FishFace

Do you not think rants belong here? What about linking to a rant on a blog?

I mean I am a programmer and while I don’t submit many changes written by AI, I review plenty written with its help, and plenty of those are actually decent. Plenty are shit or pointless, some execs’ expectations are utterly insane (“you’ll be submitting a hundred changes a week”). But plenty of others are sensible.

Tech companies, as a general rule, do not filter the internet of their employees, because those employees generally need to do a lot of stuff with the internet (or networking besides the internet) and filtering it would cause a lot of problems.

Production machines (where the data lives) can be much more restricted than work machines. Strong access controls mean that compromising a work machine doesn’t give you access to production data.

Do you think tech companies filter their employees’ internet?

They have tens of thousands of employees, a few of them are bound to download some porn at some point. And the amount downloaded is about 20 files per year on average.

@[email protected] remind you of anyone?

I had no problems so no idea about troubleshooting. Wired ones show up on plugging in, Bluetooth ones are detected fine. We even have an official Nintendo controller working.

Thanks so much, with this I will be a counterstrike pro in no time!

I do have a desktop so no real point. I use a dock to plug in to the TV and connect controllers though.

Can’t be bothered connecting a mouse and keyboard to my deck for that kind of game.

Turns out that “that kind of game” also includes more than I expected!

That’s what I do inside my head…

A hammer is a completely different tool, but different defaults in a single program are not.

Point is there’s no objective standard for “perfect”

Putin when he tried to restore the majesty, pride and legacy of the USSR

Child or troll?

My experience doesn’t match this in terms of resuming after an interruption. For me at least, most coding doesn’t require that high state of understanding everything in every detail, so an interruption is usually not that big a deal.

It’s helped if the interruption is on my own terms though.

All languages convey information at roughly similar rates. Those with less information per syllable tend to say more syllables per minute. It’s a fascinating area of linguistics!

All natural languages have ambiguities where the meaning of a word depends on context due to changes over time.

Something that you can’t trust can be good if it is possible to verify without significant penalties, as long as its accuracy is sufficiently high.

In my country, you would never just trust the weather forecast if your life depended on it not raining: if you book an open-air event more than a week in advance, the plan cannot rely on the weather being fair, because the long-range forecast is not that reliable. But this is OK if the cost of inaccuracy is that you take an umbrella with you, or change plans last-minute and stay in. It’s not OK you don’t have an umbrella, or staying in would cost you dearly.

In software development, if you ask a question like, “how do I fix this error message from the CI system”, and it comes back with some answer, you can just try it out. If it doesn’t work, oh well, you wasted a few minutes of your time and some minutes on the CI nodes. If it does, hurrah!

Given that, in practice the alternative is often spending hours digging through internal posts, messaging other people (disrupting their time) who don’t know the answer, only to end up with a hack workaround, this is actually well worth a go - at my place of work. In fact, let’s compare the AI process to the internal search one - I search for the error message and the top 5 results are all completely unrelated. This isn’t much different to the AI returning a hallucinated solution - the difference is that to check the hallucinated solution, I have to run the command it gives (or whatever), whereas to check the search results, I have to read the posts. There is a higher time cost to checking the AI solution - it probably only takes 30 seconds to click a link, load the page, and read enough of it to see it’s wrong. Whereas the hallucinated solution, as I said, will take a few minutes (of my time actually typing commands, watching it run, looking at results - not waiting for CI to complete which I can spend doing something else). So that is, roughly, the ratio for how much better the LLM needs to be than search (in terms of % good results).

Like I said, I wish that the state of our internal search and internal documentation were better, but it ain’t.

No, AI results can be quite good, especially if your internal documentation is poor and disorganised. Fundamentally you cannot trust it, but in software we have the luxury of being able to check solutions cheaply (usually).

Our internal search at work is dogshit, but the internal LLM can turn up things quicker. Do I wish they’d improve the internal search? Yes. Am I going to make that my problem by continuing to use a slower tool? No.

More likely it’s the limitations of the data sources they could find

It’s the plug one

I call boilerplate having to repeat types because there’s no type inference, but this is not captured here.

Kolmogorov complexity would be a good denominator, but you’d end up with a higher boilerplate score if your language just uses longer keywords