vczf

Doing your own oil changes is not worth the hassle when considering the risk of a spill and the difficulty of legal disposal—unless you have a fancy engine that needs frequent oil changes.

Cabin air filter and wiper fluid, sure. Headlights and taillights used to be a no-brainer, but now those are often sealed LED assemblies and difficult to access as well.

The clear distinction between scalars and vectors appears to be the main advancement HUML offers.

I think it’s a neat improvement.

...stream for Emacs...

...terrible UX, terrible everything...

I think I accidentally enrolled for emacs and can't unenroll on the site. I guess I'll have to finally start using emacs now

Should I buy a gun? I'm an American.

No, that's unnecessary. Nobody will be taking you that seriously.

some new enrollments topped up their accounts and dropped off before the final step that makes it show up on the home page

Did they actually put money in?

That's a very clear vision on how to solve this kind of funding/cooperation problem outside of government and mission-focused nonprofits. And incidentally would be an existential threat to surveillance capitalism should it reach critical mass.

BTW your password-based signup flow isn't working (on iOS Safari at least).

I think LLMs are more like the invention of high level programming languages when all we had before was assembly. Computers will be programmable and operable in “natural language”—for all of its imprecision and mushiness.

The more general pattern is “slowly at first, then all at once.”

It almost universally describes complex systems.

5mg may be an order of magnitude more than needed.

I started my father [0] on ≈200ug and he was more animated and looked at photos around the house much more. Bumped him up to ≈500ug [1] the last two days, and he’s shown a nearly unbelievable improvement in lucidity and recognition.

He’s still nonverbal but babbling a lot more, and in a non-agitated way. He’ll make eye contact with us more and wiggle his eyebrows. He has not been running away from me during morning yogurt and other meals we sometimes have standing up. More responsive to music, photos, eye contact.

He actually kissed my mom today and then got teary-eyed. He hasn’t done anything like that in at least a year.

Tonight when he was tired, he wasn't unstoppably wandering. He was tired but still “there” and lucid. I was able to calmly walk him to his bedroom.

Lithium orotate might be the magic bullet. Lithium deficiency could very well be the cause of age-related neurodegeneration. It’s too early in my personal case study to draw firm conclusions, but this is looking absolutely incredible from my perspective as a caregiver. Subjectively, he seems no longer “lost” in the darkness.

If he is still stable and/or improving in a month, I will be making as much noise about this treatment as I can online and in meatspace. [2]

[0] Early onset posterior cortical atrophy and corticobasal degeneration. It presents initially as visual disturbances, hallucinations, and coordination problems rather than memory loss. Sometimes memory and speech can persist until the end.

[1] I opened up some 5mg Nutricost capsules, weighed the contents, and calculated how much of the mixture is needed for a given amount of active ingredient. Then I measure out each dose on a calibrated milligram scale. This is definitely out of reach for many people, so ideally companies can just add a small dose to multivitamins and call it a day. I’ve been taking the same dose I give my dad, and have not noticed much of anything.

[2] There are some ethical considerations in halting the disease process for those with late stage dementia. It would be inhumane to “cure” advanced AD if one has no hope of a life worth living, given one’s current capabilities and options. (For example, bed-bound in nursing homes with severe memory loss.) I’m very optimistic about lithium orotate, but I doubt it can do more than halt the neurodegeneration—which would mean rapidly hitting a low ceiling for cognitive improvements after beginning the therapy. That also means a practically life-long caregiving requirement. I didn’t want to be a caregiver for my father for the next 30+ years, but frankly I did not expect such a dramatic and immediate improvement in his condition. I expected nothing at all to happen.

The important part is this: “a Li salt with reduced amyloid binding”

If cells in the brain are being deprived of lithium due to sequestration by amyloid beta plaques, then a bioavailable form of lithium that is resistant to sequestration may treat the pathology.

From the abstract of the paper:

  > Replacement therapy with lithium orotate, which is a Li salt with reduced amyloid binding, prevents pathological changes and memory loss in AD mouse models and ageing wild-type mice.

Another source on lithium orotate:

  > LiOr is proposed to cross the blood–brain barrier and enter cells more readily than Li2CO3, which will theoretically allow for reduced dosage requirements and ameliorated toxicity concerns.

Sure, but the amount and form of the lithium matters. 5mg of lithium orotate (as a supplement) versus 600mg lithium carbonate (as a mood stabilizer) will have vastly different acute and chronic health effects.

I’m a newish Kagi user and I find myself using the LLM about as frequently as search itself.

Sometimes I search for things I know I am looking for. Other times I don’t know quite what I’m looking for or I know in advance that I’m not likely to find it—so I chuck it at Llama 4 Maverick and it usually gives me something useful.

I had no plans to use the LLMs until they opened it up on my tier. At this point however, it’s half the value I get out of Kagi.

The genius of capitalism is in its unique ability to harness the energy of sociopaths* to produce valuable products and services**.

If there was no freedom to amass fortunes, these people would still exist, and they would do even more damage in whatever theoretical social structure we would have.

* And to a lesser degree, the self-interest everybody naturally possesses.

** Minimizing negative externalities is the responsibility of government accountable to the people.

If all we needed was general intelligence, we would be hiring octopuses. Human skills, like fluency in specific languages, are implicit in our concept of AGI.

A layer of very thin gold can be transparent.

I agree, it's definitely still possible to get bored.

If I stop making progress on my personal projects, sinking my free time into games or online interaction is very unsatisfying.

It doesn't seem like a completely different thing to generate specifications and formally verified programs for those specifications (though I'm not familiar with how those are done today).

Artificial Super-Intelligence

You might be underestimating the potential of an automated evolutionary programming system at discovering novel and surprising ways to do computation—ways that no human would ever invent. Humans may have a better distribution of entropy generation (i.e. life experience as an embodied human being), but compared to the rate at which a computer can iterate, I don't think that advantage will be maintained.

(Humans will still have to set the goals and objectives, unless we unleash an ASI and render even that moot.)

The endpoint is that being a programmer becomes as obsolete as being a human "calculator" for a career.

Millions, perhaps billions of times more lines of code will be written, and automated programming will be taken for granted as just how computers work.

Painstakingly writing static source code will be seen the same way as we see doing hundreds of pages of tedious calculations using paper, pencil, and a slide rule. Why would you do that, when the computer can design and develop such a program hundreds of times in the blink of an eye to arrive at the optimal human interface for your particular needs at the moment?

It'll be a tremendous boon in every other technical field, such as science and engineering. It'll also make computers so much more useful and accessible for regular people. However, programming as we know it will fade into irrelevance.

This change might take 50 years, but that's where I believe we're headed.

I think you're too quickly dismissing the potential benefits. From the FAQ[1]:

  The combination of all the above properties, plus the fact that I/O is also separate from pure computation and updates, enables a number of features that are not found in conventional languages.

  The first one is the ability to "replay" the execution of a Cell program. One can easily reconstruct the exact state of a Cell program at any point in time. Obviously that's very useful for debugging, but there are others, more interesting ways of taking advantage of that. It's for example possible to run identical replicas of the same Cell process on different machines over a network, and that's something that will be crucial in Cell's future network architecture.

  The second one is orthogonal persistence, that is, the ability to take a snapshot of the state of the application (or part of it), which can then later be used to create with minimal effort an identical copy of it, that will behave in exactly the same way. In Cell, you don't save your data the way you do it in conventional languages. Instead, you simply declare which parts of your application's data structures have to be persisted, and the compiler generates for you all the code needed to save and load them.

[1] https://www.cell-lang.net/faq.html

They released a cell to cpp compiler last year: https://www.cell-lang.net/version-0.7.html

GitHub: https://github.com/cell-lang/c-cpp/tree/master/src/compiler

I'm speculating that one would have more mirror neuron activation watching a person perform live, compared to listening to a recording or watching a video. Thus the missing component that makes live performance special.

Mirror neurons?

"Hold up, let me take my face off real quick."

Sampling methods also affect this. Have you tried min_p sampling? https://github.com/ggerganov/llama.cpp/pull/3841#issuecommen...

In addition, Apple and AMD both sell hardware, and are not competing in the same spaces.

What if you encoded the whole game state into a one-shot completion that fits into the context window every turn? It would likely not make those illegal moves. I suspect it's an artifact of the context window management that is designed to summarize lengthy chat conversations, rather than an actual limitation of GPT4's internal model of chess.

As a counterpoint, the most thoughtfully designed and responsive task management app I've ever used is a Flutter app (http://quire.io) Maybe check it out and see if it changes your mind about what's possible with Flutter.

I don't think it would know how to have a conversation if it's never been exposed to one before.