gravypod

> Also the amount of commits is suspicious. In the last two months, rsync had about as much commits as in the last two years before that.

I wonder if the data looks worse or better when not doing per-10commit and instead do per-commit.

This is a really cool post but I think one metric we may want to also look at is does using agentic coding tools in one domain impact your coding abilities in another domain? A lot of people I know have been talking about getting rusty on the fundamentals recently. This is not something I am particularly feeling as I do a mix of running agents in parallel and writing some code manually where it makes sense. But if people who have been prompt-only at work come home and work on rsync and are more "rusty" maybe that could also lead to more bugs?

This would be even harder to measure.

> CI/CD?

That is less useful when the changes are editing the tests but we don't know if a human has validated the assertions.

> My non-techie friends send me screenshots of ChatGPT. I guess that’s a modern micro aggression?

I think the concern I have is explicitly not the sending the chat logs. I think it's this flow:

1. Ask a question

2. Get an answer from a team member.

3. I don't like the answer and instead of discussing I am going to go to Claude and ask the same question.

4. Copy/paste the answer into chat without seeing if it includes novel information.

In one case the engineer was asking which model to select in the agent framework we are using. I gave an answer and provided a list of reasons. They did not like this answer and asked Claude which gave the same answer.

The answer was something inherently obvious and that anyone should be able to derive from first principals.

I did read the other replies. I don't think my comment is that LLMs are bad. I use LLMs and agents for work. I think my "oh shit" moment is the dynamic that giving someone LLMs amplifies their impact (positive or negative).

For example, some people give kids tiny go karts and that's acceptable because the damage they can do with a very tiny battery powered 4 wheeler is minimal. We now live in a world where everyone has access to a tank and can plow over everything.

I think LLMs will increase anti-social behavior.

I work with someone who is very AI-forward, high confidence, and very low execution. He has started sending me large PRs of AI slop that he assured me doesn't need to be reviewed. I quickly find many minor issues from an initial pass of one of the reviews. He gets mad at the team for slowing him down.

He also will paste chat logs with Claude into our team chat. Often Claude will say the same thing I told him but he either doesn't remember or doesn't trust human engineers now.

He has spent months working on agent skills and prompring.

He has not landed anything in 3mo, and has landed nothing useful in ~1 year.

This will be the rest of my career. Working with people in ai psychosis and trying to stay productive.

What mod did you build?

If you had a business task to complete that was only possible with ai and it cost you >$1500/month of work, how long would you have to delay the task so that it's cheaper long run to buy hardware and do local models?

$1,500/mo * 14 months = $21,000.

If local models are 14mo behind as many in HN say it may be profitable to just wait. Maybe just spend a few hundred dollars of your tokens and buy hardware piece by piece.

Anything commissions based might fit the bill?

Just a note, P40 came out at $5700 in 2016 dollars. In 2026 dollars that is $8000 (wow!). If you bought 100k today, assuming a 1% failure rate per year your $800M investment can be traded in for about $30M.

I think it is reasonable to assume a similar depreciation in GPUs.

Meaning you'd need to have made more than (800M - 30M) * (1 + income tax rate) + (power + maintenance).

Some say the margines on inference are already there for new GPUs but they are right margines.

It's, from my understanding, a little bit of both. There's a failure rate of GPUs and fans. There's also changing in standards like PCIe and software stacks.

LLM inference is mainly memory bandwidth constrained so I think it's highly likely that a company will create silicon with just an insane number of memory chips and less compute. These ASICs will probably do the same thing the crypto ASICs did.

If we look back 1 decade, no one uses a GTX 950 for anything.

Where is this list from?

We did allow people to send us letters with QSO logs on paper.

Warning: This is not medical advice, I am a nerd on the internet not a doctor

For what it is worth different countries have vastly different recommendations for HBP and these drugs. I recommend discussing with the pharmacists in your country.

In the US I have been told it's a strict "never", in Ireland I was told that it wouldn't have a measurable effect on blood pressure. I've also measured my personal blood pressure (pre-hypertension to stage 1) and have not been able to measure a difference in blood pressure.

> with small amounts leading to being detained for 23 days (like in this blog post)

This seems ultimately like a very bad sales pitch for the tourism industry in Japan. I had thought I wanted to go to Japan but if I can accidentally, without malice, be thrown in a prison for 20 days that seems like a bad system.

I can't imagine the international relations of the ruling classes of various countries to the UAE would be trending in a positive direction if they arrested and punished people for walking off a plane with airplane bottles of alcohol.

People *did* write down these logs, manually, and submit them.

Ok, that makes sense. I am assuming it is not common to be arrested for possessing this kind of over the counter remedy?

My first "job" in computing, where someone else paid me for code, was in a research context where we were modeling radio propagation. Nothing about that was removing human labor. It in face eventually called for a bunch of humans to interact with each other. See: https://www.hamsci.org/basic-project/2017-total-solar-eclips...

I don't think it is fair to claim computers are about putting people out of jobs.

What do you do when you have a bad flu or cold if you don't have pseudoephedrine?

Yes! That would be fantastic! They would need to approach many people, each having the ability to question the motivation! Or, they would need to convince a judge and obtain a warrant.

This is the disaggregation of power of surveillance.

One would be scope. There's a big difference between a security camera next to a secure facility (bank, police evidence facility, school) and a 1 mi radius circle around that facility. Security cameras around a bank only track stuff within a field of view from the bank. A cell geofence could be millions of people if it's drawn in midtown.

Another would be incentives. There's no reason to collect cell location data for everyone if you aren't able to use it for anything. I think just the fact that we are all monitored constantly is its own violation of our rights. We should have laws banning these practices.

The side effects of spending funds on these mega projects is also something to consider. NASA spending has created a huge pile of technologies that we use day to day: https://en.wikipedia.org/wiki/NASA_spin-off_technologies.

I've heard this brought up many times but I don't really understand the implications of it. I can't really think of a system of government which hasn't killed large numbers of people. Even the USA's early wars like the civil war have death tolls >500k. We also lose ~7k homeless people every year.

Removing the contexts of those deaths makes it very difficult to evaluate the true causes and if the political ideology is to blame or if the centralization of power common across all governments leads to deaths.

Do you have any recommendations for science fiction books that explore interesting ideas?

There is no Antimemetics Division was really interesting in how some of the scenarios play out. I don't read much but I've been trying to do that more. I really liked the book.

Things like the memory consuming entity, async research, etc I enjoyed.

What I am saying is if your system lets me pay $x/token and open router lets me pay $y/token if x<y then someone could make money just by providing those tokens through the open router API. That would either drive up demand for your systems increasing costs or drive up supply on open router decreasing costs. Eventually the costs would converge, no?

Why wouldn't developers just do llm arbitrage against openrouter if it is a better deal?

I really can't think of a better way to respond to this situation. It is clear to me that over the next decade the amount of people who will have been hot-headed kids on the internet who grow up to fully-fledged adults who have said they no longer agree with things in ways that are not kind is going to be a lot higher. I've no doubt said things that I no longer agreed with that made sense in the context of when they were posted.

Thank you for being a good role model and setting the example that saying "that was bad, here is the context, but I don't like that I said that."

Sorry, my comment was not clearly phrased looking back. #1 is an example of an elected official who can hire/fire. #2/#3 are not. These are all from recent litigation (in the court room and within public discourse) during the Trump presidency.

Maybe a better way to accomplish this is a free yearly physical with a doctor? The doctor can then be required to share any changes in disability with the government. Missing x years of appointments also stops your benefits. If you can't come to the doctor maybe they could do a house call?

> government employees work for elected officials

This is not correct and we have recent examples to counter this claim:

1. There are government employees directly employed by various branches of the government (ex: USDS was under the executive allowing them to be retasked by EO into DOGE)

2. There are government employees appointed into office who cannot fired after appointment (ex: Fed Reserve Chair)

3. There are also government employees who are non-political appointments

I think there are also more categories. I don't think your reply was charitable.

I don't think RevEng was saying it is impossible to do technically. I believe they were saying the charging infrastructure near their home makes it impossible to do practically as the chargers are often limited to 50-100kW. Aging Wheels did a video about this: https://www.youtube.com/watch?v=ouPiwt5hxXQ