aeonfox

I was thinking more if you write a prompt into an IDE that has first-party integration with an LLM platform (e.g. VS Code with Github Copilot), it would make sense on their end to reduce and remove redundant input before ingesting the token into their models, just to increase throughput (increase customers) and decrease latency (reduce costs). They would be foolish not to do this kind of optimisation, so surely they must be doing it. Whether they would pass on those token savings to the user, I couldn't say.

Do the larger LLM platforms just do this for you? Or perhaps they do this behinds the scenes, and charge you for the same amount of tokens?

Except for the times you do want it to run the CI.

LLM issues can often be solved by being more and more specific, but at some point being specific enough is just as time consuming as jumping in and doing it yourself.

Looking at their misrepresentations and over-exaggerations regarding Erlang it now seems like a long lead-up to a sales pitch. Their motivation to exaggerate deficiencies in existing approaches is to lend chapter 7 more rhetorical punch. All the same, I'm keen to hear what they have to say.

True. And that the subtle bugs were then picked up by static analysis makes the safety proposition of Erlang even better.

> Bugs that happen regularly in prod

It depends on how regular and reproducible they are. Timing bugs are notoriously difficult to pin down. Pair that with let-it-crash philosophy, and it's maybe not worth tracking down. OTOH, Erlang has been used for critical systems for a very long time – plenty long enough for such bugs to be tracked down if they posed real problems in practice.

A real interesting read as someone who spends a bit of time with Elixir. Wasn't aware of the atomic and counter Erlang features that break isolation.

Though they do say that race conditions are purely mitigated by discipline at design time, but then mention race conditions found via static analysis:

> Maria Christakis and Konstantinos Sagonas built a static race detector for Erlang and integrated it into Dialyzer, Erlang’s standard static analysis tool. They ran it against OTP’s own libraries, which are heavily tested and widely deployed.

> They found previously unknown race conditions. Not in obscure corners of the codebase. Not in exotic edge cases. In the kind of code that every Erlang application depends on, code that had been running in production for years.

I imagine that the 4th issue of protocol violation could possibly be mitigated by a typesafe abstracted language like Gleam (or Elixir when types are fully implemented)

As pointed out in the OP comment, it's basically 'money for jam' by the point he releases the source code:

> It's an entirely different thing; he made a thing, sold it, and then when he couldn't sell more of it, gave it away. That's nice!

Carmack has extracted as much profit as he could care for from the source code. The releasing of the code is warm fuzzy feelings for zero cost, while keeping it closed source renders zero benefit to him.

Well done

More succinctly, Carmack only contributes his code to OSS, but not his time, and shouldn't impose his values on the wider community that contribute both.

> technically correct, but in context I think that's somewhat beside the point

Talking past people to argue on semantics and pedantry is a HN pastime. It may even be it's primary function.

I never said percentage.

"<small_boutique_company> has filled 8 verified positions on <product_name>"

Add a visible count of how many verified applicants got verified jobs at the verified business. Higher count increases confidence.

Has a registered business that's been around longer than 6 months.

Has a confirmed business phone number, called by a human that verifies another human is on the other end.

Has a confirmed business address (mail them the confirmation code).

Has a website that's been around longer then 6 months.

Has a Google Maps location that's been around longer than 6 months.

> You could just as easily describe someone with bootstrapping experience as being like an FAA crash investigator who investigates take offs.

Takeoff systems aren't analogous to prototype development. I don't know you'd build a prototype plane that's feasible to take to market, without having deep knowledge about how planes are built.

Early design decisions matter. And you don't get to that realisation without dealing with legacy systems where some upstart made terrible decisions that you're now responsible for.

Still nearly 14x the Pinto instead of 17x. Four fire fatalities for 34,438 vehicles is pretty concerning.

> I think the blame more rests on Apple for falsely representing the quality of their product

There was plenty of other concerning stuff in that article. And from a quick read it wasn't suggested or implied the VO2 max issue was the deciding factor for the original F score the author received. The article did suggest many times over the ChatGPT is really not equipped for the task of health diagnosis.

> There was another problem I discovered over time: When I tried asking the same heart longevity-grade question again, suddenly my score went up to a C. I asked again and again, watching the score swing between an F and a B.

> Reading between the lines

Tbf, and in support of your broader point, there's no reading between the lines, because genuine intent is indistinguishable from deception with this kind of stuff, because the latter imitates the former. There's only expecting the worst, and being only occasionally wrong.

> But if traffic/views are what you care about, regular submission would do "better".

Traffic and views can lead to comments and discussion. I was definitely looking for feedback. I think there's a minimum threshold of interest that's required before people start engaging. But it could also be that my submission was super boring, or there was something else off-putting about it – I'd be surprised if it was the latter, because the HN audience isn't known for holding back on criticism.

> it's visible in both places

Yeah I corrected myself in another comment. But I find with submissions to `shownew` they don't enter `top` straight away, or at all.

> You get a lot more traffic over a longer period of time

With a regular submission, it gets on `top` for at least a short period of time as well as `new`, whereas that doesn't seem to always be the case for a Show HN. And certainly not in my case. That said, many Show HNs do make it to top with less votes than the tiny few I got organically, so maybe I tripped some HN's filters? Or maybe it was bad timing? But if I post again, it'll be without the Show HN.

As in you think it's as bad on HN as it is on PH? From what I've heard PH is basically dead internet.

I might have been mistaken, I think if you post to Show HN, it puts you on both shownew and new. I think in my case I saw Show HNs with less points and engagement on the top list, and somehow mine never appeared and I wonder if maybe there's some cherry picking there.

It's a bit like ProductHunt. The only people that look at ProductHunt, are the people that post on ProductHunt—or at least that's its reputation. Though that's an unfair comparison, because I think on HN there is actually genuine engagement by other builders (or perhaps I'm mistaken about PH).

With Show HN vs a regular submission you're shoved on shownew, which gets a lot less eyeballs than new. If you get enough votes, you're supposedly moved from shownew to top stories, but somehow 5 votes wasn't enough for me (though I saw other posts that got there with just 2). I'd like to see someone attempt to persuade me otherwise, but I really don't see the value in using the Show HN: prefix.

The other flaw of the study is that it doesn't compare populations with no social media, to the ones with.^ Just because you're not on social media, doesn't mean you're not affected, specifically where bullying is involved. You might switch off social media because of this, but have worse mental health outcomes.

^ No-one can do this kind of comparison anymore because no such population really exists, save for the Amish and extant hunter-gatherers who would have plenty of other confounders contributing their mental health measures. However the upward trend in suicide since the late 2000s pretty well correlates with smartphone usage. As does the drop in fertility rates (which are now showing to not necessarily correlate with the oft-cited suspects like wealth, female educational attainment, etc.)

It's kind of beside the point. You could argue the change of mind/heart is unlikely, but if indeed they had changed, it would be better to encourage that. Perhaps they were lured by the money, or perhaps some jingoistic impulse, but then the reality of what the company was doing became clearer? Or their world view evolved?

That said, if some ex-Palantir worker was somehow working for UNICEF – to take an extreme example – it would be a little awkward unless they had denounced their old company in a fairly public manner.

You're missing my point. If all 'ethical' companies treat all ex-employees of 'unethical companies' as unemployable, they are effectively only going to work at 'unethical companies' regardless of whatever mindset has shifted over time.

The options are a) they have to decide between starving their family or continuing compromise their morals and increasing the capabilities of immoral company X, or b) a more ethically aligned company removes them from the resource pool of immoral company X. Which world do you prefer?

> Wouldn't it be better to just go with nuclear?

But for economics. Renewables are simply the cheapest option for generation.

For reduced land use, and hence reduced impacts (overall) on the environment and agriculture, nuclear wins hands down. But decades-long lead times, radioactive waste disposal, encumbering safety regulations, water supply etc. etc. etc. are problems you don't have with renewables.

The irony is that Scott Adam's himself wouldn't have been in favour of policing one's own thoughts. /tongue-firmly-in-cheek

> The motorcycle is real, you can buy it.

Pretty major point of difference though isn't it? They claim this will be out there in Q1 2026, giving a chance for sceptics and industry professionals to tear it down. If this is really the long con, they'd at least give a little time for the cheques to clear before their wild claims could be assessed?

The other possibility is that they have very little moat with this new battery tech, because it's so easy to manufacture. Being the first mover might be their only play, and it's only a matter of time before someone else figures it out (or it leaks).

I guess we can't really know that it's real until this guy gets hold of a whole bunch to test

https://www.youtube.com/@WillProwse