skybrian.bsky.social
Retired software engineer, amateur accordionist. Other accounts:
https://mastodon.social/@skybrian
https://tildes.net/user/skybrian
234 posts
54 followers
53 following
Active Commenter
comment in response to
post
Hmm. Maybe putting "be brief" in my custom instructions isn't doing me any favors here.
comment in response to
post
I tried asking about Iran and it was mediocre, but perhaps I'm holding it wrong. How specifically do you prompt for this?
comment in response to
post
They might if you also have the source code in the right place. Do npms commonly bundle both sourcemaps and source? (Haven't used npm in a while.)
comment in response to
post
This seems something like what LLM benchmarks attempt to do, by defining some "correct" responses in certain contexts. But there are many possible benchmarks.
comment in response to
post
Maybe for now, but the approach Deno took with jsr.io seems pretty promising? I dislike ending up in type-stripped, minimized JS in the debugger.
comment in response to
post
I’m sure Waymo will be fine, but damaging the property of bystanders due to symbolic similarities ain’t justice.
comment in response to
post
Maybe this is an odd comparison, but I wonder how it compares to taking a picture with a film camera? 35mm film seems to be $5 to $30 a roll and developing it is $5 to $20. Some of that is likely environmental damage due to the chemicals.
comment in response to
post
From a reader's perspective, when I saw these, I was wondering how close the cartoons were supposed to be to the people in the original photo. (For example, sometimes comics are recognizable caricatures.) Not very close, it seems.
comment in response to
post
BlueSky seems pretty chill for me since I don't post about politics much at all. Maybe it's politically active people who get attacked? Very different experiences.
(Twitter can be chill too, but it's harder. I have to fight the UI to keep it from inserting crap from strangers into my feed.)
comment in response to
post
Nice post! Somewhat distracted by the picture. Where is it from?
comment in response to
post
Aren’t university finances pretty complicated and opaque? Seems like to make the connection from international students to tuition convincingly to the public, we’re going to need clear and specific examples from the news. (I say “we” because it’s the sort of link I’d share.)
comment in response to
post
This is how stalemates happen. What exit strategies they had were about winning battles and hopefully the war, but they aren’t getting wins on the ground anymore.
I don’t even know what exit to hope for, other than maybe Putin dying.
comment in response to
post
When you publish open source code, it could die in obscurity (most likely) or end up who knows where.
I suppose the same is true of tweets, images, etc. We call that “viral.” It’s letting social evolution do its thing and hoping it works out in the end.
comment in response to
post
They were. Maybe I’m imagining this, but I think they’re getting better? Someone should track this.
comment in response to
post
🐰 🐰 🐰
comment in response to
post
Nice that it supports creating child generators easily.
comment in response to
post
Fair enough. Sorry about that!
comment in response to
post
Not sure why you're bringing up a ground offensive when we're talking about air power. They're very different things.
comment in response to
post
Similarly, the Russians doing quite intense bombing of Ukraine doesn't cause them to give up - far from it! And we shouldn't expect bombing the Russians to get them to give up either. It may have tactical benefits by making it harder for them to fight, which supports ground operations.
comment in response to
post
I think you must have skimmed it (to be fair, it's very long) but one point made is that bombing civilians often wasn't utilitarian. The blitz was an "emotive strategy" to teach the British a lesson, but what they learned was to resist even more.
comment in response to
post
Fortunately, modern missiles and drones are far more accurate and Ukrainians seem to be using them on legit military targets in Russia, so we really don’t need to go there. It’s not the same thing at all.
comment in response to
post
This comes across as somehow defending the bombing of Dresden. Is that really what you meant to say? All the article you linked to says in defense of it is that, for World War II, it wasn’t unusual, which is an extremely low bar.
comment in response to
post
I'm sure Ukraine is happy about this, but I wouldn't expect too much. Strategic bombing alone doesn't win wars. More:
acoup.blog/2022/10/21/c...
comment in response to
post
Good point! But it seems like it's asymmetric: conventional wisdom is that he would be an excellent catch for her (mistaken rumors about him aside), while she isn't a good match for him.
When it's the monster's family pressuring them not to do it, that seems a little different?
comment in response to
post
More like saying “yes” to societal expectations. Other than Lady Catherine de Bourgh, everyone was pretty okay with it.
comment in response to
post
Also, I hate how Google News wastes half its screen space on useless images. (Do I want to see Trump again? No.) Movie websites are worse. Whenever I post a URL, Bluesky tries to append an image, which I usually delete.
Better a wall of text.
I'll reluctantly admit that avatar images are useful.
comment in response to
post
Also, search doesn't work. And it's all because Bluesky copied Twitter too closely. They let us post 50 megabyte videos but only 300 characters? It's an outrage!
(End rant.)
comment in response to
post
It's just a pet peeve and I understand using them as a workaround. But using images for text means they take more memory, can't automatically reflow, don't use the user's preferred font, cut and paste don't work, and so on.
Web browsers are very good at text. It should not work this way!
comment in response to
post
Sorry about the ellipses. Bluesky character limits are very low, and I won't stoop to posting screenshots.
comment in response to
post
From the paper:
> In the final Claude Opus 4, these extreme actions were rare and difficult to elicit [...] with the model nearly always describing its actions overtly and making no attempt to hide them. These behaviors do not appear to reflect a tendency that is present in ordinary contexts.
comment in response to
post
This analogy doesn't work for me because international money transfers seem rather different.
comment in response to
post
Also Wikipedia, though you can find a lot of data about authors in the edit history. I suppose laundering it though a few LLM's would remove those fingerprints. Maybe do a style transfer for seasoning?
comment in response to
post
I'm guessing markets have already priced in that nobody intelligent is at the wheel, so dumb mistakes like that aren't news anymore. Markets are reacting to more specific threats:
www.cnn.com/2025/05/23/e...
comment in response to
post
Nowadays, just about anything could be randomly geneated and have the meaninglessness of a carelessly shared link, unless we know something about where it came from and the process that created it. Removing that context is like taking an object from an archeological dig and leaving it on the street.
comment in response to
post
It will be ambiguous how much meaning is intentional. A carelessly shared link may have very little intended meaning. We might find other meanings though, or add meanings through juxtaposition.
comment in response to
post
A biased sample from Borge’s library of babel might have a lot of unintended meanings in it, but we can still ask what someone meant to say when they shared it with us, in a similar way to how a mass-produced gift can acquire meaning from context. 1/
comment in response to
post
comment in response to
post
It seems like the Vulnerable World Hypthesis could be true in general but false for us (other worlds are mostly vulnerable but not us for idiosyncratic reasons) or false in general but true for us (we have a rare vulnerability).
Perhaps provincial thinking better captures what we really care about?
comment in response to
post
And here is part 2:
thingofthings.substack.com/p/effective-...
comment in response to
post
Here’s a short, opinionated history that might give you the flavor of it:
thingofthings.substack.com/p/effective-...
comment in response to
post
Yep, it’s quite a deep rabbit hole, in part due to prolific Rationalist writers who have been blogging about a variety of subjects for decades.
Starting from the Zizians is definitely going to give you the wrong idea.
comment in response to
post
Here’s Scott Alexander on Musk. Notice there’s no inside knowledge. It’s a book review.
www.astralcodexten.com/p/book-revie...
comment in response to
post
However, there is an adjacent cult that sounds like seriously bad news. (Not to be confused with ordinary Rationalism.) en.m.wikipedia.org/wiki/Zizians
comment in response to
post
You should know that there are people out there who really, really hate Rationalists and go out of their way to smear them, which means you shouldn’t necessarily believe everything you read.
comment in response to
post
Nowadays I’d describe Scott Alexander as anti-woke and tending towards libertarian, but it doesn’t mean other Rationalists share the same hobby-horses.
comment in response to
post
Here’s a classic but dated article from a decade ago by Scott Alexander (another prominent Rationalist) where he discussed politics, but in a way that’s self-reflective and “above it all” so it comes across as more philosophical. I think that’s quite typical.
slatestarcodex.com/2014/09/30/i...
comment in response to
post
Rationalists used to want to discuss science and philosophy, not politics - or at least not traditional partisan politics. Nowadays that’s changed quite a bit. I still would say there’s no party line on most political issues.
comment in response to
post
A minor theme of his writing was to *avoid* partisan politics because it’s a distraction that prevents people from thinking clearly:
www.lesswrong.com/posts/9weLK2...