Luca K. B. Masters (@lkbm) header
Luca K. B. Masters (@lkbm) profile picture

Hi, I'm Luca Masters and you're not. Neener, neener!

Marshall NC
Joined October 2007
Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)Photo by Luca K. B. Masters (@lkbm)
My daughter refuses to even try this weird, pink fluff that is obviously dyed cotton fibers, not food.
Some day, my daughter will taste cotton candy and think "oh, it's grape flavored candy!" And then she'll try "grape" flavored candy and think "wtf is this awful crap?"
24
I like telemetry because I'm mad at product managers who add bad features and want them to see that I've disabled them.
1
3
268
It's weird that the same culture that sings odes to the global supply chain thinks the take away of Petrov Day should be some weird celebration of individualistic behavior rather than a reflection of the need to build better systems that don't depend on people like Petrov.
4
3
235
"If you can't spend the time to write it, why should anyone read it?" is somehow treated as a sane statement, but it's definitely does *not* come from someone honestly considering why they read things.
2
47
"Asimov is a famously fast writer, therefore his content is far less worth reading than other writers" is something no one has ever claimed, because we all understand that the value you get from reading isn't "did the author take a long time writing it?"
1
32
"If you can't spend the time to write it, why should I take the time to read it?" is also a completely unoriginal statement that we've seen a thousand times. You didn't take the time to write that! It's less original, creative, or thoughtful than anything any LLM has ever said!
1
13
Well, if we're going to evaluate risks based on who talks about them rather their merits, then I guess we should look up who has spent a long time warning us about impact on cognitive fitness, cybersecurity, etc. and listen to them. Better for you if you think things through.
Why are the leaders of AI companies saying "Our products are going to extinguish humanity. Invest in us!" They come from an eccentric Berkeley subculture in which this eschatology is a sacred belief. Good analysis in a NYT op-ed by Cal Newport. nytimes.com/2026/09/05/opini…
1
57
Some day, my daughter will taste cotton candy and think "oh, it's grape flavored candy!" And then she'll try "grape" flavored candy and think "wtf is this awful crap?"
1
56
The Twitter discourse system trains us to fill in the context, and if you get mad at people for filling in the wrong one, you're mad at the wrong thing. Many tweets are incoherent (or can be read two opposing ways) unless you know whether the author is left or right..
Oh crap Wait "So you hate pancakes?"/ "No, did you read the words I actually wrote? Why are you guessing totally separate and distinct sentences rather than just reading what I wrote?" Is it possible the answer is "Because I was systematically taught to do that?"
1
2
66
And that's when the thing is left/right oriented. So many things aren't, so there's less of a clear way to fill in the correct interpretation.
1
1
19
Now, to be fair, this is not limited to Twitter discourse. I think Twitter makes it worse, but this is also something people constantly do irl, including when it's completely unnecessary, and I wish we did it less.
7
I find it weird when stores put a FB/etc. icons on their signs. I mean, sure it signifies "you can look us up on FB", but that's already assumed. Just bought a nutcracker a Google+ icon printed directly on it. (Along with FB, IG, and YT, and the company phone number.)
38
To experiment with improving Opus' writing style, I tried asking it to speak in @KelseyTuoc's voice on a random question I had, and it included "load bearing" (italicized, no less.) When called out, it blamed her.
1
2
38
It also blamed her for stating % confidence, but that's in my base instructions: "Prefer numeric statements of confidence". Of course, it's in there because of Rationalist/Rat-adjacent reasons, so it *could* be her fault.
14
omg, @Firefox Smart Window behavior: t⏎ => go to x.com te⏎ => Google search "te" tes⏎ => Ask "tes" test⏎ => Ask "test" Non-deterministic behavior in your UI is never the right move. This may be the dumbest UX design decision I've seen in any browser, ever.
1
75
Oh, look, a rate limit. Maybe if you didn't randomly route people's Google searches to the LLM you wouldn't have to rate limit people so hard!
1
16
They also forgot to test asterisks. Good thing no one ever uses the calculator functionality of Google. (Which will also default to Google for some arithmetic expressions and Ask for others.)
7
OpenAI's talk ending with "We accidentally created a nuclear weapon, which we're selling to the general public. As in industry, we must urgently build missile defense shields also using our technology." is very pikachu surprised face, equally for the first part as well the second
2
45
"Strike a deal" = "deal successfully agreed upon." Other contexts: "Strike" = "delete" It's a bad language, sir.
27
We went through all that dumb "American food is terrible" discourse where people kept citing the canned chicken no one has ever actually seen in real life. And no one ever mentioned that WE SELL "FAT-FREE HALF-AND-HALF" IN EVERY STORE.
3
161