This is an amazing article. The problems it describes are exactly what we found when building a production system that used LLMs to (most of the time) produce reliable results. Extensive tests are necessary, and stakeholders have no idea how their suggestions fail in production. They just see the handful of times they tried something and had it work, not the long tail of cursed results. (“How hard can it be? Why don’t you just…”)

We didn’t get to the point of self-built prompts, as the article suggests, but it’s an intriguing idea.

This guy is on the ball with the problem. Totally correct: Like my friend Coda says all the time: "the textual nature of prompts leads us to take the intentional stance towards systems which aren’t conscious, and thus miss the essential nature of their non-meaning."

I don't know if his solution (""We should all go insane building interlocking evaluation and optimization pipelines, instead.") would be the long-term solution. Instead perhaps something could be trained into the models, i.e., he is describing a process at inference time that could be done at training time. To make their weird errors less frequent / make them more human.

cf. https://arxiv.org/abs/2008.04071 "On Controllability of AI" However, as I said, you can't make it perfect but you can make it better. (You can't make humans fully aligned with human society's interest anyway, including the humans controlling the nukes.)

> I don't know if his solution (""We should all go insane building interlocking evaluation and optimization pipelines, instead.") would be the long-term solution.

TFA could explain this one part better I think. The whole process proposed is real with lots of stuff in the literature, but by definition NOT a long-term solution in the sense that this process actually has no end. None of the approaches can get you a static answer for a moving target/platform.

So the "interlocking pipelines" for eval/opt would not be some stepping stone you can throw away, and they aren't something you'd run periodically. They'd basically be always on forever and spending 10-100x on system complexity and on tokens. Unless of course you're ready to freeze everything else about the whole system forever (including the backend model, and the whole nature of the "average" context window, the plugins/other prompts in the mix, etc).

Are most people in position to freeze requirements/platform forever? Not really, because if they were they'd just build a fairly static system and probably have limited use for AI. Are most people in a position to just casually accept 100x complexity/cost? Not really, that's the "it's not yet webscale" kind of advice that sounds good but isn't necessarily reasonable for average use-case or average org. Since specializing your own locale for this is usually a mistake.. the likely future direction is eval/optimization as a service

Were the people you were interviewing already unemployed? Work trials are harder when people are already employed. There can be legal issues as well as scheduling ones.

If you’re doing traditional on-site interviews, you’re already asking for multiple days (unless they live nearby).

It can be a challenge sometimes, but not as often as you might think.

I’m sure there are heavy-handed employment contracts that prohibit side work, but hasn’t been a problem in practice.

This will always be the problem with trials. Anyone with a job, often underemployed between software roles these days, will be excluded.

Still, it does benefit the hiring company, just not the applicant in these situations.

> Still, it does benefit the hiring company

Not sure it benefits the hiring company when they are artificially reducing the pool of applicants to only people who don't already have jobs.

Not that there's shame in not having a job, especially in today's market (shit happens, I'm sure lots of very qualified people are currently looking), but there are solid statistical reasons why social proof of already being employed is seen as a positive.

Net benefits perhaps. It absolutely lets the hiring company get more information by increasing the amount of time spent by the candidate before a commitment occurs.

I have to agree with @fleshdaddy. The explainer page itself looks like slop. I haven’t seen the original project but I’m afraid this makes me less interested, not more.

So funny, I didn't even expect the page itself to be such a blocker/issue!

I replied to another sibling: Based on your feedback, I scrapped it all and started over with the actual, unredacted Claude Code transcript plus a personal letter.

https://alexisrondeau.me/mosey-site/conversation/ (Feel free to skip the letter and the final cut)

How does this feel?

Thank you as well, just wrote below to the same point.

And I agree with your impression.

My goal here is to understand how can I demonstrate that I did in fact put work and care into the "thing" (video in this case).

Basically show that I did spend energy over time, that the final product took a few dozen iterations with a lot of dead ends (as opposed to a lazy one-shot "make good video, no mistakes")

So I do think this route can have merit given there is actual work, but how would a page like this look if the current one is not yet giving you a trustworthy vibe?

Idea: Just show the full Claude Code transcript on the left side and the incremental outputs (images, clips, videos) on the right side?

You'd definitely see some raw and embarrassing stuff there. Like frustration, anger, greed, ignorance, incompetence, creativity, curiosity.

Is that maybe what we need to see to believe that a human (me n this case) was involved meaningfully?

You’re almost there.

People shit on things that are low effort, and especially low-effort trends. And especially especially when the low-effort trend involves something that offends them in some way, for example if they think AI is taking jobs, hurting the environment, etc.

Of course people are shitting on low-effort AI. It would be more surprising if they didn’t.

Ironically, Comic Sans is more effort than Times New Roman.

But people would shit on a Comic Sans poster more than a Times New Roman poster.

> free cooling

Vacuum is an insulator. I think you may have been mislead.

Perhaps you've heard of radiant cooling?

SpaceX has already deployed thousands of functioning computers in space. I think if the vacuum were an issue, we would have heard about it.

it's not a trivial engineering challenge to scale that into space. we're not talking a single sattelite we're talking 100k GPUs and servers

Mislead or … work at an Elon company. Shudder.

I appreciate it when people say something is slop. Saves me from wasting my time looking at it.

How many slop ideas have come to the front page, never to be heard from again because the execution isn’t actually any good? I’m guessing most of them.

Just because the ReadMe is slop, doesn’t mean the code is slop. People are starting to make apps for themselves now and open-sourcing them so they’re not putting much thought into the ReadMe or distribution.

This just means ReadMe’s are less important now. I just have my terminal agent dig into the code and tell me what features are there. If the app is actually useful.

Even before AI, there were so many projects with subpar ReadMes, no screenshots, etc. But once you use the software, you realize how good it is.

Source: I maintain a massive collection of open-source alternatives and quality of open-source alternatives have increased a lot

If you do not respect your project to write your readme yourself chances are i will not care though.

For many new coders, LLMs are so good at writing the code, asking it to write the ReadMe sounds like a good idea. Clearly it’s the first impression your project makes so handwriting it is important.

Being turned off by the project because of the ReadMe is your prerogative. I’m just suggesting you dig into the code sometimes, the ReadMe is not the be all, end all.

> For many new coders, LLMs are so good at writing the code, asking it to write the ReadMe sounds like a good idea. Clearly it’s the first impression your project makes so handwriting it is important.

The point is that if they looked at the result they'd question themselves. And the fact that they don't is a violation of the social contract: I can hardly be expected to care about your work, if you don't.

But it takes so little effort to wade through an LLM readme and cut out or edit it down to something palatable.

Just the bare minimum effort to do that would be nice.

I apply the same rule to the code, unless it is on the job and I have to follow along with the herd, or else....

Your comment might be sarcastic, but AGENTS.md already exists for the explicit purpose of agents reading and using it.

yeah, my point is just that we agents can have their own docs and we can also have a nice curated experience for humans.

You’re right that a poor quality readme doesn’t mean a poor-quality product, but it seems more likely than not to me.

Slop is an instant tab close for me. If something’s good, it’ll come around again. I’ll catch it when there's some evidence that it’s worth my time.

You can produce good code and good projects with Claude doing all the work. This might even be an example of one, but the description on GitHub makes so little sense that I stopped reading before I figured out what it even does.

If the author is too lazy to even write their own commits, I doubt there would be any effort put into the codebase.

Hmm, I wonder if we can automate that

Texas has no income tax. It makes up the difference with sales tax and an unusually high property tax.

That seems nice. I'm not sure why some guy says on a different comment "People think TX is some libertarian utopia because it has no personal income tax on a W2 salaried person, but do the full cost of living comparison with property taxes and other taxes, vs what you would calculate as a total in states that do have income taxes (such as OR).".

For reference, I pay about 0.1% property tax (live in Romania), while income tax is about 43%. For $100 I make, I get to take home $57, forget all the "you're only paying part of tax, employer pays another. No, I pay everything because without me there's no income, I can't believe some people still fall for "Out of $100 I take home $1 but I live in a tax paradise because I only pay 1% tax although company pays 98%. But that's the company, not me".

So while the level of income tax is definitely nothing I'm happy about, at least once I draw a net salary, it stays on me. At least I don't have to rent my own apartment from the state after I already paid an arm and a leg to the bank for it in real estate mortgage. Same goes for land and I mean arable land, of which I own some 10 hectares (well shared with my 2 brothers). Not only I'm not losing money on it, I actually get subventions for cultivating it (again, outsourced, I'm not working it myself but lending it to someone who does it, while the terrain retains / increases in value - never in the history of humanity has arable land in a viable area decreased in value over time).

"A 10-hectare (about 24.7 acres) piece of land in Texas typically costs between $40,000 and $430,000, depending heavily on the location and region." says Google AI. Well, where I live it's around $100k. At least I wouldn't sell it for less, not that I would sell it. If I sell it, I have money for a day then inflation and bad investment makes it disappear. Not to mention that it's a highly asymmetric market: I can sell the land in no time but even if I have money, A LOT OF MONEY, people won't sell. Maybe for 3x-5x the market value, some might be tempted. Which suddenly raises my non-liquidation value of the land to about how much it values in Texas.

Income tax scales to your income, so you can always pay it. No income, no tax. Property tax is unforgiving.

Given the backlash amongst gamers against AI, how does being an “AI gaming” company work?

It only works if one measures their success in Twitter likes.

I look forward to this and the new technology that advances for Ai gaming.

>Given the backlash amongst gamers against AI, how does being an “AI gaming” company work?

They are delusional or know nothing about how development works. I can promise you there's not a single major game studio left out there not leveraging AI in significant parts of their workflows at this point.

Brown dogs chase cars != all dogs chase cars. And especially, != black dogs chase cars.

80s hobbiest programmers are GenX, and maybe most execs are GenX, but that doesn’t mean both groups share the same opinions.

Slop, and with nothing new to say. The “something” is the AI bubble.