aeontech

AI: "this is a known open problem, I can not solve this, sorry."

Prompt: "Can you try solving it, instead of not solving it?"

AI: "Oh, sure, here you go!"

The tendency of Wikipedia editors is that, when an article is many years old, they would rather flag it for improvement rather than simply throw away years of fellow editors' work.

That's not been my experience, tbh - in my view the deletionist fraction of the editors has essentially "won", if one can put it in those terms. I _think_ there is a (maybe small) group that have decided it is their mission to guard Wikipedia against what they view as cruft or non-notable, regardless of how many years of work these articles may have accumulated. They do not need to be paid for this - they enjoy it. Destroying is always easier than creation.

I seem to recall some study showing that the vast vast majority of edits/deletes on Wikipedia are the work of just a few hundred long-standing editors (citation needed) - which to me confirmed my gut feeling that most new editors bounce off and give up on contribution in short order.

I contributed for a few years, but gave up eventually - it was exhausting to spend time collating sources, collecting information, editing, rewriting, and then having someone come along and propose discarding your work with very little investment from their side.

Stackoverflow has gone through similar calcification - it's nearly impossible to contribute now, or build reputation as a new user, as posts get closed as duplicates or not-relevant.

Hmm, that seems to contradict the article directly - insecticides were used to try to battle screwworm initially and were not really effective - the solution was using sterile male flies to stop reproduction - which would work in South America just as well as it did in North (with sufficient scale)

Official site announcement discussion thread at https://news.ycombinator.com/item?id=48772573

Looks very interesting though!

Lemma 1: you want to protect your users privacy, and are also beholden to regulation enforcing that commitment (GDPR).

Lemma 2: you are obliged by other regulation to offer equal access to user data to third parties, so others can build equivalent functionality (DMA).

Lemma 3: malicious third parties will absolutely try to abuse the access and trick the user into sharing their data by all means possible. You will be held responsible in court of public opinion at minimum and legally at maximum if/when a malicious third party abuses said access.

This is a hard, possibly technically unsolvable problem no matter how much money you might have, because the root issue is not technical, it's the fact that you legally have to give third parties access and no way to control what they do with it - and as others have mentioned in the threads, it's exacerbated by the fact that the regulation doesn't say "this is okay and this is not", it is vague and judges things "by outcome", so you may spend all the time in the world implementing a solution you think will work, and then get hit by fines/lawsuits because the implementation is judged as not sufficient after the fact.

Very interesting - I just installed pg_hint_plan [0] extension a few months ago to get around a query that was confusing the planner too much. Edge case, but when you need it you really need it.

Haven't seen pg_plan_advice before, TIL!

This is beautifully written - don't know how this got on HN, but thank you for sharing it.

Time to revive my account... (scrobbling since 2003, heh) - started in audioscrobbler days, before last.fm

My usage has gone way down since switching to streaming services, this made me search for ways to backfill the data...

For anyone using Apple Music, looks like you can request an export of your data from privacy.apple.com, then use https://github.com/nerveband/Apple-Music-Play-History-Conver... to convert it for Last.fm import (haven't tried it yet, but looks promising).

Duplicate of https://news.ycombinator.com/item?id=47427388

Well, sure - braking is mostly relevant when merging to the slower lane, when merging to faster lane I generally do not need to - since that lane is already moving faster, just need to speed up slightly and time it for the right moment.

My point is, it feels safer and easier to aim to enter a new lane with the aim of "following" someone, rather than trying to rush in "ahead" of someone. But maybe it's just me.

I'll tell you what I specifically and intentionally do when I need to change lanes. I brake slightly, signal, and wait for the person on my right or my left to pull ahead of me, then change lanes immediately _behind_ them. Then sit there for a moment until my following distance evens out a bit.

This ensures that

a) I do not cut anyone off accidentally, and minimize the amount of stress in my immediate part of the universe

b) I will (most likely) have plenty of room behind me after I change lanes, reducing chances of anyone else running up on me

c) If there's noticeable traffic, the time I spend signaling and waiting for the person to move slightly ahead of me gives plenty of warning to the people _behind_ them that I'm about to enter the lane.

Ultimately, yes, of course in principle you're right, when I change lanes, I enter the lane in front of someone.... but I _can_ control whether I enter as far as possible ahead of them.

This is awesome, and deserves its own post!

Since people are posting links to alternatives, another awesome source is the noun project. Has a mix of royalty-free, Creative Commons CC-BY-3.0, and paid license icons.

https://thenounproject.com/

This is a very cool app and list of resources for learning Japanese. Does anyone know of similar top recommendations for learning Chinese (Mandarin or Cantonese?)

While I agree that the training/learning ecosystem is pretty heavily centered in Python, going from that to "Ruby is awful" seems like a very drastic jump, especially if we are talking about the LLM interaction only.

I probably wouldn't write a training system in Ruby (not because it's not doable, just because it's not a good use of time to rewrite stuff that is already available in python ecosystem)... but hooking up a Ruby system up to LLM's for interaction is eminently doable with very little effort.

I am assuming your situation had some specific constraints that made it harder, but it would be nice to understand what they were - right now your comment describes a more complicated solution and I am curious why you needed it.

You don't need it until you need it, and needing it often comes in the form of a lightning strike from blue sky. The counterargument is that having everyone pay a higher amount makes it feasible to actually have this coverage available, when needed, without bankrupting the insurance companies, because the rare astronomically expensive care is covered by the premiums paid by the vast majority of people who are relatively healthy and are unlikely to need it.

Now whether the on-paper prices for medical care in this country actually have any relationship to objective reality is an entirely separate question of course. In general coming from an outside perspective, combining healthcare and for-profit motives in a single system seems particularly likely to lead to all kinds of perverse incentives, but, it's the system that exists, and it seems unlikely to change any time soon.

Oh my god, that makes the Matrix world make so much more sense :)

They almost exclusively compare their model to prior models from 2024

As another comment here noted, the title is missing (2024) - this model was released almost a year ago, last December, so it's not surprising that that's the models they compare to.

Yep. You'd be shocked (or maybe not) at how many people I looking at their phones on a freeway.

I wonder if there's any statistics comparing deaths and injuries from drunk driving versus distracted driving over the past 20 years or so. Is it a comparable at all?

I hope they find a way to open-source it! Seeing years of hard engineering work disappear into a black hole would be truly sad.

teacher chose my essay specifically to repro onto a transparency and place on the overhead as an example of bad writing

Oh man, regardless of how "bad" someone's writing is, this is terrible terrible teaching. Public shaming in front of peers, especially on something subjective like this? Some people should not be teachers. I'm sorry you had to go through that.

guiltily looks up from HN while stopped at a red light

that's beautiful!

Very cool idea. The links are posted to the original source, which is not in Bengali though - is that a barrier?

Or, with the subject matter being tech, most potential readers are fluent anyway?

Oh, I just realized I am guilty of the drive-by-free-association comment without actually saying anything about the subject of the post - sorry!

Very cool to see a team use Jepsen for super early pre-release testing of the system.

I wonder if you wish you had waited for the runtime to be a bit more stable, or you feel this was already well worth the effort, even with some of the identified failures being in "known incomplete" areas? (I could see either side of the argument - waiting longer might give you more valuable failures, but testing early gives you a chance to catch problems before they become baked into the foundation and become more difficult to fix...)

Another tool that feels like sci-fi to me any time I hear a mention of it, is Antithesis [1] - written by the people who built FoundationDB. Could be another interesting integration to investigate in the future to help bulletproof the language runtime?

[1] https://antithesis.com

Aside from obvious Smalltalk influence, this also brings to mind Darklang (that switched to an open-source model recently [1]).

I wonder how this will pan out... very interesting to see new approaches being explored.

[1] https://news.ycombinator.com/item?id=44290653

The ending tag in the demo video is "For stories only you could imagine"

That's... uh... a pretty bold description for a tool where you are in fact outsourcing the "imagination" part to the machine.

I mean... this is a deterministic task, can just run it through autoformatter? Why ask AI to do indentation of all things?

That sounds like stuff of science fiction, can't believe it works. The best part is that it works long distance without having to have satellites in the sky... and is probably un-jammable?

Thanks for sharing this, so cool to learn about it!

Wow, that is a brilliant approach. Thank you for sharing. In the US we seem to be stuck in a local maximum (or local minimum) with little chance of changes to the existing system - I wonder what kind of solutions exist to transform an entrenched institutional process, barring drastic upheaval...