As a statistician who has implemented models nowhere near the scale, and as someone who has experience in distributed computing, a lot of the actual work comes from global optimization, which means that all the state has to be communicated somehow. Now this state can be distributed, but it needs to exist, so now you need to have the same resources required to train the model, plus the resources required to shuttle the state around in a distributed manner. It’s cheaper and faster and more ecologically friendly to build better hardware.

I agree that Slate has an interesting idea but they have an uphill battle with the marketing. Unfortunately, the US, to a great extent, particularly the rural US, sees electric cars as being “more tech”, which they equate with more frustration. Naturally EVs actually could likely have simpler computer systems than ICE cars, but, for instance Tesla chose to slap a huge iPad in their cars when consumers want dials.

For all the technical superiority of EVs, it’s hard to convince the average person that they should drive a novel looking “high-tech” vehicle, when a Toyota Camry is an easily available, affordable option. Most people with F150s don’t need them, they want them. Slate isn’t competing for those people, it’s competing for the people who buy Impreza’s, Corollas, and Camrys. I’m not sure it’ll work out.

For an industrial on-site work truck for running errands, I’d bet they sell like hot cakes, but that’s a much smaller market.

They both are beautiful and impressive artworks.

Personally, your latest render evokes a sense of warm cleanness. I take it as a sort of illumination of a dark but hopeful place.

Your earlier render feels more outdoorsy (but not necessarily more organic, if that makes any sense). It conveys to me more a spirit of perseverance in contrast to the first one.

That is to say, while everyone may have their own opinions, both pieces of art to me, are incredibly complex works, and you’re clearly insanely talented, so you make it hard to have a preference!

Your latest work does seem to have an incredible amount of meticulous work and effort put into the lighting, and this one seems to focus more on the background, but either way, they are stunning!!

deleted by creator

deleted by creator

This github page isn’t visible on my mobile device because the ads block the view.

The concept sounds truly interesting but distribution is everything. The AdSense you have is probably not very profitable and is actively hurting your recognition. As politely but bluntly as possible: If you want appreciation and adoption, remove the advertisements. You’re selling us on your idea, not whatever bottom barrel consumerism Google wants me to buy

Behold: PRQL. I only know it exists not if the errors are good, my SQL needs are simple, but perhaps for some complex data wrangling it could be nicer idk

Perhaps the textbook example is the Simpson’s Paradox.

This article goes through a couple cases where naively and statically conclusions are supported, but when you correctly separate the data, those conclusions reverse themselves.

Another relevant issue is Aggregation Bias. This article has an example where conclusions about a population hold inversely with individuals of that population.

And the last one I can think of is MAUP, which deals with the fact that statistics are very sensitive in whatever process is used to divvy up a space. This is commonly referenced in spatial statistics but has more broad implications I believe.

This is not to say that you can never generalize, and indeed, often a big goal of statistics is to answer questions about populations using only information from a subset of individuals in that population.

All Models Are Wrong, Some are Useful

• George Box

The argument I was making is that the NYT will authoritatively make conclusions without taking into account the individual, looking only at the population level, and not only is that oftentimes dubious, sometimes it’s actively detrimental. They don’t seem to me to prove their due diligence in mitigating the risk that comes with such dubious assumptions, hence the cynic in me left that Hozier quote.

“Wet sidewalks cause rain”

Pretty much. I never really thought about the causal link being entirely reversed, moreso that the chain of reasoning being broken or mediated by some factor they missed, which yes definitely happens, but now I can definitely think of instances where it’s totally flipped.

Very interesting read, thanks for sharing!

I feel this hard with the New York Times.

99% of the time, I feel like it covers subjects adequately. It might be a bit further right than me, but for a general US source, I feel it’s rather representative.

Then they write a story about something happening to low income US people, and it’s just social and logical salad. They report, it appears as though they analytically look at data, instead of talking to people. Statisticians will tell you, and this is subtle: conclusions made at one level of detail cannot be generalized to another level of detail. Looking at data without talking with people is fallacious for social issues. The NYT needs to understand this, but meanwhile they are horrifically insensitive bordering on destructive at times.

“The jackboot only jumps down on people standing up”

• Hozier, “Jackboot Jump”

Then I read the next story and I take it as credible without much critical thought or evidence. Bias is strange.

I think many people don’t like it conceptually because the advertising for Brave is:

Built in Privacy + Crypto + Ad Blocking

Firefox + uBlock Origin suffices well enough for most people. It’s stable, suits the purpose, and separates them from a company entangled with crypto.

Everyone is just trying to do their best to balance convenience with the social impacts of their actions. People make change because they care, either altruistically or personally, but it always comes with some sort of personal cost. Putting your neck out there and trying to make a change is more important than any specific browser choice

I’ve done a bit more searching and it seems ltex-lsp-plus is the best out there for lsp grammar checking. It’s 1000x better than nothing, though the false negative rate is a bit high for my taste :)

deleted by creator

I’m not sure what kind of diagram you’re after, but Typst has Cetz which is graphing + arbitrary drawing of shapes, paths, splines, etc.

Typst also has fletcher “maker of arrows” for diagrams which is my personal fave for the work I do

https://typst.app/universe/package/fletcher/

Word definitely has its niche.

However, I find for many of my tasks, LaTeX or Typst just make sense. I don’t need to worry about out of date figures. I can customize styling instantly. I can track my changes with Git. Grammar checking is rough tho. lsp-like grammar checking would revolutionize my world lol.

I can personally attest that I transitioned to LaTeX from Word, when Word wouldn’t handle equations correctly, or would crash when I had too many. It doesn’t matter if I can put out 50 word equations faster than LaTeX if I’m breaking my flow state to restart my editor.

They overlap in their ecosystem niches but in no way is one a complete replacement for the other. LaTeX has a larger niche than Word which makes it a really safe default.

“Nobody ever got fired for choosing React”

Do you know of any Nix projects which are basically nix-but-as-if-was-brew?

I get that this violates the Nix philosophy, but it’s hard convincing collabs to install a root package manager, which has install commands like:

nix profile install nixpkgs/nixos-24.11#hello

I get that it’s flexible, but I would like something more like:

nix install hello

I want three things:

1. rootless
2. can manage “casks”
3. global cli with support for per-project flakes

Do you know if this exists / is being developed?

I would love to see an actual source/docs stating 1875 is a commonly used epoch, rather than microblog posts. Either bring hard facts or shut up, because…

Arguing over the epoch completely misses the obvious refutation. If there’s errors in the database, that might be because there’s hundreds of millions of individuals represented in the DB, and no data should be made to be perfect at the cost of people starving. I would posit the bigger a tech system gets, the more social constraints it will acquire. The errors in the database could mean there’s underserved people, and we should fund efforts to represent these people, so their needs can be met. What errors don’t mean is that the SSA is being defrauded to such an extent that it should be shut down, but the way Musk has worked his claim makes that implication natural.

The epoch could be yesterday or at the building of the tower of Babylon and it doesn’t matter, they’ll just deflect and say there’s people who are ageless in the dataset who are defrauding the system and it’s all corrupt

This article gives validity to opinions of idiots meddling. This is implicit, and perhaps accidental, complicity in an outrageous government outreach. Musk is actively tearing down the government.

Sure tech is cool, but it’s nowhere near as important as the social issues surrounding it and for a tech based newspaper to ignore that basic fact is embarrassing

data.table is so cool! I was shocked to realize it has no dependencies (outside of base), very elegant and very fast for sure.

tidyverse is more than a pandas analogue. It’s more like Pandas (and a little more sugar) + Expression + Altair (or matplotlib) and a few others less used. It’s very beautiful. It aligns well with R and is quite functional stylistically and is usually pretty clean syntactically.

The books are really good, but the docs(tidyverse and R) are kinda poor compared to Python (and Rs documentation tools are very limited — PDFs mostly). R package management is much worse than Python’s.

It’s certainly powerful, it’s certainly elegant, and Wickham is an incredible technical writer.

There’s lots of really incredible research done in the R ecosystem.

Caution: Lots of docs are affiliated with Posit, including Wickham’s. Posit wants to sell their cloud offerings. This often leads to over-optimism in documentation.

The language is definitely capable of serious work, and is pretty good at dataviz, psych, and gis.

I highly recommend giving it a try, if you like functional programming or want to see some cool data science ideas and statistics research.

Perhaps I’m just lost or the fool here, but wouldn’t it be usual that the overhead of spawning a thread would be much higher than just drawing the next pixel? If the post is true, could someone explain to me how a renderer with so much thread contention could optimize drawing on a CPU?