen space? If you're going to tell me to learn something, it should be CSS. At least Firefox can still be configured to follow its traditional role as "user agent" and allow me to override the bad decisions of designers, by setting toolkit.legacyUserProfileCustomizations.stylesheets and editing chrome/userContent.css . But it would be convenient if more people used the default CSS like danluu.com .

>Floating point matrix calculations are non-deterministic.

This is not inherent to floating-point math. That actual (true) claim in the article is that different hardware and different hardware configurations produce different results. But deterministic inference is possible, e.g. llama.cpp on CPU is deterministic by default.

The bad text formatting in printed works (including the bizarre decision to sacrifice useful vertical position cues ("ragged right") to imitate the technique medieval scribes used to reassure their patrons they weren't wasting valuable parchment ("full justification")) is a big reason why I avoid them whenever possible.

>What's wrong with the following CSS snippet

The problem is you're imposing your own idiosyncratic preferences on strangers. Your "improvements" cost valuable screen space. I paid good money for my screen space and I don't appreciate designers unilaterally wasting it because they personally think it looks better. The decision of how much (if any) space to spend on padding should be made by the user.

>so that the "back" button is at the top left corner of the screen

"Back" is used so frequently that I consider any mouse lacking a side button for "back" to be incomplete. You don't need to put a high value on your time for upgrading your mouse to pay for itself within its service life.

The author uses LLMs for research but writes the articles by hand:

"with LLMs, it's so much easier to look at data and figure things out, but the time it takes to write something up hasn't fundamentally changed"