prionassembly

I have Discord and Telegram groups for my blogs. I think most people that reach it bounce after seeing how loooong and hard to parse (at first glance) my posts are, but there's three or five people who AFAIK are reading me.

Ima just go ahead and spam it: https://asemic-horizon.com -- I don't think it's for the median HN dude/tte, but this forum is a meeting of diverse minds...

Yea, but this complexity is not inherent to code.

Expressivity and automatic enforcement are completely orthogonal. Some pseudocode in academic papers is sublimely expressive, but can't be typed directly on a computer.

Haskell is very powerful, but I don't think this article presents anything that's widely available in popular languages right now. I'd be very skeptical that TypeScript isn't expressive enough to do Maybe and Either; I wouldn't be surprised in PHP is there already.

Tuples (or namedtuples/dataclasses if you must have named fields, which in Haskell too is additional syntax) are product types, enums are sum types.

You should look into functools.singledispatch if you're doing business logic with isinstance. But you should also learn about object-oriented programming, inheritance vs composition, mixins, etc. People get sour on OOP because they should have learned FP first and then imitate/translate FP idioms -- instead of doing something like modeling domain ontologies in object graphs. Likewise, people get sour on FP yadda yadda instead of trying to literally write their programs as proofs to theorems.

I used to go to Brasilia 3-4 times a month for work. If you got on a Tuesday morning flight you could rub shoulders with congressmen, top-level cabinet ministers, supreme court justices and so on. Many would easily engage you in conversation, too; maybe because I didn't open with political subjects and mainly complained about having to go to Brasilia.

The first problem with Brasilia is that it's horribly located. We did avoid the Stockholm fate of having our historical cities razed, but moved the seat of power to the middle of nowhere. I know that overpaid lower-ranking bureaucrats have houses in the outer rings, near the lakes; but no one who has an actual choice lives in Brasilia. I don't know if it fully qualifies as a city. If the capital ever moves, it will be a ghost town in three weeks and start crumbling (bad architecture and bad construction) within the year.

Yeah. I was kind of infatuated with the UMAP paper when it first came up (and working through it to build intuition led me to persistent homology and other fun stuff), but it's far too complex to work as a "plug-in tool" that people naive to statistics can blindly use.

This is the power of ANOVA, CCA, linear regression and factor analysis, etc. It's much, much, much, much harder to shoot yourself in the foot, particularly with experimental data (usage in the social sciences is another bag of johnsons.)

You keep using that word.

UMAP.

PCA is pretty close to optimal in situations where you want to preserve global structure and variance-covariance matrices are highly informative. This comes down to the min-max theorem for Rayleigh quotients; I won't grandstand going over the math here. It won't work well for gene sequences or text embeddings because the desired structure there is local (UMAP seems to reign supreme these days for that).

An older explanation for PCA is that it's a sort of default factor model prior to rotations that introduce structural assumptions. (Varimax etc. factor analysis is really underrated in exploratory statistics; but then, by now data science training never introduces people to the FWL theorem, identification, etc. With large enough deep learning pretty much anything is possible -- also because using deep learning implies having truckloads of data -- but the median xgboost guy is way out of his element.)

Doesn't seem to be a Metafilter.

Qapla'!

(Are you somewhy trying to emulate rms's looks?)

I only briefly knew R in grad school (circa 2007) but I lived inside Stata for about 7 years, and yeah, while specific frustrations vary, the general tenor...

Then -- I once made a meme to that Oliver Stone Vietnam movie that said "This is my copy of Stata. Without me it is useless. Without it, I am useless. I must cherish it as I cherish my life..." (In the original said of a rifle.) I was good with Stata, fast and precise and never "ugh, okay, let's open a quickie notebook... there goes my morning..."

This kind of discourse sort of exculpates the outraged. People have agency, however imperfect. We've gotten in a rut of blaming platforms for people's behavior -- both assholes that post in social media and brittle-porcelaine people who can't stand this torture.

Something needs to be done policy-wise about Twitter and Facebook, but we should also be telling each other to chill the frak out -- and to consider the truths (and every ideology has some truth to it, that's how it acquires verosimilitude and grows) behind the asshole's worldview. Maybe those damn transphobes are looking at some things that we've become blind to.

These things shouldn't make you this angry.

- BLM is unusually opaque about how it uses donations, particularly for something that's promoted by XKCD (usually a seal of quality).

- The current understanding of gender is very very new, and JK Rowling is 56. Also, are we saying no trans kids ever regret their choices?

- There are complaints by female athletes -- possibly just sour grapes -- that transwomen are unreachably strong. The usual rebuttal is that hormone therapy "undoes" in some sense the muscular advantage that testosterone produces, but the timeline for that is unclear, and not all of the peak male physical superiority is due to sheer muscle (bone structure and mass, etc.) These aren't talking points, it's other people's lives.

At any rate, your complaint is that someone is wrong on the internet! That does happen every once and then.

What aggravates me the most is seeing people argue that "first amendment applies only to government, corporations are free to do as they wish".

Yes, but didn't we used to be in favor of free speech as an abstract dignity of humans?

I made my own graph-based note-taking tool where you don't write notes, just titles. It's built on Streamlit and really rickety, but (when I have time to tend to it, which I haven't for months with the baby) it's been a very interesting way to compose thought fragments and find ideas for essays for my blog.

Source: https://github.com/asemic-horizon/sursis

Working demo (used by like two close friends besides me; please don't vandalize it):

http://legibet.casino-rhizome.com:8501/

[There's lots of fun stuff where nodes have "weight" so we color them by the Gaussian gravity field. But the idea of notes without body is already novelty enough for one sitting, I guess]

Gate gate paragate parsamgate bodhi svaha!

Respectfully, can I ask a couple questions?

- Do you see yourself as a "new" or "recent" woman who's learning about female things now? (E.g. did you know what a pessary was? Did you read women's magazines in the before times?)

- Did your preferred sex (e.g. women for straight males) remain the same? Do you identify as straight or gay (or none of the above?)

It's none of my business, other than it helps me situate you (and maybe inductively other trans women) in these male-female ML things.

"Unusually strong neodymium magnets" were a popular young-adult toy in the late 2000s to early 2010s. I had them both in flat discs and little balls.

Now that I have a baby crawling around I pray that I haven't misplaced any in some crevice of the house...

Table presents p = Prob(knows word | male).

The unconditional probability P(male) is around 1/2 or a little less.

Then Prob(male | knows word) = Prob(knows word | male) P(male) / P(knows word).

---

Now, we're not given P(knows_word), but assuming answers are from a reasonably balanced sample, we know that

P(knows word) = P(male)P(knows word|male) + P(female)P(knows word|female) = [P(knows|m]+P(knows|f)]/2

Going back:

Prob(m | knows) = 0.5Prob(knows | m) / 0.5[P(knows|m]+P(knows|f)]

Which gives us a formula. E.g. for peplum, Prob(m|knows) is

13%/(13%+64%) = 13/77 ~= 16.8%

For "shemale":

88%/(88%+54%) = 88/142 ~= 62.0%

So sometimes the actual "maleness" or "femaleness" of the word is overstated, while sometimes its underestimated.

This isn't a critique of an article, it's a literal comment.

--- Edit:

The drive to procrastinate today is strong. Here are the probabilities for all words.

https://colab.research.google.com/drive/1-UP3qTJ3GZ3BpsA0ZNa...

I mean, "not safe for work" doesn't mean "immoral". Beach volleyball isn't something you should be watching at work, at least not in the kind of open office plan where everybody sees your screen. There's a lot of talk about workplace "lad culture" being hostile to women these days -- possibly enough to mean liability; also you probably shouldn't be hostile to women in that specific way.

It's possible to hold the opinion that it'd be better if we had waited 10 or 20 years more.

Large-scale societal changes -- even ones you think are well past due -- are filled with Chesterton fences.

Now, be nice.

What he seems to be saying is that it was bad for democracy back then.

Maybe it would be terrible if Napoleon had access to nuclear weapons in the early 1800s. Am I saying we should have let the Rape of Nanking go on indefinitely?

This is a common malicious misrepresentation of conservative commentary in general: even a slightly counter-progressive reading of history (maybe things haven't just gotten better and better in every possible way) is often painted as wanting to turn back the clock in the most extreme and extremely plausible of ways. Search old HN to see discussion about Curtis Yarvin to see this.

I haven't seen screenplay typography in a while... how did you typeset that?

Scientists do not necessarily see the global picture here -- if anything, they'll be inclined to downplay geopolitical and nat-security issues (or at look the other way) if it produces more and better science. This is excluding any consideration of funding -- just what a scientist is supposed to care about.

Some people need to take science more seriously, but there's a kind of crowd that needs to stop worshipping scientists.

    key_value = Infix(lambda u,v: dict(zip(list(u),list(v))))

Let's goooo.

What would be an acceptable error rate?