(Please don't use sarcasm, it's rarely constructive)
I'm saying that HN has a vocal population who are so very sure of themselves and their own opinions that they'll rapidly post messages here declaring that innards of a large, complicated piece of software are so obviously wrong and that they know what the right way is. They seem to live in a world of utter certainty, never a doubt that they might not know all the considerations that could weigh in on the design of some inner working of a complex system.
HN would be a better place if these people could entertain some humility, perhaps a little bit of self-doubt, perhaps an inkling that other people may know something they don't, and that the world is not always the black/white, yes/no certainty that they seem to experience. Or we can just drown in the continual "This is stupid! they should do X" style comments.
The thing was, I didn't actually have any certainty about anything. I just thought that the best way to talk about an issue was to pick a "side" (often at random), and let other people pick their "side", and then defend your "side" every time it gets rebutted, gradually researching the issue from the perspective of your side in order to best steel-man your own position, while assuming others are doing the same in reaction to your own arguments.
In other words, basically pretending internet comments are a debate club, or a court-room, where you're not seeking truth in an Aumann's-agreement-theorem "both people gradually converge onto the truth" way, but rather an adversarial discovery system where the truth comes out for the audience in the compilation of evidence presented by both sides, whether those sides see it themselves or not.
I don't know, it was fun at the time. Maybe I was a teenager with a lot of testosterone.
Until you see such a thread go four or five levels deep, it's not really clear that that's what's happening, and so you might want to join such a thread as a sane, calm individual, only to have your own argument taken apart "by the seams" in response (e.g. asserting that due to errors in your deduction, you didn't prove your point, even if you may be right) without engaging with its content. Good for lawyering, bad for truth-seeking.
If you then are willing to get baited into playing the game in response, honing your own attack to be bulletproof like a lawyer would, then the truth might eventually find its way out. But if you didn't want to play this game, you'll probably just quit when you realize what's happening, maybe with a petulant "you kno what? Screw this." at the guy who just can't seem to get your words through their head (when they never even believed in their chosen position, so there's no convincing to be done.)
If you explicitly indicate that you want to play this game, it could be great fun for all parties. But I know from experience that the other side of this—being baited into playing without knowing the game is afoot—is painful, and can burn people out on a forum community.
Personally, I think the adversarial system is inherently and downright bad and wrong. The idea isn't to get to the actual truth, it's merely to win the argument at all costs. There might be an idea that this efficiently gets to the truth, but I don't think it does at all; this system just finds who's better at arguing. An inquisitorial system I think is superior at getting to the actual truth. See your own comment about "good for lawyering, bad for truth-seeking": I think this is exactly what's wrong with the common-law system.
Personally, do like to play devil's advocate a fair amount, but not to the extent where I need to actually win the argument, I just want to make sure all sides are covered and that people aren't missing something major by hopping on a bandwagon or engaging in group-think. There's a big difference between this and just endlessly arguing one side without actual regard to the truth or the merits of the other sides (and there may be many--another problem with this adversarial approach that always assumes only 2 sides).
Also related, personally I think formal academic debates are similarly bad and wrong, though not quite as bad as lawyering because at least with academic ones someone's life or welfare is not directly impacted by the contest over who's a better arguer.
I disagree with this one. The best, most productive discussions I've seen on HN are those where people have been clear about their views and expressed them succinctly without the hedging and careful politeness that's expected these days.
I've more often heard, or gotten the impression, that discourse is getting more polarized and vitriolic these days. So which is it?
I do quite a bit of work on improving the safety of various code, both professionally and in my side projects. The techniques I advocate are those that I've found effective. But I can't do everything on my own, unless you're offering to pay my contracting rate for me to work on a new kernel.
This is a false dichotomy. Some discussion can be constructive while some is not.
Whereas, suggesting a rewrite can lead to results when the tool is new, small with significant use + bugs, or small to medium with safety- or security-critical usage. Examples of last ones include DNS, web, key, file, and other servers that are mission critical and widely deployed. Example rewrites that actually happened... we'll go with DNS... are IRONSIDES DNS in SPARK Ada and bluejekyll's TrustDNS in Rust. Maybe add Lobo web browser since it ambitiously wrote a browser in pure Java. Dead now but impressive how far project got.
So, yeah, there's certainly pointless discussion that waste space on HN. Arguing for a Linux rewrite is always one of them unless someone with a pile of money dedicated to rewrites is asking which mainstream OS to rewrite in safer language. I'd prefer FreeBSD but probably recommend Linux for higher impact.
Google does use rust in the user layer of Fuchsia.
This is not true. There are even bindings written in Rust for Magneta kernel in fuchsia.
> Is this supposed to be taken as Google engineers have infinite wisdom on how to build kernels and therefore we should copy what makes sense for them?
Main benefit of using Rust (memory safety) is probably not high enough comparing to eventual costs - that's why they don't use it in kernel. You make bold claims about decisions of fuchsia developers. I've checked your comments but only small percentage of them are technical and only saw that you write about yourself as "javascript programmer". What's your expertise in building kernels and operating systems that you think we should not copy it? How many kernels and operating systems did you built? I assure you that this team have a lot more wisdom than you and me combined in this space. Have a little respect.
> and there are lots of explanations for why it is in C other than C is better than Rust.
No one proved that Rust is better for writing kernel to this day because there are 0 kernels written in Rust used in high scale production systems today. Until this will change, C is proven to work in this space and considering that - it's better/safer choice there today.