travisgriggs

I’m overwhelmed by LLM/Agent signal on HN lately. I actually liked this question. It’s trite and whimsical, but frankly nice break.

Can you elaborate why adding guards makes things faster?

Tax social media and other forms of attention products. Time users spend on these products is time taken away from earning revenue that can be taxed. Balance must be restored. If viewers’ available time/ability is harvested so that they have less taxable income, transfer the tax to where the wealth generation is being transferred to.

Same for AI.

In family and friend relationships, this all resonates completely.

Where I struggle and find my ego self defensively screaming “But…!” is in work relationships. Product managers, where their wrongness makes my downstream life more miserable. Basically any relationship where I have a (self perceived) need for the outcome to be a certain way to protect/enhance my well being. Asymmetric relationships.

Q: what kind of collection is this real world example illustrating?

A: Copying Garbage Collector (semi space). Chapter 4!

Great book. I was always fascinated by bakers treadmill. Always wanted a real world case where I could implement one with Fibonacci sized mills.

SpaGrox

Paying $60B for this is idiotic.

Isn't that kind of Elon's thing?

It’s ok. In the future, no one will do math. Mathematicians will be directors, with a team of math bots that they administer and direct. Instead of being managed, they will become the managers of mathematic autonomons. Universities need to get with the program.

/s

This is awesome. Bag “vibe coding”. Today I will start coding in what I’m going to call “Roy Kent mode”.

I find that junior engineers like to use it “up the authority” of their arguments when my experience clashes with their desires. OTOH, I am humbly aware that sometimes my experience is wrong and a curse to me. I DO need to be careful to not “hold on to old ways”. But I’m not convinced this is the right way to level the fields between wisdom/experience and innovation/freshtake.

Reading the BBC article about poisoning the real time AI well of info yesterday, I was super struck by this point

"We're moving towards this 'one true answer' world. Before, Google would give you 10 blue links and you would kind of do your own research. But AI just gives you one answer. It becomes so easy to just take things at face value. You need to be careful."

In a world of insecurities, and a world where we crave out-facting or out-proving our fellow discussioners, this “one right answer” is like synthetic drugs to the social experience. And we suspect “it’s not good for us” but it’s just so damned addictive.

What is unclear to me is how less skilled people gain useful experience, when using these amplifying tools. I’ve been at this for 35 years; I like to think that sometimes i get some pretty amazing results.

I work with two pretty green developers. The rate that they can make a mess is now phenomenal. And the sense of confidence the tools give them with early successes, means any experience I might have to offer means less now. Which is ok, I’m not going to be that “my experience has to be useful to you so I still fell relevant” old guy. But I do find myself curious how “lessons are learned” that lead to greater and greater tool exploitation in this brave new world.

People are never going to quit doing this. I'm surprised we still get "incensed" by being "watched without consent". There is

No. Way. It's. Going. To. Ever. Get. Stopped.

The only way to level the effects are to radically increase the surveillance so that everyone ends up in a Dark Forest "I know shit about you too" deterrence stand off. And/or flood the sensors with so much input/noise that meaningful signal is tough to suss out.

quality code

Probably where the mismatch is in this discussion. The measure of what is quality code is all over the place. For some, some form of "good enough" is quality. And for others, metrics like terseness, readability, vacuous amounts of comments, cleverness, various fuzzy measures of "idiomatic", etc, make "quality code" much more of a moving target.

Anyone have any stats on just what the headcount is at Anthropic and OpenAI these days?

Been using Kagi search for more than a year. Been happy. I use GPT/et al for the little things (e.g. unit conversions, rather than search for and then try to use an enshittified web page from 10 years ago). But for actual real tech leaning content, Kagi has been pretty good.

There will rise a PETA like organization that will protest this. Probably call themselves the Lorax and protest that we're practicing colonialism on plants now.

All satire aside... this is pretty cool. And so are groups that look out for the little guy.

It’s hard to forecast this. Support calls occur chaotically. So staffing to support them is difficult to do in a way that keeps a steady margin.

Rust is mostly ~20% bigger. Except comments. Where they basically doubled... what's with that?

In a world where juniors (or seniors in new territories) are incentivized to publish or perish, how will any of us gain proficiency any more? I can see the agent assisted journey accelerating some familiarity, but not proficiency.

I’ve used AI tools to do i18n translations to Spanish and Portuguese (somewhat ashamed to admit this). I’ve grown more familiar with the structure of these languages, and come to recognize some of the common vocabulary for our agtech domain. If anything, I feel more clueless about both languages now than I did before, when it comes to any sort of proficiency.

There are a lot of tools that do parts of this. I want someone to take them, put them all together and fit them up.

But just a few inches earlier, the author stated:

Everything tools always turn into crap.

This seems like a contradiction to me.

I'm rather happy with Zed.

I use it for Elixir and ansible stuff. I may eventually be open to using it instead of PyCharm for python and/or Nova for C.

If there's one area I still feel that Zed lets me down is in pane management. Maybe I need to just learn more key shortcuts. But I spend a bit of time "managing" the secondary panes and having to switch back and forth between outline, files, search. I'm not sure what the solution is. Just wish the secondary panes weren't a scarce resource that had to be mux'ed betwixt.

I really like(d) the agent integration, but we're currently experimenting with Claude Code Desktop, and I really miss not having the tight integration. My guess is that I'm going to switch back to using the Pro subsidized version. I was getting by with ~$40-$50 a month. Now the company is paying $125 for Claude Team premium seat, and it's a lesser experience.

The bit in the middle where it decides to make its control loop be pure P(roportional), presumably dropping the I and D parts, is interesting to me. Seems like a poor choice.

I try to fly about once a week, I’ve never really tried to self analyze what my inputs are for what I do. My hunch is that there’s quite a bit of I(ntegral) damping I do to avoid over correcting, but also quite a bit of D(erivative) adjustments I do, especially on approach, in order to “skate to the puck”. Density going to have to take it up with some flight buddies. OR maybe those with drone software control loop experience can weigh in?

All of this article, both the good (critique of the status quo ante) and the bad (entirely too believing of LLM boosterism) are missing (or not stressing enough) the most important point, which is that the actual programming is not the hard part. Figuring out what exactly needs programmed is the hard part.

HARD AGREE. But…

Taken as just such, one might conclude that we should spend less time writing software and more time in design or planning or requirement gathering or spec generating.

What I’ve learned is that the painful process of discovery usually requires a large contribution of doing.

A wise early mentor in my career told me “it usually takes around three times to get it right”. I’ve always taken that as “get failing” and “be willing to burn the disk packs” [https://wiki.c2.com/?BurnTheDiskpacks]

Fortran, Basic, APL, Beta, Odin, Self, C, C++, Objective-C, C#, C--, D, Scheme, Clojure, F-Script, Eiffel, COBOL, Ocaml, Haskell, Snobol, Crystal, Forth, Python, Lisp, Brainfuck, Java, Oak, Javascript, TypeScript, Wasm, Logo, Elang, Elixir, Gleam, Elm, Zig, m4, Tcl, Simula, Smalltalk

Fun challenge. Unlike the author, I have nothing really to add.

I just wanted to say that "I did NOT write it with ..."

Our small team has a lot of commit messages like this. For a while, we had a guy on the team who had come from a site that expected more. The pet peeve he brought along was that commit messages end with a period (my guess is that someone at their previous work place had reasoned that forcing periods encouraged developers to actually write meaningful sentences). When I look at that period of development, I see lots of messages like “stuff changed.” And “more stuff changed.” And then it goes back to just “stuff changed” around the time they moved on.

Mac OS has become what would happen if Harley Davidson merged with Volvo Truck and some high up said that to "reduce costs" and "homogenize the brand", the design groups needed to be merged and put forward a unified design. If I was less lazy, I'd have a !AI thing whip me up a mashup drawing.

When you eat a strawberry do you feel like an impostor for not growing it yourself?

I don’t think this is the right question. What you posit is a consumption dilemma. It’s a valid question, but it focuses on what values we might arbitrarily ascribe to how we source what we consume.

The OPs dilemma is more akin to giving a cutting board for Christmas that you bought vs handmade. Or some other. I think these cases of how we present what it appears we created is the dilemma OP is facing.

Huh. This seems selectively simplified. At least according to https://en.wikipedia.org/wiki/1981_Professional_Air_Traffic_....

Multiple economic write ups have concluded that Reagan’s “stick it to the upstart guy” cost us tax payers way more than it would if they’d just acceded and maybe even thrown in a gracious bonus to say thanks.

Larger sociology say the intangible cost to labor balance laws actually were much more.

Reagan’s trickle down (great euphemism for “piss on”) movement was the beginning of the demise of the GOP IMO. Disclaimed: I voted both times for him and many GOP followers.

It’s also the only industry that is legally allowed to practice ageism. You have to start before or up to 31 years of age. You’re out at age 56. This figures into how the benefits are structured.

You can still do contract ATC work after 56.