visarga

> As in, what work or enjoyment does someone actually get from this? Why do we look at art, watch videos/movies?

I like to generate songs from obscure poems

I agree 10K LOC is too little, I was saying "a few 10,000's of LOC"

A pluggable loop with events and hooks.

Yes, a 15 year old code base written by hand obviously, not by a coding agent from the start.

At the same size I only heard of AI generated projects reimplementing compilers and browser engines which come with amazing tests written by hand, but even these projects generally don't become reliable enough to use.

There is a limit, after a few 10,000's of LOC ai projects start disintegrating and becoming unmanageable.

Funny you mention Chinese Room and LLMs in the same response, I would say LLMs proved Searle wrong, agents now make cutting edge discoveries and meaningful problem solving. They not lookup tables though and you need to pay for inference, so the intuition of syntax doing the work of semantics without understanding was wrong.

AliExpress and Temu would refund you if you complained about a delivery. Zero risk.

> If you want to follow a process or a checklist you probably shouldn't use an LLM

I like to externalize tasks as markdown files with checklists, they are still planned by agents but I can pass the plan around to judge agents and fix some errors before implementing.

I also have the coding agents comment on each closed checklist item, so the same file becomes a log of what happened. This goes to the implementation judge. I can also switch agents anytime, or resume a task days later no problem.

I am avoiding internally provided tools for todo lists and planning because they do not leave the same artifact trail which makes judging with separate agents easy.

I can't get an answer from Fable on "How does digestion work?"

Used to straight out bail to Opus on this, now it's "Honing" and "Pondering" for 10 minutes on it. Then I get a bunch of "This response didn’t load." and it still churns ahead.

> I’m sticking with gemini for now given the 20TB cloud storage and youtube premium that comes with it

I used to have the Google One with Gemini and Drive space and YT Premium separately for my family. A credit card expiration lapse lead to closing YT Premium and being locked away. Why? it was because Google One plan bundles YT Premium Lite as an extra. So they actively blocked me from getting Premium back for a month.

Now I moved to YT Premium on my wife's account and downgraded One to lowest tier. Never heard of a company forbidding users to upgrade their plans before this. I suspect Google really wants users on Google One no matter what they want. I was barely using Gemini anyway, already have claude and codex plans.

> That forces people to look beyond the walled garden.

Every time I get "you used your quota, come back in 3 hours, or 2 days" -> that is experimentation time with their competition, leading to changed service plans. When they said "claude -p" will be billed at API pricing even for plan users I moved my harness off claude. After I integrated codex, then it was never going to be a full claude project again.

What business encourages users to try their competition and adapt their usage to the competing products?

What you are saying is impractical, we cannot uninvent AI or force all countries to abandon it, a simple laptop is sufficient compute to run it. Licensing to access AI cannot be implemented. Why stop at AI, maybe we should license people to use the internet as well? Own computers? Electricity could also be a powerful weapon, should we only trust some people to have electricity?

> Who wants to take 101: Hit Approve all day on your 20 claude terminals?

Obviously put all 20 in tmux sessions and have a monitor agent do the approvals.

> they don't care about the negative effects [...] rush to secure whatever advantages they can, regardless of the negative externalities

I think we don't yet know the long term view - what is harmful or damaging in the long run. Too early to tell and pass judgements. Every new technology from writing to internet had its detractors and they all pointed out negative externalities, as they seemed to appear at the moment.

Should have let the kids use AI during exam too, later in life they will be solving tasks with AI not manually. Just make the test questions difficult enough even with AI.

I see our current approach like a driving school training its students at fast running.

> I don’t mean to imply that making things with these tools is lesser, or that it doesn’t require some skill, but it is fundamentally different to programming.

Exactly, it is a skill, and it is different from programming, but in the age of cars why train to outrun them on foot? train to drive the car.

Using AI for hours and hours every day has got to leave a new skill set in the human users. It's hard to sit there and steer every minute, just the volume of verbiage and speed of change is amazing. That time and effort spent into AI steering is your new training.

> There’s no reliable record of human intent. Prompts are discarded

I agree, but I just log all user messages in a chat_log.md and review them regularly with an agent to detect deviation from user intent.

I find it bonkers most coding harnesses simply discard one of the most valuable streams of information, those user messages can also identify harness weaknesses and project issues by reflection.

Yeah, great logic, that way they are sure there are no extra copies around.

If AI is so great it can do your job it is good enough to provide for your needs directly by automation. Just buy a robot and a plot of land and you don't have to worry about jobs.

App Store is a wasteland, I don't think there are more than 100 good apps in millions. All want in-app-purchase, none is fair to give me an upfront price. I don't want for example my kids being bugged to buy apps that cost $100/week.

Say you invent a crap product, and shove ads under my nose to bug me while I want to buy a decent product. Why should I accept that?

Just get your products reviewed by trusted reviewers, if you can budge them maybe you deserve a bit of attention from us. Someone who has a stake in being trusted for their fair reviews.

Ads are a scourge.

No wonder Amazon sued Perplexity to forbid agents on their site, agents could cut through the crap and find what users wanted instead of what Amazon peddles. They even won the suit.

I think our only hope against platform enshitification is using agents that optimize our interests not theirs. They don't deserve human eyes on those SERPs.

I think the Opus 5 formula is to be the little professor treating your ideas like an essay for grading, or like a buyer analyzing merchandise for purchase.

I would download it if I could, no question about it. And 64GB more RAM if I am at it.

It is probably skill loss on one side and gain on another. In a few years it is going to matter even less if you can code manually. The hard part now is understanding what agents do. How can you gain a bit of confidence in their work? I think it relates to testing, you can trust the parts you test.

Medium term it could lower the price of memory and compute.

It is indeed well written but I feel it has LLM smell, not that is a bad thing.

> When folks like uncle Bob says they don't review AI code, I can only think they are burnt out or being unprofessional. Indeed it saves some time writing the code, but overall, I think we just moved concerns from one place to another.

I think you might be getting yourself drunk with plain water here ... reviewing a code is just vibes, "LGTM" type of vibes from a human instead of AI, but not better than vibes. Yes it might catch some implications or bugs if we are lucky, but it is not a reliable way to verify code.

my idea is "skin is all that's left"

I think a product should answer a specific need, it cannot be fully defined internally, it is an expression of some need in the environment which makes it a legitimate product. A product with no viability soon ceases to be a product. Generating code is not the same with answering a need, it is orthogonal. Good products have market fit, not necessarily are themselves well written apps.

The vibe-coded demos everyone's tired of are code with no environmental need pulling on them.