mywittyname

Following the bugs in your candy theme: natural red 4 is carmine, which is made from crushed insects (which are gross looking, even for insects).

So the point I was trying to make is that I'm maybe in the 25th percentile of developers and am using multiple agents for a task, and Anthropic researchers are at the 99th percentile and are using thousands of agents per task.

So the median developer is probably using 5-20 agents at a time for a task. I'm not sure if that crosses the threshold of being a "fleet of agents".

I usually have a high-level agent drafting plans, then spawning sub-agents to build parts of the plan, other sub-agents to check the work of those agents. Then the PRs have different code review agents looking at the code, on top of the agents for people on my team also reviewing my code.

And I'm probably on the more amateurish end of AI development. As I learned these techniques from YouTube.

When you hear about the work Anthropic researchers are doing, they seem to be using hundreds or thousands of agents to solve problems.

Do you expect to be murdered? Repeatedly tortured and gang raped in custody while it's filmed? Disappeared to some other country?

Also, any of this can happen to you even if you're a lawful citizen.

The situation in the USA is dire and it's going to get, much, much worse.

Hedging won't help, so we might as well be blunt.

It still sucks to need another machine to play games that a MBP should easily be able to handle.

Especially give how good recompilation projects are anymore.

I think Ohio, Pennsylvania, and Virginia are the more interesting states. These are affordable places with solid job opportunities. So I think job migration is more complicated than just cost of living and taxes.

Also, it would be interesting to see these absolute figures as a percentage of overall income.

> Our American obsession with jail is nuts.

We rarely imprison the people who actually deserve it. Even when we do, it's barely a punishment if the person is wealthy.

Our jails are full, but a lot of that is we harshly punish mundane crimes against people with no real ability to defend themselves. And that's mostly because the grift of for-profit prisons.

Properly implemented, it's less friction than the current model.

Just like how Apple Pay is less friction for online purchases than entering my credit card info is. Take that a step further where Apple can supply personal information JIT for shipping.

Yeah, the workflows need manual overrides for things like different billing vs shipping addresses, etc. But the happy path can be streamlined.

> Edit: I am wrong

To be fair, this is a really common myth. Unless you actually read the BLS methodology, you wouldn't have known. And in most places, anyone attempting to combat this myth is met with down votes.

My salary took the dumbwaiter to the basement :(.

As an American who grew up in one of these underserved markets, you couldn't pay me enough to work there. My mom works at a rural hospital and loathes every minute of it. The reality is, American workers value more than just dollars. Especially professionals with options.

Immigrant doctors may also hate working at these facilities, but the prospect of raising their family somewhere in the USA is a huge value add for them.

I've stopped calling out Claude mistakes on team meetings because this is so true.

I mean, sure, I could have predicted in what ways an LLM would fuck up, but there's just so many ways I can't keep up.

We just had a major production issue because someone's LLM wrote queries against dev databases. Which are very obviously dev databases because they are labelled with dev in the name, and in the table descriptions. AI reviewer didn't catch it, neither did the human reviewer for that matter.

> AI cannot be held accountable.

And this unaccountability is transitive. The amount of time I've seen people successfully justify issues based on the fact that Claude/Astra/Codex wrote it is absurd. And it comes from the top.

We've had AI ship made up data to clients and tech leadership was like, "haha, that's AI for you."

> If you’re toying around, LLMs do a great job.

This too. We have a lot of business guys that develop tools with Claude that look like they work, then tech team gets pressure to deploy them immediately, because they assume that everything must be a prompt away. It's not (though, we do have some wizards on the team who make this true enough).

> AI overdose is a thing and it directly puts an expiration date on your skill set.

Agreed. But I'm not in a position to push back. It's not just managers, but tech leadership who are all in that on the fact that humans shouldn't program anymore. I still fully believe that I'm better than Claude / Astra in my specific domain, but people give me a hard time when my PRs contain what look to be human-generated code.

Overall, I agree with the article, but I'm still pretty sure I'm falling behind in my apprehension to fully trusting Claude/Astra and not giving into the approach of burning millions of tokens daily to generate PRs so large they break github (two of which I approved today).