Lightnews — Scholar-powered news

Mike Samuel 🟣 @mvsamuel.bsky.social · 9h

Incoherent.

"I was HUD secretary. I know the answer [to housing prices]" he says, continuing to not articulate that answer, instead pivoting to point out that the NY mayor has no sway over US/Israel policy, undercutting his message that NY Jewry should worry about Mamdani's thoughts on Israel.

Mike Samuel 🟣 @mvsamuel.bsky.social · 18h

Thanks for taking the time to call this stuff out despite the abuse.
Lack of clarity on election process plays into the hands of gerrymanderers and into electorate apathy.

Mike Samuel 🟣 @mvsamuel.bsky.social · 19h

Just accept that your partner will probably never internalize a thing you say about the joys of methodicality or continuous improvement.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 1d

I've tried to visualize different PLs' approaches to constructor chaining.
Afaict, every failed theory in biology corresponds to an actual chaining semantics.
Phylogeny recapitulates ontogeny: C++ has it.
Biological essentialism but missing bits: Java.
Aquatic ape theory, I'm sure it's out there.

2

Mike Samuel 🟣 @mvsamuel.bsky.social · 3d

An aunt of mine spent 40yrs translating medical journals btwn French and English. There's a lot of nuance in technical language work but the hard part of her job was reorganizing so readers find things where they expect. I worry about this rush to obsolete the only people competent to judge quality.

2

Mike Samuel 🟣 @mvsamuel.bsky.social · 4d

The entropy here is just log2(|$chars|) x ($length), unless I'm missing something?

If the first is 6 some (case-sensitive alnums) and the latter is 64, you've got pretty good entropy.

Recommendations for strong passwords tend to fall in the 128..256b range: security.stackexchange.com/a/257535/227

How many bits of entropy should a password have to be reasonably future proof (10+ years)?

I’ve seen estimations that the NSA is capable of at least 1 trillion (PGP pass phrase) guesses per second, which would mean a password with 80 bits of entropy would take, on average, over 15,000 ye...

security.stackexchange.com

Mike Samuel 🟣 @mvsamuel.bsky.social · 4d

Columbus day is big; there's a big Italian-American population in NY state.

Pride parade has been skipped by most politicians until recently. SF's Folsom St. Fair is still controversial.

The Proud Boys marched in DC in January, the KKK back in 1999. What differentiates a march from a parade, idk.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 4d

There's also some advantage to getting all the random bits you need at once instead of calling for each output character.

Just get an array of random bytes and then do whatever conversion to safe ASCII you need once.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 4d

Yes, the static methods in there delegate to a strong source of randomness, or a well seeded strong source of pseudo-randomness.

1 1

Mike Samuel 🟣 @mvsamuel.bsky.social · 4d

Do R users really kvetch about its 0-basedness enough that this is worth calling out?

I suppose Fortran users get to feel smug either way.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 5d

Colour photocopiers will not photocopy currency. There are a few providers of the hardware and they all agreed to install recognizers for currency and reduce copy quality when it's detected on input.

Is there no similar hardware-provider bottleneck for non-state actors?

2

Mike Samuel 🟣 @mvsamuel.bsky.social · 5d

Yudkowsky is not unaware of how economic leverage trumps decision making quality. At 2:25:00 in the below he seems to have a nuanced view of leverage.

www.youtube.com/watch?v=s-Ek...

YouTube video by Foresight Institute

www.youtube.com

Mike Samuel 🟣 @mvsamuel.bsky.social · 5d

Do they assert LLMs are different from other ML approaches? I.e is a model trained from a large natlang corpus qualitatively better at reasoning than one trained on a medium-sized tagged corpus?

They seem to brush off AI&AI ethics work from <late '90s and I wonder if it's irrelevant from their POV.

Mike Samuel 🟣 @mvsamuel.bsky.social · 5d

If your dynamically imported module has a static import back to the module that imported it, do you run into trouble?

Or are import cycles errors?

1 1

Mike Samuel 🟣 @mvsamuel.bsky.social · 5d

Can you not do this with some kind of explicit meta-programming?

Or is the problem that dynamic import needs resources unavailable at compile time, like source files deployed alongside the program?

1 1

Reposted by Mike Samuel 🟣

Robyn Pennacchia @robynelyse.bsky.social · 6d

Taking in refugees isn't just a nice thing we do — we have a legal obligation. The 1951 Refugee Convention came about specifically because countries like the US refused entry to Jews who were trying to escape Germany, because the world didn't ever want that to happen again.

3 86 310

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Brazilian Orkut was wild.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

But my impression is that sub-communities of marginalized people might be geographically distributed but still form a strong clique on a graph cluster because they reach out and support one another.

1 3

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Yeah. You'd probably want some decay; people move.

But if you're trying to use ML to disambiguate, the more signals the better.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Do you have access to enough post time metadata to infer a likely sleeping time zone for the writer?

Probably would have lower accuracy in South America. Not too many timezones between the La Pazes and the San Juans.

1 1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Bombay, NY
Melbourne, FL

I imagine the first is complicated since the official name, according to the Maharashtra state authority, for Bombay is Mumbai.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Hiring and promo decisions are so arbitrary but AI adds a new layer. Thinkers might not be as proactive at BSing about all the ways they've used AI to increase their productivity as their less thoughtful colleagues. Luckily, you can outsource that to AI.

1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Aren't there a ton of homonyms even ignoring all the Alexandriae. I think there's a Washington, ON and one in Cuba. Also, a Washington island?

1 1

Mike Samuel 🟣 @mvsamuel.bsky.social · 6d

Non-fictional Jewish leaders in Oklahoma City understand perfectly well regardless of their personal politics.

1

Reposted by Mike Samuel 🟣

Tomas Petricek @tomasp.net · 14d

I'm teaching 𝗪𝗿𝗶𝘁𝗲 𝘆𝗼𝘂𝗿 𝗼𝘄𝗻 𝘁𝗶𝗻𝘆 𝗽𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺(𝘀)! again. I'll be posting the videos & tasks on YouTube too.

In the first lecture, I explain what's a tiny system, why write one and show plenty of demos!

🎞️ Playlist: www.youtube.com/playlist?lis...
👉 More info: d3s.mff.cuni.cz/teaching/npr...

Write your own tiny programming system(s)! - YouTube

The goal of this course is to teach how fundamental programming language techniques, algorithms and systems work by writing their miniature versions. The cou...

www.youtube.com

1 17 50