michaelcampbell

> Every card remembers every rating you give it, as well as the time / date. This allows for Anki to solve for a 'forgetting curve', and predict when different cards have a chance to be forgotten.

True to a point; every card has its ratings, but the "forgetting curve" algo of FSRS is only tuned to the deck (or "option set") that the card is in, not per card.

I would love to hear/see a definitive answer for this, but I read somewhere that the relationship between MS and \A is such that the copilot version of the \A models has a smaller context window than through CC.

This would explain the "secret sauce", if it's true. But perhaps it's not and a lot is LLM nondeterminism mixing with human confirmation bias.

Can you (or anyone) explain how this might be? The "agent" is just a passthrough for the model, no? How is one CLI/TUI tool better than any other, given the same model that it's passing your user input to?

I am familiar with copilot cli (using models from different providers), OpenCode doing the same, and Claude with just the \A models, but if I ask all 3 the same thing using the same \A model, I SHOULD be getting roughly the same output, modulo LLM nondeterminism, right?

I have a few shell scripts using `getopts` that have a `-!` flag to make it go; the default is dry-run.

And it's pretty nice. The downside is if you get used to that behavior in things that don't have it the consequences can be bad. (Like the common `alias rm='rm -i'`. No, that's just a trap; don't do it.)

It feels like our (US) political system; people in their camps refuse to believe any data proposing a benefit of the "other" camp.

For me, the rise of the TUI agents, emerging ways of working (mostly SDD, and how to manage context well), and the most recent releases of models have pushed me past a threshold, where I now see value in it.

> all fintech being built was being built on top of crypto and NFTs

This seems WAY less true than the assertion of software being built with LLM help. (Source: Was in FinTech.)

Like, to the point of being a willful false equivalence.

> Everyone is building tools for tool builders to make it easier to build even more tools.

A lot of hobby level 3d printing is like this. A good bit of the popular prints are... things to enhance your printer.

Oddly, woodworking has its fair share too - a lot of jigs and things to make more jigs or woodworking tools.

One thing I learned in 30+ years as a software developer is a static blog site doesn't need to "time out".

It's better than most, but not 100%. As I see this the clock hands are all correct, but the numbers only go 1-8.

> Remember, there used to be a time programmers productivity was measured in LoC per hour.

Do you remember such a time or company? I have been developing professionally since the early 1990's (and hobbyist before then), and this "truth" has been a meme even back then.

I'm sure it happened, but I'm not sure it was ever as widespread as this legend would make it sound.

But, there were decades of programmers programming before I started, so maybe it just predated even me.

Yegge's been around a long, long time and this is about within a standard deviation of his normal writings, at least in style. I haven't read much of his LLM/AI related stuff, but none of Gas Town left me with any sort of "huh" reaction, knowing the author.

> > Indian students

> Is this cultural?

Could be, but there are a number of very popular Youtube and other video based classes/bootcamps (taught and targeted from/to Indian students) that teach how to work with git and github that show how to create PR's and comments in repos, and then a lot of students do that, on public (and popular) repos.

There are a couple very famous examples of this.

Like "meteoric rise". Meteors come down, not up.

And yet, we all know what it means. Language is fun.

Then you've totally missed the entire point.

All of us did in the day, and most of the strips were common from multiple people for a given office, for the obvious reason.

This seemed so blatantly obvious (and the answer if asked "why did you put this one up?") that I'm agog at anyone thinking otherwise.

> That API design pattern would be a nightmare!

And yet people have been using it for decades without much trouble. I can't remember ever seeing any complaints about API explosion from adding methods to numbers in ruby, and rails does a fair bit of it.

Possibly one reason is with a fluent API that's pretty understandable, you don't need to rely on autocomplete. (And indeed, all this happened before that was a big thing in tooling.)

> had these features 2-5 years before Android did.

"first" isn't always more important than "best". Apple has historically been ok with not being first, as long as it was either best or very obviously "much better". It always, well, USED TO focus on best. It has lost its way in that lately.

Thank goodness for reader mode. The transparent background where the text is with the wiggly line background is... challenging.

FF working as expected for me on Ubuntu

Lots of good stuff here. Literally none of it required working at Google to learn.

> Don't they miss the feeling of..... programming? Am I the weird one here?

Our company is "encouraging" use of LLMs through various carrots and sticks; mostly sticks. They put out a survey recently asking us how we used it, how it's helped, etc. I'll probably get fired for this (I'm already on the short list for RIFs due to being remote in a pathological RTO environment and being easily the eldest developer here, but...), but I wrote something like:

"Most of us coders, especially older ones, are coders because we like coding. The amount of time and money being put spent to make coders NOT CODE is incredible."

Did he actually ever write anything substantial, other than that book?

We got my son vaxxed for this when he was able, and the doctor doing it was quite reluctant to do it. (US; ~2001)

Have we though? I'm glad we're not shouting about it from the rooftops like it's some magical "win" button as much, but TBH the things I use routinely that HAVE been rewritten in rust are generally much better. That could also just be because they're newer and have the errors of the past to not repeat.

It is; last couple months.

FSRS has dials to tweak to show you cards either more or less often, so if it's feeling "too hard" you can tweak that dial to show you cards when "easier" (at the risk of seeing more cards per day since you're seeing them more often).

Failing to do that, one might consider that instead of focusing on how hard each card feels, but rather the size of the corpus that they have "under their belt". This is the case if you're constantly adding new cards - if you are and your cards/day is stable, then you have an ever increasing mound of memorized knowledge.

If you aren't adding new cards, then the cards/day will inevitably go down, barring some actual cognitive issue.

It's a matter of what you focus on as your measure of success.

Thanks for posting that; as soon as I saw the title here I was going to look that up if no one else had already. Sabine Hossenfelder too, though there's far too much content from her on this to put a list, but anyone interested might like some of her takes.

A few debates between Brian and other notables; Hossenfelder, Eric Weinstein, and Roger Penrose to name a few; have popped up in my youtube feed lately which are typically also engaging.

Same here; I've gotten a lot of utility from Udemy. Actually kind of got me a job (tbf, the manager I'd known for years and he'd've hired me regardless, but I was able to actually DO the job he wanted me to do basically day 0, even when he was willing to hire me and let me learn as I went).

There's some crap there; I've returned a couple courses not to my liking, but by and large I've been happy with them.

For anyone following this thread, that link goes to a download itself rather than a page where you choose whether or not to download.

I was in Switzerland on holiday and noticed some of these. Pretty fun, non-destructive and interesting.