HN user

arh68

1,213 karma

alec at mailc dot net

Posts7
Comments676
View on HN
Julia 6 months ago

FYI Don't miss out on the <!-- HTML comments --> .

Wow. AMA, you say? When do you prefer loops, or lines? Additional lines, or carriages? How much do you tear down at once? if you don't mind me asking

I'm like median on Metro, ~60 hours over years (though perhaps just the one hour, 60x, &c). Never too late to learn some strategy, I guess. Never played Motorways.

They ought to get source code, too !

Crazy to think we'd pay for software, ask for source to run a Fortify scan & whatnot, and get told to kick rocks. "Proprietary ... trade secrets ... &c&c". Just mark it green

Well, now I'm confused.

Fira Code uses uniform length for +, =, and -. - and _ share similar length. The /\ characters join together and render smaller compared to the other fonts.

This "joining" is a ligatures thing, I'm almost certain, at least for `<>`. I can't for the life of me get anything on macOS to render `/\` as joined, though. Stumped. I've no preference either way, it's just weird to see a familiar font rendered so strangely. Maybe it's a Windows font rendering thing ?

A very fair comparison, though I'd argue legibility isn't always worthwhile; the MICR (?) fonts on checks are quite legible (perhaps machine-legible) but too weird to use.

also, TIL IntelliJ bundles Fira Code for quite some time now

one clear, sharp picture. Better than cable. Here's the reason. OTA broadcasts must meet a legally defined broadcast quality: they're all at the top quality that HD can provide

I wonder about this part. I think it's probably still true for the "main" station, like full 1080i for 9-1, but the "extra" stations like -2, -3 ... -6 are usually noticeably compressed.

From my limited understanding, all the extras are sharing the same bandwidth, and more channels = more ads, so it's more like 1080i + 5x 480i. Some channels will look 1080i on both -1 and -2, then maybe 720i on -3.

I don't live near an ATSC 3.0 station, but it would be cool to get 4K/2160p. Soon ... (naturally, I'm curious)

[1] https://en.wikipedia.org/wiki/ATSC_3.0

[2] https://en.wikipedia.org/wiki/List_of_ATSC_3.0_television_st...

Maybe he "just" got it wrong. Maybe they're typos, and the manuscript was correct. Or...

Maybe Pemulis gave Hal an obviously wrong derivative, and when uncorrected, drove Pemulis to abruptly end the tutoring. Maybe Pemulis said it right but Hal heard it wrong. Or...

Maybe it's "just" another sign they're in an alternate universe where even the math is different. That's pretty much how I feel about it

Exactly. By GP's logic, time zones are absolutist, and every town could & should observe solar noon independently.

I think folks are averse to "world time" (for reasons, largely inertial), so maybe the baby step is try 1 timezone per country (like China's done for .. 75 years ?).

I'd even argue Local Time has only ~4 useful times: dawn, daytime, dusk, & nighttime. Where I grew up, the parks closed at dusk every day. Nobody complained

If I take a walk at 7:05:35 PM every day, it seems very precise but doesn't indicate whether I need sunglasses or a flashlight. It's meaningless precision, like 0.6235 slices of pizza. If I'm coordinating a walk with you, I might as well use UTC: it still won't tell light from dark, but at least nobody'll be waiting for an hour due to DST. It'd make more sense to schedule our walk at `1 hour before dusk`, or "just" settle for UTC, IMO.

Wow, a 5060Ti. 16gb + I'm guessing >=32gb ram. And here I am spinning Ye Olde RX 570 4gb + 32gb.

I'd like to know how many tokens you can get out of the larger models especially (using Ollama + Open WebUI on Docker Desktop, or LM Studio whatever). I'm probably not upgrading GPU this year, but I'd appreciate an anecdotal benchmark.

  - gemma3:12b
  - phi4:latest (14b)
  - qwen2.5:14b [I get ~3 t/s on all these small models, acceptably slow]

  - qwen2.5:32b [this is about my machine's limit; verrry slow, ~1 t/s]
  - qwen2.5:72b [beyond my machine's limit, but maybe not yours]

What do I picture? 2D pieces, to be honest. Computer chess is just so prevalent. I picture what I'd call "USCF style" [1] because that's what they'd use in the Chess Life magazine to annotate games. I also picture the "old style" pieces [2], used in other periodicals & some books (especially puzzles).

I bet a lot of people picture the default set on Chess dot com. I find it very hard to adjust to new sets, for whatever reason.

As far as real pieces, I picture plastic pieces & vinyl board. Either what I'd call the "triple-weighted set" [3] (my favorite), the plastic "Dreuke set" [4], or the "basic USCF set" [5].

I had no idea there were so many variations, especially in the last 100 years. Most I've never seen before. I still dislike the real abstract/bauhaus style but there's a lot of artistry in the sets.

[1] on lichess, it's called "companion"

[2] on lichess, it's called "leipzig"

[3] https://www.chessset.com/collections/weighted-chess-pieces-h...

[4] Player's Choice, like https://www.wholesalechess.com/reproduction-of-the-drueke-pl...

[5] https://www.uscfsales.com/single-weighted-regulation-plastic...

Great article. Informative, qualified. I had not realized how splintered the [E]BNF syntax is, like in the way I already knew timestamps are (3339 vs 8601 vs mm/dd/yyyy &c &c).

Q: what's your ideal way to write Unicode characters clearly? In the W3C/XML spec they'll have stuff like [#x200C-#x200D] but I have no idea what those are, without like a dictionary on hand. Points for specificity, but it doesn't scream "readable".

Your point about standards-not-publicly-available is unfortunately similar to, well, laws. In some areas, "the laws" themselves are not public (!) though perhaps it's a digression better to not get into

pedantically, s/unabiguously/unambiguously/g

I feel like the prices have gone up for years & years. I thought my Spyderco was a cheaper alternative to Benchmade, but now they're basically all $100+. $65 is now $130 or so.

My guess is collectors that'll buy at almost any price. Some knives from the '80s that used to cost $25 are simply eye-watering today (well past 10x).

Underperforming states escape scrutiny. Our biases prevent us from asking, for instance, what’s going on in Oregon. Or Vermont. Or Maryland.

Uh, is this true ?

I feel like there's been plenty of asking as to why Oregon schools are as bad as they are. Going back 10 years. Hell, maybe 30+ years. It's not a secret. I've never felt we-can't-talk-about-it, as the author claims.

I'm glad Mississippi is making progress, but a lot of the article reads like a straw man (or, weirdly, a reading comprehension test)

LLMs specifically, have been shown to have the capacity to make such copies

Exactly. I asked my Gemma how long of a quote it could give me of a given book, if I were the author & gave express permission, and I was a bit surprised it readily admitted it could

Without Permission (Current Limit): Single sentence.

With Broad Permission (Full Reproduction Allowed): I could theoretically quote the entire book.

Eye-opening (for me, at least).

Meanwhile, a ~year since Riot required Vanguard, League/Valorant playerbase isn't really any smaller. If anything, it's continued to grow.

The players do not seem to mind. At all. Basically a non-issue for 99%.

I'm not terribly surprised so many here find kernel anti-cheat "unacceptable" or "ridiculous" or whatever, but there's just no there there. Microsoft has kernel access: are they selling my stuff, too ? I don't pay for Firefox: are they auctioning off my passwords? I guess it's fun to speculate; I can make unfalsifiable claims all day

Yeah. I wrote so many hamburger essays that 1) were terrible & 2) got As, I wouldn't blame a student for generating that crap. I have written approximately 0 hamburger essays post-high school. They don't want good, they want their made-up rules to be followed

you're not always going to have a calculator in your pocket was the old fib, now it's s/calculator/llama/

So much of education when I was growing up was pointless box-checking, I'm a little satisfied that, finally, LLMs might swing the pendulum back towards valuable work. If any assignment an LLM could handle is limited to 10%, and 90% of the final grade is determined by oral exams, that seems positive.

Is "2x4" equal to 4 + 4, or 2 + 2 + 2 + 2 ? There's only one correct answer! "You ... pick up that can" wow such learning

If LLMs love to produce 3-bullet-point-bold-font copypasta, it's probably because it's the exact varietal of crap that garnered so many upvotes on Quora. Why would I be asked to write a 500-word essay when 50 words would suffice? Maybe let's move beyond regurgitation & rote drudgery

I believe they're talking about something like ECC

To use ECC, all parties must agree on all the elements defining the elliptic curve, that is, the domain parameters of the scheme. The size of the field used is typically either prime (and denoted as p) or is a power of two

like "25519"

An EdDSA signature scheme is a choice: ... of finite field F q over odd prime power q ... Ed25519 is the EdDSA signature scheme where q = 2^255 - 19

[1] https://en.wikipedia.org/wiki/Elliptic-curve_cryptography#Do...

[2] https://en.wikipedia.org/wiki/EdDSA#Ed25519

FWIW, my llama suggests that the original usage of the term `political correctness` was somewhat inverted:

The term "political correctness" was first used in a political sense by Maoist factions within the American New Left movement during the 1970s. It was employed to criticize liberal critics who were perceived as compromising revolutionary principles for the sake of mainstream acceptance.

So the original sense was a too-centrist/too-mild/too-pragmatic sort of INcorrectness. I found that interesting.

Is wokeness / anti-wokeness the new heresy ? Cool beans. I'm not really interested either way.

Atomic means nondecomposable [1]. I dare say array-typed values are not atomic.

Re: "(instructor_id, List<skill_id>); PK(instructor_id)”, are your Lists sorted ? Are your Lists actually Sets?: what prevents `insert values (100, [2, 2, 1])` ?

If you think those are isomorphic, how about “(json_data); PK(json_data)” ? With, you know, all the same data in the JSON. Tomato, tomato ?

I’ll just link this [2] PDF that seems to describe the present

Does this mean that SQL:1999 allows databases that do not satisfy first normal form? Indeed, it does, in the sense that it allows “repeating groups”, which first normal form prohibits. (However, some have argued that SQL:1999’s ARRAY type merely allows storage of information that can be decomposed, much as the SUBSTRING function can decompose character strings—and therefore doesn’t truly violate the spirit of first normal form.)

[1] at least, according to Codd

[2] https://www.cl.cam.ac.uk/teaching/0304/Databases/sql1999.pdf

I think `COBOL` is (nowadays) a catch-all term for `COBOL` &/ `IBM assembler` &/ `IMS/hierarchical databases` and/or batch jobs. I can't fault a journalist for referring to all of that inexactly by `COBOL`.

Second, maintainability is absolutely not given freely. A gov't merely "purchasing software" does not necessarily get them the source code, let alone proper documentation. Seriously. Contractors have no incentive to make things easy for you, and all the incentive to make it hard. The tail often wags the dog.

"The powerpoint slides didn't say anything about lock-in" &c &c &c

Nah, enshittification is a way better word. Platform decay sounds too natural, like a tree falling in a forest.

It's like when the platform itself tries its hand at shitposting. When the software becomes shitware. Not so much a passive voice, but an active somebody-threw-it-down-the-shitter kind of vibe.

I remember white/black-dithering (at even twice the dpi) being a lot smaller than the greyscale jpeg, when I was scanning documents. Greyscale was a bit smaller than full color, but dithered was a big step up (er, down).