Yeah this seems like the best tradeoff. You avoid the central control infrastructure and you provide information to clients. It's also a great match with free computing devices, which can then utilize the new information, empowering users (eg parents -> parental control on device, or individuals who want to skip some kinds of content).
There are issues today with this approach such as lacking granular information for sites that have many kinds of context, but if you stop investing in the central control infra and invest in this instead that could be remedied.
Oh wow, I had to try this and as expected it's amazing. Trying to design interesting algorithms that also look good is a lot of fun, and the result is surprisingly readable.
It takes some getting used to symbols that can be confused when upside down such as b or brackets (like the symbols for begin/end)
Like others I am curious about doing it for a lisp or Forth
Isn't it more about the demo of what you can build than the paragraph? I feel like the text would only give a small amount of the information this demo gives
As Sam Harris has discussed several times, if you run the thought experiment of what is definitely real and not to its conclusion, subjective experience is the only thing that cannot be faked as it does not make any claim beyond the experience itself. Even if there is no real world out there, you're just sitting on a hard drive in a super computer being simulated, your subjective experience was proven to be true the moment you experienced it.
There is a distance between your experience and the shared reality (a bunch of signals, perception and processing) that we use science to try to overcome
I think one of the big missing pieces in your parent comment about choice and negative externalities has to do with this network effect and I do not think the argument that it can be overcome locally is a very compelling counter.
It's convenient to run the lint step faster/sooner than at CI/CD time. Depending on your setup, the separate linter deps handled by pre-commit can be more convenient than hassle both locally and in your CI/CD pipeline (re the makefile script you mention).
Having done it both ways several times I lean pre-commit for now
I'm not sure how Twitter works, but for YouTube it's not like their recommendation system is a perfect user-friendly algorithm.
While it is your choice what to click, what you end up watching is a result of the interplay between you and what is recommended to you and if the recommendation system was different, both what you click and the resulting feed would be different as well
Your earlier argument about unsubscribe being effectively an informal termination only works if you properly separate that side from all promotional/marketing/new features/etc material, otherwise it seems you are avoiding the main point and purpose of the unsubscribe button
I've been in this situation many times too, but I have to say it feels like a weakness. When a single person works on a problem I see improvements being left on the table compared to when two people effectively collaborate. I've experienced this in a wide range of skill/experience levels so I don't think that is the problem.
That said, I don't know how to change the situation if you have devs with skillsets that don't seem to overlap much, which seems inevitable for some companies
I could see this the other way, where the reason this path happens is because of a problem with how the team operates so others are not in the loop on the problem or , even worse, the entire system. I think what the parent is proposing is trying to combat that entire issue
I can see how it scores poorly on the number of scrolls per paragraph of information, maximum words read per unit of time, etc, but clearly that is intended with the choice of spacing and font size which is different in the docs (typically optimized for providing information, similar to what your metrics seem to try to measure).
I think it is more usefully analyzed as a landing page, with a lot of emphasis on how it looks, at the deliberate tradeoff of the text being harder to access.
On the practical side, personally I'd expect to find docs, getting started, and source repo which I found immediately.
Some surprises though: I didn't notice the scroll until after looking at links (intended?). The way some elements disappeared if you scrolled just slightly past them entering the page. I didn't find the scroll jacking surprising after first seeing it, I read it as "not a normal text document" shortly after.
Of course you might in the end not care for the visual side, or not expect it to come at such a great expense in terms of scroll # or similar metrics. I think that's a risk they take with this kind of page.
As someone who is somewhat dismissive of labels, I've started thinking more about this. In a context/culture where the way you act isn't as accepted/appreciated it seems labels are a useful thing to defend your way and for people to rally around (since a single term for something makes it easier to recognize, for instance).
I think it's natural for them to show up in the kind of culture clash that happens when groups of people who don't feel as accepted/appreciated try to make a positive space for themselves.
The grandparent they are agreeing with: "They are trying to get on his nerves, he should go to the police and stop seething on the internet cause he's giving them plenty of what they really want."
seems to suggest trying to make repercussions happen without feeding a harasser with reactions. That doesn't seem to be ignoring the threat, just trying to deal with it more effectively.
I think it warrants investigating exactly how and when Copilot reproduces code, but using one example to write it off as just copy and pasting seems excessive.
Also when talking about rights, whether or not Copilot copies doesn't seem sufficient to make a call. For instance, if it has to be coerced by the programmer to produce these kinds of snippets in an obvious way, then it seems fine to lay the blame on the programmer similar to when using regular autocompletion (or copy+paste for that matter).
I think the discrepancy between what actually happens in the code and the main ideas in the paper is a great point, and it touches on the parent commenter's goal of attaining knowledge. For instance, even if the results are reproducible, maybe the proposed key idea in the paper is not the piece of code with the largest impact on performance. Though, with the code available you might be able to discover exactly what the discrepancy is if there is one and you might be stuck at "it doesn't reproduce" otherwise
I agree with the sentiment, but I also think designers and builders of all kinds ignore the most advanced users at their peril regardless of if they're HN, some game's best players, someone who uses a library in production instead of as hobby, etcetc.
The impact is just different and sometimes causes big issues if ignored
There are issues today with this approach such as lacking granular information for sites that have many kinds of context, but if you stop investing in the central control infra and invest in this instead that could be remedied.