claudius

Sure. If you then take personal responsibility for every accident even when going at or below the speed limit, I could quite live with this. "Personal responsibility" of course means making good for any and all damage you caused, something which you are typically quite unable to do.

The fact that people still have "accidents" and that these "accidents" are ruled "accidents" and not "willfully crashing their 2 tons of metal into something" suggests that people are not ready to take responsibility for their actions and that they cannot be trusted, even with the current speed limits already in place.

From personal experience, I know that most drivers I interact with on a daily basis are reckless and needlessly endanger others for their personal convenience and comfort. This starts at parking on pavements and cycle lanes, not stopping at zebra crossings, going above the speed limit and all sorts of other incursions.

For example, there was recently a statistic floated around that half the people in the UK had been in an accident. Assuming that all these accidents were one-on-one collisions (unlikely, people also "lose control" all by themselves), this suggests that at least 25% of drivers drive recklessly, as they have caused an accident at some point in the past. I do not understand how 25% are "pathological".

Add to this the number of people who also drive recklessly but for sheer luck have not caused an accident yet and those people who have caused accidents without involving other drivers and the above 25% go up even further.

So yes, I stand by my previous comment.

It will simply eventually become illegal to run without these limiters just as it is already illegal to use particularly polluting cars in cities (for example). Of course it will also be illegal to disable them. I don’t see how what you’re describing is any sort of credible loop-hole.

Hu? The "evil" thing here are cars (emitting poisonous gases, driving climate change, making loud noises etc.) and their reckless drivers (racing those cars in inner cities, not paying attention, claiming losing control of their vehicle was an "accident"…). Anything that limits their use and the freedom of their self-centered users is good.

So I am supposed to check the script Google sends me every time I access the page that this script does not now (only in an A/B test! Only in beta! It was a human error!) send all the page content to Google as well? Ridiculous.

I bought a winter coat there for 24.99€ in November and so far it lasted quite well. Given that it’s predecessor was 10x as expensive and was quite worn after 8-ish years, it was ok.

Absolutely agree on tech, "Aldi-Notebooks" tended to have a bit of a bad reputation (not sure if they still sell them…).

Aldi siphons away marketshare from other stores with large overheads and from brands which spend more on advertising than the products themselves. Personally I very rarely notice a difference in quality between Aldi-bought products and those from more expensive chains (e.g. Edeka or Rewe), though the latter are more hit-and-miss (never need to check the expiration date at Aldi, always have to at Edeka…) and much more chaotic. This second aspect, which makes it hard to shop quickly, combines with very slow people working the tills, meaning that the time it takes me to buy a single item in Edeka is longer than what I need to buy a whole week’s worth of food at Aldi.

The random collection of stuff is mostly useless, yes, though there are some items there with sufficient regularity and quality to be worthwhile, e.g. WD-40 and tape. Clothing is typically also ok though neither fashionable nor pretty.

In my limited experience so far, both my (public) employers were happy to pay for emergency exit seats due to my height.

1) Ask the user if they consent. If they don’t, let them continue using the product and do not collect any data. Make both the "Yes, I consent" and the "No, I do not consent" buttons equally large.

2) If they do consent, consider every individual part of data you save. Do not save complete user sessions, instead, before doing anything, decide what you want to test, which information you require to do so and then save only this information. E.g. (using the example elsewhere) if you want to check how long people stay on individual pages, collect a signal on each page how long the user stayed on this page (and nothing else).

3) Anonymize the data as quickly as possible. For the example above, do not store data for each user how long they stayed on each page. Instead, have one counter per page which is incremented by the time the user stayed on the page (and the individual time subsequently immediately discarded). This way you can still figure out which pages are left early but you cannot tie this data to any individual user.

4) If you want to look at individual user sessions, pay people to use the website while you stand behind them (physically), do not collect data from random customers.

All of these.

Do not store user data on your end unless you absolutely have to.

Do not give user data to third parties unless you absolutely have to.

Do not do anything without the user explicitly or implicitly consenting to it.

Example: You have to momentarily store the users IP address in order to serve their request for a website. Remove the IP address as soon as you served their request, because you don’t absolutely need it any more.

Example: You have to hand user data over to your ISP (and their ISP etc) in order to serve their request for a website. Do not hand this data over to Facebook, Google, your mum or anybody else, because you don’t absolutely need to.

Example: If someone is visiting your website, it is fair to assume implicit consent to the above two bits. However, if you provide a service where they can store data on your server (e.g. Dropbox), you should inform the user on how the data is stored so that they can sensibly consent to this (or not). So if you’re storing data unencrypted, inform the user that this is the case. If you’re storing data in your mum’s basement, inform the user that this is the case. If you’re storing data in some country with strange laws, inform the user that this is the case.

> In your prod<2>(a, b, ... example, if p1 and p2 are in the same space, how would a*b know which one to contract? Or do they have different names from when a was created?

ITensor introduced this concept and I mostly just followed their lead – spaces have unique names and tensor legs have a name label and a "prime level". So for example an operator O: A → A would have one leg labelled a[uuid]-prime0 and another leg a[uuid]-prime1. Similar to how one might write O: A → A has elements O_{a a’} when writing it down on paper.

> I made a thing to audit that you are consistent with your indices [1] as an alternative to explicit named-tensor objects. And found, but have not used, another package in a similar spirit [2].

Cool, I will have to check this out!

> Although to be honest, simply writing A[n,μ,ν,c] etc. (using different letters for different spaces) makes it pretty easy to visually check that you are getting this right. This is one of the attractions of the notation, even on paper. Unfortunately np.einsum's string notation makes this harder to see, as the indices aren't adjacent to the variable name.

The problem is not so much the letter-space association but also handling the ordering of the spaces inside the tensor. For example in my code, to do a contraction over two indices, you could do prod<2>(a, b, "tlx,tr,p1,p2|tlx,p1,tl|tl,p2,tr") where the result would then have index order tl,p2,tr. The problem was then that changing the index order in one place (e.g. for performance reasons) meant having to re-check all other places where this is used. If you want to contract a tensor network like (d) in [1], this quickly gets complicated. With named indices, the above becomes a * b and if I change the index order in any place, it gets automatically changed there, too.

[1] https://journals.aps.org/prb/article/10.1103/PhysRevB.81.165...

ITensor [1] has somewhat pioneered this concept in tensor networks for condensed matter physics. If I understand correctly, Uni10 [2], a similar project, is even working on a graphical interface for such networks so that you can "draw" the network and have the computer figure out the best contraction order.

In my own code I’ve recently also implemented such named indices (only had "einstein summation"-like contraction specifiers before) and they make tensor contractions so much simpler to write, especially since you can simply overload operator * and have it figure out which legs need to go together.

As a side effect, it also enforces your algorithms to make sense because you can't simply add two tensors living on different (but equal-dimensional) vector spaces together anymore.

[1] https://itensor.org

[2] https://uni10.gitlab.io/

LDAP?

For your info: Hacker News being responsible for the data its users provide is not a problem for the users, it is a problem for Hacker News and similar platforms who leech off content from the general public for their own profit.

Further for your info: Hacker News does not "need a filter lest you quote a sentence from some movie", it only needs to take responsibility of you commit copyright infringement by doing so.

I’d happily let Hacker News go if it means other such platforms go with it.

Even though a hobby may be competitive, you don’t have to do it competitively. Maybe you just want to go fishing and catch fish and you’re happy if you caught one and want to catch more.

As a hobbyist cyclist, I like having a decent bike simply because it’s more fun to cycle with less effort. Of course I could get an old crappy one to show off my skill and constitution, but that’s not the point for me. The point is to have some fun and while part of this fun is due to achieving goals, achieving a hard goal with a little help such as a good bike is still more fun than not achieving the hard goal, especially if it’s due to comparably-bad tech.

Whether your threshold for bad tech is now a really old bike or a not-newest bike is merely a quantitative difference and in the end subject to personal preferences.

Phoning home by default, necessary to display such "snippets" of course violates people’s privacy. That this is done not to pointlessly wish a happy new year but with commercial motives makes it even worse.

I quite like having functions which take generic "scalars" (double/float/std::complex<double>/boost multiprecision etc) without me having to worry about the specific type everywhere. Of course, sufficiently far down eventually one has to care about the scalar type, but on a high level, it’s nice to ignore it.

The same goes when building abstractions based on this. Tensor<T> instead of Tensor_double, Tensor_float etc.

And then you have all the typical usecases of polymorphism in the wrappers and callers and executables and whatnot which are also to some degree part of "scientific computing".

"If you don’t pay me X we’ll report you under criminal law and you’ll have to pay much more."

"If you don’t pay me X we’ll report you under environmental protection law and you’ll have to pay much more."

"If you don’t pay me X we’ll report you under labour regulations law and you’ll have to pay much more."

How would GDPR be special?

Well, none? To install GCC, Homebrew wanted me to install XCode first. That’s when I gave up because nothing was installed there by default except for an outdated Bash.

There's no pre-installed compilers on it (at least as of 2013) and Homebrew needed me to install XCode first before it could do anything.

Admittedly I stopped looking into it and just shelfed the laptop until requesting a saner one.

Install a compiler without an Apple account?

> It's really even worse than that, because when you say you make $60,000 in the US vs. $40,000 in Europe, the $60,000 doesn't include employee benefits like employer-provided health insurance, whereas the $40,000 doesn't include the taxes you have to pay for the equivalent government services.

And the 40'000€ also doesn’t include your employer’s contributions to public social security (approx. half of health insurance, half of unemployment insurance etc.).

> You do realize the whole concept of a startup is about raising money because you know you can’t both grow fast enough to conquer a new market while at the same time making profits ?

But undoubtedly a start-up during that phase also benefits from public education and roads, from state-subsidized internet and transport, from security provided by the police force etc and hence should also pay its fair share towards those costs?

Check out scipost.org. Two-way open access, community-run and public peer reviews.

The path doesn’t need to be hardcoded into the binary, the server could instead unveil those paths you specified as data directories.

Relatedly, I found it useful to pass -g1 and not just -g to GCC. The resulting debug information is a fair bit smaller and GCC itself also takes less RAM and less CPU time to compile template-heavy code (going from 10s of GB to 1-2 GB on one particularly nasty object file).

> Well firstly, I don't why you're putting "academic" in quotation marks, it would be a wild assertion to try and claim these are anything other than academic journals.

Not really, given the miserable quality of the papers they publish. Maybe ‘newspaper‘ or ‘journal’ might be ok, but ‘academic journal‘ requires a certain amount of quality.

> The problem with this article is that what it's methodology tests whether it's possible to get low quality papers into certain academic journals, but it's stated theory and conclusion are that academic journals are deliberately publishing bad science to push an agenda that this article calls "Grievance Studies".

Academic journals publish papers deliberately. They don’t ‘accidentally’ publish bad papers, I can assure you that nobody accidentally clicks the ‘publish’ button. Whatever further agenda the article may or may not push pales in comparison to the actual problems it uncovers.

> The bad guys here are the random authors who have designed a bad experiment, deliberately set out to undermine academic institutions, and then rather than presenting real evidence have written a political hit piece.

Claiming these people undermine academic institutions is like blaming your upstairs neighbour stomping on the ground for your house of cards imploding. If a journal accepts and publishes such submissions, it was never an academic institution to begin with. The completely sufficient evidence for this is the acceptance and publication of those papers.

While authors ideally should act in good faith and only submit good papers, the whole idea of having journals and editors and peer review is to filter out bad papers. The problem here is not some people "deliberately submitting papers" which "have problems", the problem is that those papers were accepted by "academic" journals. This is not "manufacturing" evidence, this is simply evidence.

The bad guys here are not random authors but the journal editors and peer reviewers accepting this stuff.

The problem with callgrind is (http://valgrind.org/docs/manual/cg-manual.html#branch-sim):

> Cachegrind simulates branch predictors intended to be typical of mainstream desktop/server processors of around 2004.

In other words, the data produced by Callgrind may be suitable to find obvious regressions, but there still may be more regressions which are only relevant on more modern CPUs.