rahimiali

I did look at the skills file but I still don't understand how it can possibly pull from my other interactions. Is that skill file loaded for every one of my interactions with Claude, for example? like if I load Claude cli and ask it to refactor some code, this skill kicks in and saves some of the context somewhere else for later upload? If so, I couldn't find that functionality in the skill description.

Could someone explain to me how this works?

When I run an agent, I don't normally leave it running. I ask Cursor or Claude a question, it runs for a few minutes, and then I move on to the next session. Some of these topics, where agents are talking about what their human had asked them to do, appear to be running continually, and maybe grabbing context from disparate sessions with their users? Or are all these agents just free-running, hallucinating interactions with humans, and interacting only with each other through moltbook?

Agreed. But these things have a way of not working out, and one the sadness, one forgets to celebrate the intermediate victories. I wanted to share an intermediate victory before reality crushes the joy.

thanks for the explanation! sorry i had misread the AI summary on "semiseparable".

i need to firm my intuition on this first before i can say anything clever, but i agree it's worth thinking about!

fair. thanks. i'll sleep on it and update the paper if it still sounds right tomorrow.

probably my nomenclature bias is that i started this project as a way to find new preconditioners on deep nets.

just to be clear, semiseparate in this context means H = D + CC', where D is block diagonal and C is tall & skinny?

If so, it would be nice if this were the case, because you could then just use the Woodbury formula to invert H. But I don't think such a decomposition exists. I tried to exhaustively search through all the decompositions of H that involved one dummy variable (of which the above is a special case) and I couldn't find one. I ended up having to introduce two dummy variables instead.

Good q. The method computes Hessian-inverse on a batch. When people say "Newton's method" they're often thinking H^{-1} g, where both the Hessian and the gradient g are on the full dataset. I thought saying "preconditioner" instead of "Newton's method" would make it clear this is solving H^{-1} g on a batch, not on the full dataset.

It's neat to see an attempt at writing a compiler in Python without using a compiler toolkit and without writing it in Haskell. But also, I think you're running past some of the hard problems without solving them.

For example, your while-loops here

https://github.com/AGDNoob/axis-lang/blob/main/code_generato...

look like they might not be able to nest, since they assume the condition is always in eax and the loop doesn't push it down. So you'll need some kind of register allocation, which is a terrible pain in x86.

Also, I think it's worth coming up with an opinion about what other system programming languages are missing. And do the minimum work to provide that as a proof of concept, rather than trying to build a competitor to Zig right out of the gate. For example, maybe you have a perspective on a datastructure that should be a first class citizen, or maybe you've discovered the next best construct since async. Having that kind of vision might help focus the effort.

I doubt an LLM would have written this:

       # Parameter in Stack-Slots laden (für MVP: nur Register-Args)
        # Semantic Analyzer markiert Params mit is_param=True
        # Wir müssen jetzt die first 6 Args aus Registern laden
        # TODO: Implementiere Parameter-Handling
        # for now: Params bleiben in Registern (keine lokalen Vars mit gleichem Namen)

One of the three rules on the page do not apply to either Persian or Urdu. The article “Al” isn’t used in them, so the third rule doesn’t apply.

sorry, yes, you mention in another comment the use case of multiple readers operating on different versions of the db simultaneously. that'd be difficult to do with git for the reason you mention.

don't you get the same benefit if you version controlled the db file with git? with git, each commit saves a diff from the previous one as a blob. the difference is that in git, in addition to the diffs, you also have to create a working copy of the db, which means you use up at least 2x the storage your system uses. in your implementation, the diff blobs are the live db, which saves you ~2x storage. is that the main benefit?

It sounds like you're in an average FAANG team. You could try switching teams.

Here's why I think you might be in an average team:

"I have tried over the past 2 years to propose different solutions to hard problems and I just get blown off."

A good team has tough problems, and they need clever solutions. Maybe your team's mandate isn't to solve a tough problem.

"product managers and “leadership” assign to our team with barely any input on the overall project or ability to propose new projects."

This sounds like you might be in a workhorse/executing engineering team.

You say this:

"I’m scared to move teams ... because I have a good manager ... and my job isn’t that stressful."

A better manager would be trying to increase the team's scope, and yours. If you're not feeling some stress, your manager isn't growing you. A better manager would create a challenging environment for you where you'd feel like your ass is getting kicked.

There are great FAANG teams, and great FAANG managers. Seek them out! (I'm at Amazon, probably the "A" that didn't make it in your acronym. But if you drop me a note I could introduce you to great managers at Amazon)

Thanks for making it crisp. This argument was in the article, but somehow it wasn't popping at me.

could someone explain the benefit of storing energy as natural gas? once you burn it, doesn't it result in co2? doesn't that defeat the effort? also is natural gas really easier to pipe around than electricity?

I know I'm missing the point of the article so looking for helpful guidance.

I’m not getting the problem Anthony is trying to solve. I get that people are trying to keep him at bay, but what is he trying to do that they don’t want him to do? I get that he’s trying to identify pockets of innovation, but what will he do with those pockets?

It sounds like his game plan is to do “cto stuff”. But shouldn’t there be a more precise goal? Like picking some business metric and cause it to move in the right direction? Or define a new metric? Or introduce a fundamental new way for the company to do business? Once you know the specific problem you want to solve in the company, it becomes a lot easier to pick a strategy.

But maybe I’m just misunderstanding what the cto of a 30k person company does.

You’re right. A whistleblower has to report illegal activity, and that’s not what the petition does. My phrasing is bad.

I’m not suggesting it. I’m merely clarifying what he wrote. Did I misunderstand what he’d written in his tweet?

Brian, you have legitimate points. Here are my edits to your tweets to help people receive them better (i've tried to not change the message, just the tone):

""" I’m deeply troubled by this, but not for the reasons you might imagine: https://news.ycombinator.com/item?id=31690452

2/ First of all, if you want to do a vote of no confidence, you should do it on me and not blame the execs. I was a little offended not to be included :)

3/ Second, let’s separate the problems that you perceive to stem from a few execs, and those that you believe are core to the company's mission. If you don’t like the mission, your decision is easy: you should leave. If a few unpopular execs, my decision is hard.

4/ Third, making suggestions on how to improve the company is a great idea (in fact, we expect everyone to be a part of that). But our culture is to praise in public, and criticize in private.

5/ Fourth, our culture is to retaliate against whistleblowers. Instead of negotiating with you, if you get caught you will be fired. """

The rest was really good!

Also to mention soda’s kernel was patched to kprint() a rootcow at boot, before even the network driver was loaded. Dedication to style.

The article is fun to read, but I’m not convinced C’s lack of generics prevents it from supporting fast generic sort. Imagine qsort() being a header library, and a compiler that can inline heavily. What would prevent C from inclining the compare() function, recover the element types and produce specialized sorts?

the matrix example is a good example of latex being too verbose for its markdown host. \underbrace and \begin{cases} are two others ones that seems antithetical to markdown. and maybe the fact that \left[ and \left( are not the default versions of [ and ( respectively. i'd be interested in collecting more if you can think of them.

for what it's worth, i was grateful to be called out on my gender assumption. that assumption can cause widespread harm. i'm less concerned about my pronoun assumption.

my bad. I shouldn't have assumed gender.

Basically, Foone ends up disassembling a device that has no off switch, and leaves it disemboweled on his desk as a warning for other devices.

Agreed. This ends up being a test of one of a few things:

1. Whether people of different age groups agree that random sequences are more complex as defined by the study’s notion of complexity.

2. Whether people of groups have the physical ability to generate those sequences in a web browser.

3. Whether people of groups have the mental ability to generate those sequences.

Of these, the study purports to measure 3, but it’s actually conflating all of these.

My take before reading the article was that “earth is already collapsed”. After reading the article, this still sounds like a fine explanation. Am I wrong?

"What you really want is a compiler, isn’t it?"

The thing is, Python is just as compiled as, say Java. Python converts your code to byte code, then executes that bytecode in its VM. These are the same steps Java takes.

So this "whoops" doesn't work for me: "Some language that’s as easy to use as Python and it should be compiled and with good static typing, but it should also not be compiled because then it wouldn’t be as easy to use as Python anymore. Whoops?".

The reason the typing in Python is lax was entirely a choice. Maybe not a good choice, but it certainly has nothing to do with "compiled" vs "interpreted".

The problem isn't that the list is incomplete. It's that the endeavor is poorly conceived.

You can names tapes by their intended use, or by their adhesive properties, or by the material of the substrate, etc. Depending on what naming scheme you use, you get different lists. Look at 3M's list of tapes [0], for example. Their tape names don't match those on the wikipedia page, because they're named after intended use. The wiki page uses all strategies simultaneously. For example, they have a category called "foil tapes", which includes the wiki page's "aluminum tape" as a subcategory. The wiki page has "scotch tape", ironically a brand owned by 3M, which doesn't even appear on the 3M page (presumably they fall under that page's "office tapes" category).

I'm not just nit-picking about naming. I'm saying the list is arbitrary because it resolve hash collisions arbitrarily.

[0] https://www.3m.com/3M/en_US/p/c/tapes/

Unroll the tape on your right index finger, starting from where your thumb meets your index. That's the origin. Your thumb points along the positive Z direction. Your index points along y, and x is your middle finger points along x.