alganet

What makes progress reports useful and career-helping are valuable notes, not game-like achievements. Those are hard to craft, and often not digestable to a larger audience.

I'll think the next steps for you. If things continue the way they are, it's likely someone will try to skip the influencer stuff and create automated systems that evaluate whether someone is really making progress on something novel or not. That sort of thing could change the way we look at developer presence.

In that world, you want to cultivate a smaller audience that can understand your notes, build upon them, and resonate value within that smaller audience. I'm saying a smaller audience because the larger it is, the higher the chances of it derailing.

Sure, Mr. Palpatine. Any time now.

https://github.com/fosslinux/live-bootstrap/

Other efforts around the same problem are trying to make it more architecture independent or improve regenerations (re-building things like automake during the process).

It's free and open source, you're welcome to fork it and try your best with the aid of Claude. All you need is an x86 or x86-64 machine or qemu.

The project and other related repositories are already full of documentation in the markdown format and high quality commented code.

Here's a friendly primer on the problem:

https://www.youtube.com/watch?v=Fu3laL5VYdM

If you decide to help, please ask the maintainers if AI use is allowed beforehand. I'm OK with it, they might not be.

A while back, I didn't even knew those problems existed. It took me a while to understand them and why they're interesting and lots of people spend time on them.

I have tried to adapt the problems to the LLMs as well, such as shaping the problem to look more like a thing that they're alreay trained on, but I soon realized the limitations of that approach.

I think in a couple of decades, maybe earlier, that kind of thing will be commonplace. People training their own stuff from scratch, on cheap hardware. It will unleash an even more rewarding learning experience for those willing to go the extra mile.

I think you're missing that perspective. That's fine, by the way. You're totally cool and probably helping lots of people with your work. I support it, it allows people to understand better where LLMs currently can help and where they cannot.

I can invest lots of time in Linux, for example. I don't know how to write a driver for it, but I know I could learn how to do it. If there's a bug in a driver, there's nothing stopping me except my own will to learn. I can also do it in a potato, or my phone.

I can experiment with free tier LLMs, but that's as far as I will go. It's not just about me, that is as far as 99% of the developers will go.

So, it's not uninteresting because it's boring or something. It's uninteresting because it puts a price on learning. That horizon of "if there's a bug in it, I can fix it" is severely limited. That's a price most free software developers are not considering worthy. There's a lot of us.

Lots of people don't have resources to invest in LLMs (either self-hosted or not). They rely on what other people say. And people get caught in the hype all the time. As it turns out, lots of hype nowadays is around LLMs, so that's where I'll go.

I was skeptic about LK99. Didn't had the resources to independently verify it. It doesn't mean I don't believe in superconductors or that I should have no say in it.

Some of that hype will be justified, some will not. And that's exactly what I expect from this kind of technology.

I can see this call now has a lot more tokens for the reasoning steps. Maybe that's normal variance though.

(I don't have a particular interest in proving or disproving LLM things, so there's no incentive for me to get a key). There was an ambiguous point in the "proof", I just highlighted it.

Can you remake the call explicitly using the value `none`?

Maybe it's not using Python, but it's using something else. I think it's a good test. If you're right, then the response shouldn't change.

Update: `auto` is ambiguous. It doesn't say whether is picking from your selection of tools or the pool of all available tools. Explicit is better than implicit. I think you should do the call with `none`, it can't hurt and it can prove me wrong.

> "tool_choice": "auto"

> "parallel_tool_calls": true

Can you remake the API call explicitly asking it to not perform any tool calls?

That got a highlight in the article, but not a lot of text to explain it.

> they have to fix a bug in an unknown codebase

Creating those scenarios costs a lot of time and resources.

---

I think the LeetCode stuff is just fine. There are some issues with it:

- People can cheat by being LeetCoders that specialize on solving LeetCode, not being actual day-to-day programmers. - You might end up with a monoculture team.

Overall, though, it seems to be the safest approach.

"dirty ass programmer" contrasted with "beautiful woman" is a stereotypical example, but not a rule. It represents an asymmetry in typical social interactions. If you're getting much more than you deserve, it's a trap. Got it?

Same with friendships, all sorts of things. Someone that's super into the same things as you and appears to help you and encourage you like a friend you never had? That's not a friend, guarantee you.

There's a bomb in the bus. It will activate once we go past a certain speed. Once it activates, we can't go below a certain speed.

Also, the bus driver is injured, and a random lady with no driver's license or bomb-defusing experience is now driving. Furthermore, a part of the track is missing. Also, all passengers are hysterical.

That's more like what's happening right now.

Promise: AI will change the world.

Delivery: 1000 year old vice.

But who am I to say anything, right? We have things that are reminiscent of "open your eye":

https://imgur.com/a/r0MtyVN

And we also have "linkedin" which is pretty much the chain idea.

So I guess anything goes nowadays.

Is there any risk of the laser going through the glass and damaging parts of the machine? Or things such as refraction?

As I understand, those machines have a sheet of materials they can work with. This is probably the safest way to go: just stick with the materials the laser is made to safely operate on.

I'm not saying anyone predicted those or something. It's just that the notion of doomsday is quite vague.

I'm trying to broaden some notions here. Prediction might not be exactly absolute prediction, and doomsday might not be exactly absolute doomsday.

Perhaps some great threats were averted precisely because someone predicted them (for example, the great leaded gasoline poisoning).

There's something about this idea of experiencing art as a sommelier for human feelings that disturbs me.

This has happened many times with me. Person recommends a movie, describes a scene and how that scene makes him feel and that's why the movie is great. I watch it, and what I feel has nothing to do with what I heard from that person.

There's a joke in the metal community. It goes like this:

"I never respected Dragonforce. Then I heard the Metallica guys saying how they were impressed by the new Dragonforce song. Now I don't respect Metallica."

I think it conveys well how sometimes we can't project our experiences into other's perspectives. Maybe I don't feel the sublime, and I never will, and I'll never understand this article (although I'm sure lots of people will relate to it).

However, there is a chance he could be underestimating that audience, or at least part of it.

Finding a new type of comet is a scientific breakthrough, and I think his work points in that direction (still a guess from him though, but an educated one). He is trying to cake up those potential genuine discovers with sloppy sensacionalist makeup on top, and that's why I call it a circus.

If in a few months we confirm that 3I/ATLAS is a new kind of comet, he could use the papers he wrote to say he found evidence of that new type first, and also described its landmark characteristics. It would "legitimize" him. But the alien stuff would probably continue to be garbage. He can then say the scientists were skeptics, but he was right.

Now, what angle the aliens narrative serve? Why would a scientist subject himself to being a clown? I don't exactly know. In his case, I don't think it's good stuff.

I chose Tabby's Star to satirize him because my description of a mothership deploying an origami-like occluder matches the overall conclusion from the research at the time (a disturbed exomoon). It's an object from that system that changed is shape. In fact, "disturbed exosatellite" and "unfolding mothership from a planet" are quite compatible descriptions. What matters here is epistemology (we can't know if it's natural or not). Also, it's a good demonstration that we (general public non-astronomers) don't need his antics to imagine things.

> I suppose you include a script to update staging

Essentially, yes. But if I write a script to do it on staging, why don't I write one for prod as well? Maybe I could even go beyond that. Make a simple tool to handle those scripts for me (that's migrations).

It requires some version control and deployment discipline, which is something you probably wouldn't have if you're adding columns directly on the prod db.

I learned very early that no developer should have access to prod db. Not even read access. No writing, definitely. Schema changes by a single developer outside version control should be impossible.

Have they gone through the experience of having a persistent staging environment that slowly drifts from production (1)? If they haven't, they can't possibly understand why that is a bad idea. I'll just go with the flow until they realize. Maybe I'll hint at the possible issues sometimes.

Have they figured out that copying prod is a bad idea (3)? If they haven't, same thing. They can't understand why that sucks and why that's not true reproducibility.

Finally, (2). Fixtures! It's also a journey. There are so many things that can go wrong. Knowing those things depends on having gone through those journeys with a persistent staging and production copies.

There is no relief from the pain. No magic bullet. No product or solution that will ever solve this. You have to go through those stages. If you're lucky, someone will guide you through them (in practice). The journey can be sped up, but I haven't seen a shortcut that works (like forcing the team to adopt a practice without them internalizing it).

One of them is personal privacy. For example, an activist being individually targeted.

The other is behavioral targeting, which has no business in catching criminals. It wants to know how large flocks of people behave online.

If you could just spam annoy until you win, we'd be all dancing to remixed versions of Macarena.

---

Side note:

For some reason, the Hammurabi laws always reminded me of the old Sumerian city laments:

https://etcsl.orinst.ox.ac.uk/cgi-bin/etcsl.cgi?text=c.2.2*

There's something about the structure that looks, to me, very similar.

The IT world is waiting for a revolution. Only in order to blame that revolution for the mistakes of a few powerful people.

I would not be surprised if all this revolutionary sentiment is manufactured. That thing about "Luddites" (not a thing that will stick by the way), this nostalgic stuff, all of it.

We need to be much smarter than that and not fall for such obvious traps.

An identity is a target on your back. We don't need one. We don't need to unite to a cause, we're already amongst one of the most united kinds of workers there is, and we don't need a galvanizing identity to do it.

To a more naive, metaphor-blind audience, your mention of Don't Look Up makes it look like the scientists are warning about an alien comet and I'm the one ignoring it.

I'm very familiar with apocalyptical narratives of all kinds, but what I'm approaching here is much different. I'm talking about the integrity of scientific endeavours. In particular, space exploration endeavours.