Remote

David Nash

@dpnash@c.im

mastodon 4.7.2

Home: Part of an amazing #Polyfidelitous family (3 adults, 4 kids)

Work: #DataEngineering currently. #WebDevelopment and #DevOps in the past. #Chemistry a long time ago.

Developer of the HYG + AT-HYG star catalogs, and star chart software (https://codeberg.org/astronexus).

Expect lots of posts on #Cats, #Astronomy, #Computers, #Books, and other geeky topics here.

_Glacies delenda est._

418 Followers

389 Following

50 Posts

Joined October 04, 2023

Website:

https://www.astronexus.com

Codeberg:

https://codeberg.org/astronexus

David Nash @dpnash@c.im

· 2w ago

Replying to

@jackwilliambell@rustedneuron.com @Enema_Cowboy@dotnet.social @cstross@wandering.shop The liturgy requires there be a ceremony involving consuming something holy. ❌ Communion wafers ✅ Chocolate Frosted Sugar Bombs (There is a schism between the denominations that eat the holy meal plain, versus those who insist on adding a cup of sugar before consumption.)

4

0

0

0

David Nash @dpnash@c.im

· 1mo ago

Replying to

@sundogplanets@mastodon.social Specific example: I’d been doing amateur astronomy actively as a hobby for about a decade by then, and was quite comfortable finding my way around the sky without a reference, even in ultra dark skies with too many stars to count. One night at that particular star party, I found myself looking at a bright star that I couldn’t quite identify. It took me a minute or so to realize that I was looking at Polaris. The sky was *so* dark that the stars of the Little Dipper (Ursa Minor) were almost completely swamped by dozens of other stars nearby that are completely invisible from the city and even not readily visible from “ordinary” dark sites. You’d think the stars of the Little Dipper (though kind of dim by overall star brightness standards) would be bright enough to be clear under such a dark sky. From a kind of, sort of, dark sky, sure, that’s what you experience—but from there, they were literally swamped almost beyond recognition. The kind of sky where you almost feel you need to hold onto something because you feel like you’ll float off and drown in the stars if you’re not careful.

26

2

10

0

David Nash @dpnash@c.im

· 2w ago

Replying to

@0xabad1dea@infosec.exchange @lanodan@queer.hacktivis.me Actually 16.833… (16 5/6) colors. They omitted one hex digit from the 6 digits for the full color, so they started with purple but ended up with maroon. Use your imagination a bit with those poor non-purple graphics until they release the patch to add that digit back.

2

0

0

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@loren@flipping.rocks Have you bought a bunch of stars that seem close together in the sky, but many of them actually weren’t close to each other in space? You may be entitled to constellation.

68

0

22

0

David Nash @dpnash@c.im

· 2mo ago

Since #Caturday either is already here or is imminent, I have to announce a scientific result that I suspect is not original (especially given the number of catologists on Fedi), but still important to have in the scientific literature somewhere:

Cat toe beans can operate modern mobile phone touchscreens.

At least, based on a sample of one phone (a several-year-old iPhone), one cat (Loki, shown below), and one action (simple scrolling). Clearly, more research is needed. 🎓

8

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@davidgerard People who really should know better: "Yes! Let's use these things in professional astronomy and astrophysics!" The banal reality: Among many other topics LLMs fail dismally at, they don't know one of the most challenging practical problems of modern astronomy: how telescopes work. They also fail incredibly complex astronomical tasks like "Can I see this object from such-and-such location?" on the regular. https://c.im/@dpnash/116439620966675196

29

6

13

1

David Nash @dpnash@c.im

· 1mo ago

Replying to

@sundogplanets@mastodon.social Northwestern Nebraska (Sand Hills country) in the mid-90s for one of the first annual Nebraska Star Party outings. Skies so dark you had a hard time identifying the traditional constellations, and the Milky Way was so bright it cast shadows. I got some estimates of the brightness of the faintest stars I could see there, and the site handily beat some other outstandingly dark skies I’d been to before, even including a location at the edge of the Atacama Desert I went to for a total solar eclipse the year before.

4

1

0

0

David Nash @dpnash@c.im

· 5mo ago

This #Caturday, Loki and Hazel have an important message:

Loki: MRROW meow MRROW meow MRROW MRROW meow

Hazel: MRROW meow MRROW MRROW MRROW MRROW

That is all.

20

0

6

0

David Nash @dpnash@c.im

· 1mo ago

Replying to

@grumpydad@infosec.exchange @lennybacon@infosec.exchange @overholt@glammr.us It…Californicated?

2

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@stux@mstdn.social Badly dated, alas. Here's how it works today: Junior: Asks ChatGPT to help them draw a circle. It spits out a shape that looks vaguely like a heptagon with cat ears. Carefully pastes the exact output into their application. Middle: Asks Claude Code to create a circle-drawing program. The code is full of mathematical formulas and looks correct, except for drawing a square about 4% of the time and inviting Little Bobby Tables into the corporate intranet's database every other Thursday. The junior is tasked with fixing the squares and refilling the damaged database with proper circles. Senior (mid-thirties): Uses agentic AI to create an entire department of circle-drawing agents. Their code to replace all occurrences of the letter "E" in the company intranet with circles wins the agentic-AI leaderboard one month. They are eventually called in to explain why the corporate web site is all circles, no text, but are given a promotion because of how quickly their agentic process updated the web site. Senior (graybeard): Coffee-cup method, as before. They are fired because agentic AI is incapable of applying coffee cups to paper.

7

0

4

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@NuclearOatmeal@beige.party @hal_pomeranz@infosec.exchange Star charts. Bonus points for dragging in distance information and doing spiffy views of what things look like hundreds of light years from Earth. (This is a bit more than a “just starting out with a language” problem, but this particular one helped to get me started in software development in the first place, and also to bootstrap an early and somewhat unfocused web dev career to something more substantial.)

5

1

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to on universeodon.com

@Edmonds_Scanner@universeodon.com I keep trying to grow out my antlers, but the cats really like to use them as scratching posts, and no amount of redirection to “proper” scratching targets makes any difference. End result: I get half a centimeter of visible antler at most. Anyone else have that problem? I’d hate to have to rely on costume-store antlers to get a similar effect.

5

0

1

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@davidgerard Juicing the tarball is probably illegal in at least a dozen US states by now, thanks to the spread of reactionary cultural conservatism under the Trump administration.

11

0

3

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@catsalad@infosec.exchange But the only way to quickly verify that it’s “banana” bread is to put it in the spectrometer! It’s one of the best ways to identify individual elements. The most common impostor, “bana” bread, looks and tastes almost identical in all respects, but contains only half the sodium.

6

1

2

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@NuclearOatmeal@beige.party @hal_pomeranz@infosec.exchange Although I haven’t done these to bootstrap a brand new language so far, a couple of smallish projects I did just for fun that would be good for this task:

Sudoku solver. This was a problem in an online puzzle series I did a while back and it’s pretty straightforward to do a somewhat naive but useful solver and then, as you describe, make it better as you understand more techniques.
Classical cipher codebreaking (the kind you can break with just stats; a surprisingly large number of the old-school pre-Enigma or so ciphers, even ones that are much harder to crack than newspaper-style cryptograms, are fun to figure out how to break that way).

3

0

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@schrotthaufen@mastodon.social @zzt@mas.to Given how readily generative “AI” comes up with physically impossible chemical structures (“Five bonds on one carbon atom? Hah! That’s for amateur chemistry-mistake-makers like Intro Organic Chemistry students! I’ll give you eight-bond carbons all day!”) — I’m pretty sure some of them would be great UV radiation absorbers. If only those structures existed.

2

1

1

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@catsalad@infosec.exchange Cat, as requested.

3

0

1

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@astridpoot "How do you exit vim?" "You hit meow-colon-w-q. Or just meow-colon-q-exclamation point if you don't want to save changes." "Excuse me... 'meow'-colon?" "Yes. The 'meow' key. Doesn't your keyboard have a 'meow' key?"

6

2

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@rainer @cmsdengl@mas.to Modern English: "Don't use passive voice -- it sounds weak" Ancient Latin: "Oh to hell with that -- we use passive voice and make it sound active AF"

2

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@TimWardCam@c.im @Thorium@social.linux.pizza @petrillic@hachyderm.io There is a theoretical basis for this, in that many reactions won’t start unless the reactants have a certain minimal amount of energy, and (to oversimplify things a lot), the number of molecules in that high-energy regime grows (roughly) exponentially with temperature. For a certain value of that minimal activation energy, the increase does work out to about 2x per 10 C, which is the basis for the guideline you mention. So the situation here isn’t so much that things are a lot hotter with a small temperature increase, but rather that a small temperature increase can dramatically increase the amount of material that can react easily.

2

4

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@FluentInFinance@mastodon.social 2026: no more free snacks or drinks 2027: no more free toilets 2028: no more free air 2029: no more in-cabin seats. In fact, no more fuselage. Everyone gets a wing seat. Bring a coat in case it gets a little chilly.

5

0

0

0

David Nash @dpnash@c.im

· 2mo ago

New word in the team group chat today: “Claudious”. Means something shows signs of being written using Claude, as in “this commit looks suspiciously Claudious.” Possibly best said in a posh British accent so it almost rhymes with “odious”.

1

0

2

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@middleclasstool@phire.place You’re assuming it works like “index”. If it works like “rex” instead, it’s “Windeges”.

4

5

2

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@djnavarro@hachyderm.io This is *roughly* where things stand for me at work (I'm not under quite the same degree of pressure, and the company I work for makes billions of dollars of things where there is at least a comparable duty of care, so the pressure to use generative "AI" day to day is a bit muted. Not zero, though, and it seems to be growing over time.) I have tried a number of generative "AI" tools for coding and found that they often do reasonably well at generating correct syntax for uncomplicated problems, but they *very rapidly* go off the rails when confronted with something slightly out of the ordinary. For example, I gave Claude Sonnet a softball task of parsing a column in a spreadsheet for truthy/falsy values (e.g. "TRUE", "t", and "1" all counted as "true"). Softball indeed for both novice human and stochastic codebot. But once I said that the column could contain words in languages other than English, it hocked up a bizarre claim that I should check for words meaning "false" instead of "true" because most languages have more consistent words for that concept, which is one of the most techbro-brained bits of horseshit I've heard recently. Claude Sonnet then rewrote the code accordingly, without being asked to. I do not relish the prospect of having any significant amount of code with bizarre logic bombs like this building up in it. The only defense, realistically, is to read everything (absolutely *everything*) the bot produces, not just the code, and that negates a lot of the time savings. My current conclusion is that there *may* be some use cases for ultra-repetitive, stereotyped code that isn't especially edifying for humans to bang out over and over again, but that's about the extent of things. If time pressure gets ridiculous, I may cautiously go into that use case very selectively at work. For personal projects: fuck no, generative "AI" is never showing up in those, whether they be code, art, or writing of any sort.

1

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@catsalad@infosec.exchange “…but, wait,” you say, “aren’t bananas supposed to be rich in potassium?” You’re thinking of “baknaknas”. The “k”s are silent, like the one in “knave”, so it’s pronounced the same as “bananas”. Easy mistake to make; I saw it all the time in Chem 101.

2

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@12thRITS @catsalad >M*A*S*H?…nah. C*A*T*S

3

0

1

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@david_chisnall @davidgerard The context for my remarks about LLMs struggling to answer the basic astronomical question “can I see this thing from this place at this time” was asking the Big 3 (ChatGPT, Claude, Gemini) to tell me the best time to see 5 celestial objects from southern New Zealand. One was fake, and the other 4 were in the far northern sky and completely invisible from New Zealand, for much the same reason that the Southern Cross isn’t visible from the northern United States. The number of invisible or nonexistent objects claimed to be visible ranged from 2 (Claude) to all 5 (ChatGPT); ChatGPT also offered helpful advice such as pointing out that one of them was in the northern sky from New Zealand, so you should look *south* to see it.

3

1

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@cmsdengl@mas.to @rozeboosje@masto.ai “Saxons, they get in the bin?”

1

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@renardboy@mastodon.social @grickle@mstdn.social Judging from the cartoon, that cat is sitting in a way that they could use their tail as a capo if they wanted.

1

0

0

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@WPalant@infosec.exchange Just barely an hour ago, I got a phish simulation training email at work that *contained a QR code* with the usual urgent call to action to “do the thing”, in this case, scanning a booby-trapped QR code. Jeezus, Google.

2

1

1

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@philip@mastodon.mallegolhansen.com @mcSlibinas@mastodon.social @ryanc@infosec.exchange Present perfect: snaphaveshot.

2

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@catsalad@mstdn.mx A common beginner mistake, especially for monoglot English milks just learning the language, is to accidentally accuse some other milk of being a milky plaything. “Es toy milk.”

1

1

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@markarayner@mas.to Impressively large root vegetable in that salad. But it looks a little fuzzy. I’d want to check it carefully for mold before buying any of the salad.

1

1

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@NanoRaptor@bitbang.social Unbeknownst to me, my cat bumped one of the sliders while I wasn’t looking. They managed to slightly increase the strong nuclear force strength. I only noticed when the barbecue grill turned into a small supernova when I pushed the igniter. I’m afraid to fix things by pushing the slider the other way—I don’t remember the original position, and if I move it slightly too far the other way, I’ll accidentally turn off the sun.

1

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@CatsOfYore@mastodon.social

When a cat does a jump:

physics, math, adjust the spring tension, math, physics, shift the feet just so, physics, math, hunch the back juuuuuust a bit, math, physics, one last butt twitch to compress the spring….JUMP

When a dog does a jump:

oh boy oh boy that looks like fun I can run so fast oh boy yes this is the most fun ever…JUMP

1

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@petrillic (Boss adds Project J as “high priority”) (Project Q falls off the “high priority” list and falls to the floor and makes an audible ‘clunk’ sound) (somewhat puzzled, boss picks up Project Q, puts it in the “high priority” list again) (Project Z falls off. Clunk.) (irritated, boss re-adds Project Z) (Project Σ falls off. Clunk.) (the process repeats with Projects Δ, G, Ж, K, 不, and many others. Despite a deafening roar of ‘clunk’s, the penny never drops, although the boss’s true sense of priorities is amply revealed by their actions.)

1

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

Tired: Spirit Airlines Wired: Spirit AIlines

Benefits:

Sweet, sweet VC cash
Immediately able to use “travel agents” to mean something new and exciting to the technology press
One fewer “R” in name saves money on signage

1

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@futurebird Mirrored two-dimensionally, like a floor plan: no problem. Mirrored three-dimensionally, right down to the molecular level? Possibly a big problem. https://en.wikipedia.org/wiki/Technical_Error

en.wikipedia.org

1

0

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@algernon@come-from.mad-scientist.club Just for shits and giggles, I ran a project I wrote in Go through this “detection” tool. Yep, it gave me full “AI” credit for “machine-perfect formatting”. This is *Go*, FFS. It’s kind of legendary for not only having a characteristic default formatting, but also for very strongly encouraging Go programmers to incorporate said standard formatting into their toolchain as an automated process. Also got accused of being AI for the same sorts of things that you saw in its output for Rust, like sensible naming conventions, actually adhering to standards from a linter, and (my favorite) *proper exception handling*.

0

0

0

0

David Nash @dpnash@c.im

· 3mo ago

Replying to

@CynAq@beige.party @glyph@mastodon.social @matt@toot.cafe It has really felt to me that in the last year or so, the default style of the big LLMs has gotten more formulaic rather than less. At the very least, it seems to be a lot easier to ID blandly generic text as AI- written from just a sentence or two (the number of times I’ve seen multiple possible AI “tells” in the first sentence of a document is astonishing). And don’t get me started about coding-oriented AI output. I had to deal with a SaaS product AI’s severe delusions of grandeur at work recently (not by choice), and it sounded even more heavy-handed than what I remember people showing at the beginning of the “agentic” AI fad early last year. So…real phenomenon or just an overactive imagination (or unconscious cherry-picking, or…)? And if real, does it make sense to suppose that it’s a form of model collapse already?

0

1

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@SmartmanApps@dotnet.social Good *grief*, the splaining is strong here. I know what I am talking about here. In particular: "which is unambiguously “six divided by the result of multiplying x by two”" - as is the original expression you wrote No, “six divided by the *result* of multiplying x by two means six divided by *the quantity* 2x.” That’s not the same as “six, divided by two, then *multiplied * by x”. It’s “six, divided by two, then *divided* by x”. "not the same as the ordered sequence" - yes it is No, it is *not*. There is a notational difference here. The expression 6 — 2x *specifically* with a horizontal bar separating the top and bottom expressions, means “divide the result of evaluating the top expression, in its entirety, by the result of evaluating the bottom expression, in its entirety”. It groups the two smaller expressions’ operations *before* doing the division. It has a value of 3/x, which is not the same as the simple term-by-term sequence, which gives 3*x. Yes, really, and if your first response is “no it doesn’t”, then go pull out an elementary algebra text and figure out what 3x-y —— 2+y has to mean. It *always* means “take the quantity 3x-y”, then the quantity “2+y”, then divide them”. It doesn’t insert a division between the “-y” and the “2”, the way that a simple operator sequence like “3x - y ➗ 2 + y” does! This is, in fact, the most common way I’ve seen division represented beyond elementary arithmetic. It is all over the place in literally every mathematics publication I have seen from sometime in middle school through differential equations class in college years ago. And there are reasons for this, like I said, and one of these is precisely to keep a term like “2x” intact in a division, and not have to worry if the “2” becomes a divisor instead of a multiplier because someone forgot a pair of parentheses somewhere. I’m done here. It’s pretty clear you’ve seen lots of elementary arithmetic but don’t seem familiar with topics that are much more advanced than that.

0

1

0

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@sigsegv@hachyderm.io @wdormann@infosec.exchange Super-Cali-Fragilistic-ExPi-Alidocious-Load-PC-Letter, I think.

0

2

1

0

David Nash @dpnash@c.im

· 1mo ago

Replying to

@redsad@ohai.social It's not a prohibition against giant arches. This arch is totally fine. It's a prohibition against giant arches that have a single pointy end that has to be forced into the ground before the rest can settle into place correctly. That's an old design, and it's banned for good reason (the pointy bit is murder on underground infrastructure, and rather literal murder of wildlife).

0

0

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@SmartmanApps@dotnet.social This particular one’s unambiguous, but there’s at least one of these meme problems that has something annoying like “6➗2x” in it. This is only unambiguous if you insist that this must be read as an *ordered sequence* of operations: “six divided by two, then multiplied by x” And that it must never be interpreted as a *grouped set* of operations, like the very similar-looking 6 — 2x which is unambiguously “six divided by the result of multiplying x by two” (i.e. not the same as the ordered sequence), and also how most of us write division after about 8th grade. This is also why, past a certain point, *nobody writes expressions like these the first way*, outside of silly meme problems. In particular, you almost never want “2x” to accidentally become 2/x or x/2, but that’s what happens in 4th grade order-of-operations meme problems if you’re not careful.

0

1

0

0

David Nash @dpnash@c.im

· 2mo ago

Replying to

@SmartmanApps@dotnet.social Like a lot of people, remembered PEMDAS (or the regional equivalent) to get the answer from 4th grade math class, which (because M and D are actually equal in priority) is clearly the correct answer, period. Then remembered that just about the only times I’ve seen division written like this since …well, roughly 4th grade math class, is in internet meme problems like this, because nearly every division I’ve seen in mathematics since then looks like: 3x + y ——— 2 π which nicely sidesteps most of the notational ambiguities that these internet meme problems exploit. Once I hit middle-school algebra class, I stopped writing all but the simplest multiplication-containing, and *especially* division-containing, expressions like the meme problems *precisely because it is awkward and error-prone to do it that way*, and I am utterly unsurprised a lot of people “get those wrong” even when they understand elementary arithmetic perfectly well.

0

1

0

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@n3wjack@mastodon.social @sigsegv@hachyderm.io @wdormann@infosec.exchange Get-OuttaHereWithTheseRidiculouslyVerboseCommands

0

0

0

0

David Nash @dpnash@c.im

· 4mo ago

Replying to

@renardboy@mastodon.social @middleclasstool@phire.place The two words would individually pluralize like that (“rex” for sure; “tyrannosaurus” is 19th century Neo-Latin, but if the ancient Romans had that word as well, that’d be how it pluralized as well, barring a very strange circumstance). “Rex”->”rices” is funny, but, sadly…no.

0

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@Minimus "Ecce! Ad lunam parvam volat!" "Illa...non...est...luna..."

0

0

0

0

David Nash @dpnash@c.im

· 5mo ago

Replying to

@SouthDakota 8647 >> 1488, ethically as well as mathematically.

0

0

0

0