It's both. Look up council complaints and demands for the HS2, they're full of unreasonable NIMBY stuff like asking for the line to be put in a tunnel because it comes closer than 100m to an industrial estate. They didn't get it of course, but as a "compromise" they got high berms after a lengthy consultation/bargaining, both of which inflate the cost.
The first geological repository goes online next year in Finland https://en.m.wikipedia.org/wiki/Onkalo_spent_nuclear_fuel_re... . It's not in the sparsely populated north of the country either. There are no problems to speak of when you bury something under 500m of bedrock in thick copper and even thicker bentonite clay. There just aren't.
Something else that came to mind while I was reading the comments: you do want phonetics/pronunciation classes when you're C2+. Do learn IPA and learn the sounds, it's crucial not just for your accent but for understanding others. I found [1] very useful, but only combined with a great tutor.
Something I didn't appreciate enough is that we don't actually hear sounds when we hear people speaking. We hear phonemes, which are clusters of physical sounds that make semantic difference in the language. The clusters themselves aren't fixed either, they are very loose and mostly defined through what they are not — i.e., the difference that we perceive in "lip" and "leap" is not absolute, the actual sounds might easily overlap between speakers, but we adjust to the particular accent/speaker using the fact that they probably still have two separate phonemes there.
It works very well until one starts to learn a second language that might have not just different "clusters", but a different number of them. My first language is Russian, and in Russian there are just fewer semantically meaningful vowels; I honestly thought that the word "milk", молоко, has three roughly equivalent sounds, whereas in English that'd probably be heard as two or three distinct vowels ([məɫɐˈko]). Similarly, Russian "soft" sounds like м in мята are widely heard as having "j" in them, "m-ya-ta", while native speakers just don't hear that.
Phonetics training helps to start actually hearing all those sounds, to adjust our inbuilt clustering and start perceiving things that natives do. You suddenly start understanding native accents much better, and gain a new appreciation for the language and its beauty.
It's just as much about perception as it is about accent.
Two things I observed while slowly getting better at English.
First, the "sophistication" can backfire. There're a lot of comments about reading here, but there're very fine lines between
- "simple English", think a stereotypical ESL speaker
- "well educated" English, think a native posh college alumni
- "colloquially broken" English, the way native speakers speak to their friends
- "out-of-place highfalutin" English, a hallmark of someone who didn't have a chance to experience the variety of contexts growing up in an English speaking country.
It's quite hard to balance those, but I guess it just comes with time and practice while being mindful of it. For me personally it worked in waves, from unnaturally-broken to too-correct to feeling comfortable enough to break the grammar in natural ways to noticing more unnaturalness to… you got the idea.
Second, and I'm forever grateful to the person who first introduced me to this idea, is realising that high level language acquisition can only come with a new personality attached. It's very weird and disorienting if you're not aware of it happening, but it's a natural and necessary part of it. You need to grow a personality to feel in your second/third/etc language, to react to jokes on the spot, to make friends, to dream, to live in that language context. It often differs from one's identity/personality in the first language, and that's fine, it's just as valid. Embracing the process and the difference makes things easier.
I don't think it's possible to do that through learning though.
Why speculating then? There is plenty of hard data on skin testing of various sunscreens. There are physical sunscreens, too, which were around even longer. We have MUCH better understanding of biological processes as well than in asbestos times. What is this drive that makes supposedly technologically literate people flock to the GP question and come up with unfounded paranoid theories?
I don't understand though how do you explain those transaction to your tax authorities and pay taxes. Surely it can only be so long that you can get massive (by local standards) and fairly constant salary month to month and not get them interested in the source of funds?
It would require quite a lot of it because you need to significantly change the orbit, especially if you want it to still be plenty fast on impact.
And it doesn't make much sense either. A billionaire can just pay a few millions to send a guy in a car bomb or a cessna full of explosives. It's way simpler and just as undetectable until the explosion happens.
What do you mean "it's not working"? We already have BCIs for paralysed people, they're quite bad: very invasive, very expensive, very poor user experience, e.g. [1]. What Neuralink is doing right already looks much better than those and they're only in the business for five years.
You need to separate the hype from the tech, especially so with Musk. The tech looks great.
According to [1] and [2], most German plastic waste goes to Netherlands. Moreover, apparently a third of it gets incinerated or recycled domestically. So yes, officials turning a blind eye to developing countries pseudo-recycling plastic waste is a problem, but 1) it's not as clear cut as you present it 2) surely building a few incinerators is easier than reinventing all logistical chains to not use plastics?
If you live in a developed country, your contribution to the microplastic problem is minuscule regardless of the amount of plastic packaging you use. Landfilled, or better yet burned in industrial setting plastics don't get into the oceans, they are either fixed there for millennia, or don't exist anymore. Plastics are a non-issue for countries with functioning waste management, which is coincidentally the countries that can afford caring about it the most, so in a sense there is a self-contradiction in this debate.
International treaties forcing developing countries to get their waste management together combined with targeted investment there would help way more.
There are at least three distinct use cases that Twitter somewhat covers:
a) participating in already existing, well-defined communities. Mastodon works for that, but so does Discord, Gemini, Gopher, tilda communities, IRC servers, and so on. Mastodon's peculiarities like tiny scale instances, inter-instance admin wars, non-existent search are perfectly fine for this use case. That's what it is made for, effectively.
b) connecting with friends and following their interests. This is just miserable with Mastodon: you might be on different instances that mute/ban each other, you might not find each other if you don't know their instances, if instances are interest-specific you have to follow on a number of instances, if their instance admin gets fed up with it and shuts it down you need to reconnect somehow, etc. Mastodon is not an identity-first design, it's community-first, and it's a problem for this use case.
c) following celebrities or creators famous in the world of atoms. Celebrities would want massive following, which requires massive scale, which is hard to achieve for something effectively run by volunteers. If it's smaller scale celebrities, they'd also want discoverability, which is hampered by the lack of search, federation bans, etc.
Mastodon is widely presented as a "twitter backup/alternative". It's just misleading unless the person moving to it only wants (a). There is nothing wrong with wanting (a)! However, it means that Mastodon has very little similarity to Twitter if people want (b) or (c).
You are right that simplified reimplementations make good property tests, but in this case I'd go the other way around: generate an AST, render it a in test case-dependent way (adding whitespace as you said, but also parens etc), inject known faults for a fraction of test cases, and check that the parsed AST is equivalent to the original one or errored out if a fault was injected. Rendering a given AST is usually simpler than parsing it.