tsbinz

You pay out dividends or buy back shares if you have surplus cash, and it's hard to convey to voters that businesses with surplus cash need help I guess.

In zsh you can use tab to expand globs/braces/variables as well.

He's trying to get the point across that "X don't really know on a fundamental level why X" for any group of people X and observation Y, if by "on a fundamental level" you mean "is not based on something assumed to be true". He's not uncomfortable with it, anyone with a scientific or engineering background knows that (or at least should know that). That doesn't stop predictions obtained from well-tested models from being useful.

Regarding "fully" understanding physics, I like to refer people to Feynman's answer to a question about why magnets repel each other[0].

"Assuming something" is one integral part of the scientific method. You formulate a question, you build a hypothesis from prior knowledge, you make predictions, you test them, you analyze your results which might lead you to change your hypothesis or not. "Greatness" is finding simple hypotheses (like "the laws of physics are the same as viewed from any inertial frame and the speed of light is the same for every observer") that predict/"explain" effects that were inconsistent with prior hypotheses. Of course you can never really prove a model of the world is correct - another model that predicts differences that are too small for us to measure might be the "truth", but, until we can measure it, there isn't much use in pondering ...

[0] https://www.youtube.com/watch?v=36GT2zI8lVA

The phone already is where the fancy camera is. The iPad back camera is far worse than the iPhone's cameras.

Italy stopped flights from china before the US, it didn't work so well for them. https://www.forbes.com/sites/davekeating/2020/03/12/italy-ba...

It depends. The seasonal flu mutates enough between seasons that last year's immunity doesn't help you much. If we're lucky, this one doesn't do that, or it mutates in a way that symptoms aren't as severe (since causing the host to die or to be isolated in ICU is also bad for the virus, so there's evolutionary pressure towards less severe symptoms).

This doesn't seem to be from 2016 (some identifiers suggest that it is, but it cites papers up to 2018 and it says that a tweet from 2016 is two years old).

Previous discussion (now both entries are on the frontpage): https://news.ycombinator.com/item?id=22202110

This works for a literal, but the author was probably thinking of "print a string that comes from somewhere else" where you do the first thing to avoid format characters to be interpreted.

It happened to an old lady in switzerland recently - she didn't hang up, the scammer just got quiet after saying he would hang up so she could call the police, and then had someone else answer the lady when she assumed she was talking to the police.

This is the question that should be higher up in the thread. This whole discussion makes no sense. Photons carry no hidden identity.

Where's "here"? In Switzerland you also see lots of burger replacements, imported like beyond or "homegrown" like the green mountain burger.

A friend of mine had that happen with iOS (iCloud sync settings that she disabled changed after an OS update) so I wouldn't feel safe from this kind of trickery after the OS switch.

“If you logged into our network through social media we also have access to portions of your public social profile, such as your name, email address, and photo.“

Obvious in hindsight, but the profile picture is probably a goldmine for targeted advertising. Not sure if anybody does that but using that for age estimation would probably be way more accurate than the guesses I’ve seen implicitly (lots of ads like “people in <age group> love this product”) and explicitly.

"But in this case, the researchers used a custom-made microscope that shines a light onto the sample and measures the intensity of the light that bounces back."

That's how eyes work as well. Well, they need a light source to help, but otherwise ...

You are misreading the article you are citing. Quoting your article, "Contrary to misrepresentations in the media, this risk need not arise from spontaneous malevolent consciousness. Rather, the risk arises from the unpredictability and potential irreversibility of deploying an optimization process more intelligent than the humans who specified its objectives." See paperclip maximizer. Even that is too narrow for the questions they asked IMO - someone who says that the development of an artificial intelligence will be "on balance bad" could be thinking that this AI would be in the hands of the 1% and so further contribute to inequality, even if it doesn't act unpredictably.

https://arxiv.org/abs/1906.04721 is an example paper where they perform data-free quantization after training (using batchnorm parameters to get information about the data distribution) without large performance drops.

This sounds like in theory and makes you curse in practice. Trained TF model -> deployment using TF Lite is the most robust pipeline currently as far as I know.

There are lots of improvements going into pytorch for mobile at the moment, but for the moment I'll wait and see how it turns out - I didn't have much fun with caffe2 when "train in pytorch and deploy with caffe2" was the storyline FB pushed (e.g. problems with binary size and slow depthwise convolutions) so not too eager to migrate back at the moment.

Our high school's library computer (in the 90s) logged failed log-ins in a file readable by anyone. Just the username, not the attempted passwords, but the return key on that computer was not reliable and a very common error was that the return key didn't register leading to "usernamepassword" being in the log.

Training it _with neural architecture search_. Just training it with a given architecture they cite as 192lbs ...

The tool specifically looks for data races.

It can work for more than a bare bones classifier. It's certainly not painless and sometimes you need some manual work to translate your model but work it does ...

I've lived in countries with representative democracies (Germany, Austria) as well, yes.

I live in Switzerland as well and wouldn't agree that direct democracy works much better than representative democracy. You can certainly argue that it does but it's far from obvious IMO. There's less focus on lobbying and more focus on spreading FUD about other parties' proposals, people have no time to investigate consequences of proposals so they rely on what they read in political ads, see on tv, etc. Men were voting that women shouldn't be allowed to vote as recently as 1990(!) in one part of Switzerland.

I can't say it's worse than other systems, but it's also far from obvious to me that it's better.

It (A4) has tolerances of 2mm on either side though so you can go as close to a √2 ratio as you want :) Not sure about the tolerances on letter paper.

For which use case is "how many character glyphs" the measure you you want to know?

I still need to either use chrome on android or explicitly navigate to the mobile page to get the title text on xkcd.com ... android firefox cuts the title text off with no apparent way to read the full one :/

I don't know if I agree with your first paragraph. The weather model analogy is at least misleading since the corresponding thing in QM would be a mixture of states, not a superposition.

Don't you run into sampling bias this way since not everything is published on crates.io? (I'm not saying that having that tool is a bad thing, just saying that maybe that tool doesn't give you the actual answer ...)