pieix

Good callout — it's the animation rate for your browser, which is driven by your display and conditions on your machine (I've seen between 30-120Hz on my Macbook depending on the monitor used and device power status).

Co-orbitals are top-of-mind right now! As-is they can be simulated without issue (the system is defined as a DAG where each body defines the parent bodies it is gravitationally influenced by, e.g. Sun+Jupiter+Saturn for the trojans) but I haven't yet figured out how to get the proper starting position+velocity for them in the current epoch. I foresee a deep dive down the ephemeris rabbit hole in my future...

I hear you, if we had a punchier name for the Solar System (ideally with its own TLD?) I for sure would have gone for it. Anyway, thanks for the nice comment.

Thank you! The background stars are a texture that I found on solarsystemscope.com, and _should_ correspond to the actual orientation of the firmament WRT your frame of reference. I'd love to add labels for salient stars.

You can currently pan with right click, and I like the idea of a configurable starting time! Everything is mapped into the current epoch when you load your browser at the moment.

A lot of negativity in these comments. I find this analysis of claude.ai use cases helpful — many people, myself included, are trying to figure out what real people find LLMs useful for, and now we know a little more about that.

Coding use cases making up 23.8% of usage indicates that we're still quite early on the adoption curve. I wonder if ChatGPT's numbers also skew this heavily towards devs, who make up only ~2.5% of the [American] workforce.

> I am waiting for time to forget enough of it so that I can read it again.

Incredible book and story, I read it this year too and your statement resonates.

No landing legs on Starship or its booster! Tower catch or bust.

The Shuttle cost $48 billion to develop in inflation-adjusted dollars — contrast this with the ~$5B spent on Starship development thus far. It’s hard to defend the claim that NASA is or ever has been under-funded.

NASA funding far exceeds SpaceX’s and always has. It shouldn’t take 4% of the US federal budget (NASA funding during Apollo) to run a hardware-rich design process!

Surely ~$25B/year is enough to push the frontiers of space exploration?

Is this true? If so it seems like an underexploited loophole.

1) Offline connectivity — pretty cool to be able to debug technical problems while flying (or otherwise off grid) with a local LLM, and current 8B models are usually good enough for the first line of questions that you otherwise would have googled.

2) Privacy

3) Removing safety filters — there are some great “abliterated” models out there that have had their refusal behavior removed. Running these locally and never having your request refused due to corporate risk aversion is a very different experience to calling a safety-neutered API.

Depending on your use case some, all, or none of these will be relevant, but they are undeniable benefits that are very much within reach using a laptop and the current crop of models.

The first failure wasn’t an overworked booster, it was a second stage which are always new off of the factory floor as they aren’t recovered.

OCaml is the one language I’ve used whose standard library is great to read. It’s a very developer-friendly language but in all of the ways that make it popular on HN and rarely used in the real world.

> This forced top down push for density

You have it backwards — the push for density is bottom-up, coming from the people who want to live in SF. The top-down elements are the regulations imposed to keep the city from changing.

Agree completely. For me, the author missed the mark with:

> modern, blocky pixel art often is a kind of misdirected, anachronistic nostalgia.

All of _my_ pixel art nostalgia comes from the 240x160px LCD Game Boy Advance and crystal clear blocky 16x16 sprites.

> My anecdotal receipt: None of the 15 functions we run in prod are switching to the 50% off model as product quality is degraded with gpt-4o.

Interesting anecdata — how did you validate this?

Also not Bifrost, the synthetic training data generation startup.

As you’ve identified, few people will opt into a lower quality of living in the name of the greater good. The solution is to find better power generation methods, rather than force people to accept a downgrade.

> The amount an author reads doesn’t correlate to success.

This is, in practice, not true — good writing doesn’t emerge spontaneously, it’s nurtured by reading other good writing. As Cormac McCarthy put it, "the ugly fact is books are made out of books."

The 2nd Amendment was in practice not honored in California until June 2022 when the Supreme Court ruled in New York State Rifle & Pistol Association v. Bruen. SF still didn't issue its first concealed carry permit until January 2023.

Government organizations should be wound down more regularly.

Nobody would argue that a business unit that has outlived its usefulness should be funded in perpetuity, yet there seems to be this "strictly increasing" mindset around organizations funded by the taxpayer.

> if you went and saw what was going on you might question if these were geniuses pushing the frontiers of expression or a bunch of degenerates exploiting impressionable young people.

To be fair, a lot of people still have this question about the Beat Generation.

Likely because it's absurd to state that politics in America are "beholden to white supremacist conservative Christians."

Ten years is a lot of time, but is it enough time for the current Koalas package [0] to truly die and leave the name for the next dataframe library?

[0] https://koalas.readthedocs.io/en/latest/

I'd phrase it this way: at a smaller company, you are much more likely to be treated as a human, by a human, versus being treated as a unit by a system at a bigger company.

Except, you know, a child has feelings, and a deep learning model does not. The whole reason it would be wrong to tell your kid that is because your kid is a person with feelings.

And yet, the US and capitalism are both still around, and their continued existence ensures we don't get people in comments sections claiming that we don't know how capitalism works as a system because _real_ capitalism still hasn't been tried.

A successful system is robust to tampering from external and internal actors.