They're not the biggest, but big enough to have a lot of active accounts, so I think they're likely to persist and get more than one shot on goal. (Similarly for Bluesky.)
I imagine with coding agents, maintaining private forks (reapplying patches on upgrade) will be a lot easier. Though, a plugin architecture would be better, where feasible.
If there there's a big enough community swapping patches that upstream isn't accepting for some reason, that's when a public fork becomes reasonable. (This is the Apache web server's origin story.)
The big challenge for me so far has been about setting up "breakpoints" with sufficient prompt adherence, i.e. conditions for agents to break out of loop, and request actionable feedback, rather than pumping as many tokens as possible. Use cases where pumping tokens in unsupervised manner is warranted, are far and few between. For example, dataset-scale 1:n and n:n transformations have been super easy to set up, but the same implementation typically doesn't lend nicely to agent loops, as batching/KV caching suddenly becomes non-obvious and costs ramp up. Task scheduling, with lockstep batching, is a big, unsolved problem as of yet, and Gas Town is not inspiring confidence to that end.
That was very weird. In the post where he was arguably "shilling," he seems to have signposted pretty well that it was dumb, but he will take the money they offered:
> $GAS is not equity and does not give you any ownership interest in Gas Town or my work. This post is for informational purposes only and is not a solicitation or recommendation to buy, sell, or hold any token. Crypto markets are volatile and speculative — do not risk money you can’t afford to lose.
...
> Note: The next few sections are about online gambling in all its forms, where “investing” is the buy-and-hold long-form “acceptable” form of gambling because it’s tied to world GDP growth. Cryptocurrencies are subject to wild swings and spikes, and the currency tied to Gas Town is on a wild swing up. But it’s still gambling, and this stuff is only for people who are into that… which is not me, and should probably not be you either.
In the next post he said he wasn't going to shill it any more, and then the price collapsed and people sent him death threats on Twitter. It probably would have collapsed anyway. Perhaps there was supposedly some implicit bargain that he shouldn't take the money if he wasn't going to shill? Well, there's certainly no rule saying you have to do that.
I think he's not very much to blame for taking the money from degenerate gamblers, and the cryptocurrency idiots are mostly to blame for their own mistakes.
> I think he's not very much to blame for taking the money from degenerate gamblers, and the cryptocurrency idiots are mostly to blame for their own mistakes.
I empathize with the disdain for crypto idiots, but I still think the people running or promoting these scams deserve most of the blame. "There's a market for my poison" is every dopamine dealer's excuse.
“Degenerate gamblers” is the kind of stigma that stops people and their families getting help for addiction. Even if you believe it’s a moral failing, the families deserve better.
Very true. Although, I wonder how much of that sort of thing was going on in this case? Did people actually bet money they couldn't afford to lose on this crazy scheme?
Yegge wrote in his blog post (viewed by many who ended up buying in) that it is an investment and that he wishes the investors will become "filthy rich". He wrote the post as an introduction to the concept of BAGS for an audience that is unfamiliar with it. He onboarded people to the platform and to his pump and dump scheme (in which he pumped, and dumped, then announced he's walking away from it).
You left out that part of the post and only mentioned the disclaimer he added at the top after he got pushback on his messaging. Are you influenced by his celebrity?
> I think he's not very much to blame for taking the money from degenerate gamblers, and the cryptocurrency idiots are mostly to blame for their own mistakes.
So drug dealers are not to blame for taking the money from degenerate addicts! Let's free everyone and disband the DEA, we'll save billions of dollars.
Oh wait nvm this line of thinking only applies to sv people
Maybe I'd care about his opinion if he didn't take the money. I consider this worse than OSS taking VC money. At least those don't have a scam auto-builtin to the structure beyond normal capitalistic parasitism.
Also, 275k lines for a markdown todo app. Anyone defending this is an idiot. I'll just say that. Go ahead, defend it. Go do a code review on `beads`. Don't say it's alright, but gastown is madness. He fucking sucks.
> People who love using AI to create software are loving it because they don’t value the act of creating & understanding the software.
This happens, but it's only one way to use a coding agent. I'm working on a small, personal project, but I ask it to do "code health" tasks all the time, like complicated refactorings, improving test coverage, improving the tooling surrounding the code, and fixing readability issues by renaming things. Project quality keeps getting better. I like getting the code squeaky clean and now I have a power washer.
You do have to ask for these things, though.
Some people like using hand tools and others use power tools, but our goals aren't necessarily all that different.
Indeed, one of my favourite things about coding assistants is that I can now get an easy code review on my personal projects, or once I've thought through my approach have it think up alternatives I may not have stumbled on.
I've found it very unsatisfactory (both experience and results) to use them to replace code production. But in terms of augmenting the process - used to critique, explore alternatives, surface information - they're getting really quite handy.
My favorite thing is just being able to talk through the code and problem and have someone right there too response even if it not 100% right it still gets you to think and it nice to have it push back on things you ask ect. It basically a co worker you can bug all day and everytime they are still happy to help.
In reality there are tons of tasks at work that are boring and time constrained. There are days I don't enjoy it, and days I do. It's not binary - I still love programming by hand but at times I let Agents work whilst reviewing the results.
It's a decent analogy though: there are many tasks for which a CNC router would be an inappropriate tool choice (e.g. you can cut dovetails on a CNC router, but it's a pain in the ass and you probably could have cut them by hand in the time it takes to set up the cut on a CNC), but for tasks where the CNC router is a good fit, very few tools can beat it.
It works fine for webapps and other slop-adjacent projects.
If you try to do anything outside of typical n-tiered apps (e.g. implement a well documented wire protocol with several reference implementations on a microcontroller) it all falls apart very very quickly.
If the protocol is even slightly complex then the docs/reqs won't fit in the context with the code. Bootstrapping / initial bring-up of a protocol should be really easy but Claude struggles immensely.
> (e.g. implement a well documented wire protocol with several reference implementations on a microcontroller)
I have had an AI assistant reverse engineer a complex TCP protocol (3-simultaneous connections each with a different purpose, all binary stuff) from a bunch of PCAPs and then build a working Python server to speak that protocol to a 20-year-old Windows XP client. Granted, it took two tries: Claude Opus 4.1 (this was late September) was almost up to the task, but kept making small mistakes in its implementation that were getting annoying. So I started fresh with Codex CLI and GPT-5.1-Codex had a working version in a couple hours. Model and tool quality can have a huge impact on this stuff.
The sloppier a web app is, the more CSS frameworks are fighting for control of every pixel, and simply deleting 500,000 files to clear out your node_modules brings Windows to its knees.
On the other hand, anything you can fit in a small AVR-8 isn't very big.
Yep, but I don’t intend to let that happen to my web app! It’s not that big and I intend to keep it that way.
Dependencies are minimal. There’s no CSS framework yet and it’s a little messy, but I plan to do an audit of HTML tag usage, CSS class usage, and JSX component usage. We (the coding agent and I) will consider whether Tailwind or some other framework would help or not. I’ll ask it to write a design doc.
I’m also using Deno which helps.
Greenfield personal projects can be fun. It’s tough to talk about programming in the abstract when projects vary so much.
I've been working with an agent to make a web-based biofeedback "application" which is really a toolbox of components you can slap together to support
- heart rate via Polar H10
- respiration rate via strap-on device
- GSR and EMG via arduino + web serial
- radar-based respiration (SOTA says you can get R-R intervals as good as the H10 if you're not moving)
and even do things like a 2 player experience. The code is beautiful, pure CSS the way it was supposed to be, visualizations with D3.js. I do "npm install" and can't get over the 0 vulnerability count. It's coding with React that's 100% fun with none of the complaints I usually have.
Given the amount of Arduino code that existed at the time LLM's were trained, I would have to agree that AVR-8 might be fine. For now it's on the Cortex-M struggle bus.
The experience of using a coding agent is that you're more of a "backseat driver" though. The AI acts as your driver and you tell it where to go, sometimes making corrections if it's going the wrong way.
The experience is what you make of it. Personally I'm quite enjoying using AI as a way to generate code I can disagree with and refactor into what I want.
reply