I remember what felt like 10 years ago we were griping about Google keeping search logs. At the time they didn't have a use for them, they just wanted to keep them around just in case.

So I never log in to google because I want to minimize how much they know about me. Of course they can still see my IP so who knows how effective this is, but at least I can try.

Except, wait! Then they bought youtube, and eventually merged the youtube and google login systems, so not only did you use the same account to log into both, but being logged into one meant you were logged into the other.

Now, youtube I was never that paranoid about the privacy of -- about them tracking what videos I watched. They could, but I didn't care much. So I had an account and I kept it logged in all the time, so I could see what my "subscriptions" had added recently and a few other useful things from being logged in.

But no, because google is so dedicated to privacy, if I wanted to stay logged in to youtube, I had to let google log every search I did under my own username. So to preserve my minimal privacy with google search, I had to stop logging in to youtube.

Those are the only two google services I use, but I bet there are similar stories for the rest of them.

So yeah. Google marketing a service based on "privacy"? No thanks.

This poll always totally broken, because the curators seemed to appear to totally fail to understand video games.

The poll was absurdly ambiguous about whether it was "the visual art found in videogames" or "videogames themselves as Art". You could see this ambiguity throughout the poll, from the language used talking casually about "art of games" and the choice of credits they provided for each games (typically artists and designers, almost as if they didn't understand what "designer" meant for games).

Now the results starts off with "explore the forty-year evolution of video games as an artistic medium" which is unambiguous (but makes the credits choice inexplicable), but the poll had nothing like that. So who knows what the voters took it to be.

"Secondly if you are going to invent an AI technique then picking emotive words for your jargon is a good way to ensure publicity."

Dear submitter: you are the one writing the submission summary which (a) goes on and on about the jargon term, and (b) gives them publicity. Wtf.

(Yes, I know that is what the article is about, so it is an accurate summary. It's still absurdly un-self-aware to then submit that to slashdot.)

I'd really like to do some BeagleBoard development, but it's clear from everything I read that the out-of-the-box experience is nothing like this--there's a litany of extra stuff you have to do to get it working, all of which puts me off even thinking about trying it.

Jeez, read the fucking article: "In Stroustrop's mind, making C++ compatible with C was instrumental, crucial to its success. I don't disagree."

It was instrumental to its success in the marketplace of languages. It doesn't help it succeed at being a good language. That's the whole point of the article, Mister point-misser.

"By making it compatible, he got many C-language adherents to try it, and he got it on every platform so they could do so."

Yes, I know this. That is exactly the point. I was there for the whole thing too. (And to this day I still program in C.) The essay itself is like ten years old at this point.

C++ is a language that, given the choice between a feature/goal that made a good language or a feature/goal that made a popular language, made the choice in favor of popular.

Lots of things, e.g. compatibility with C, or choosing not to rewrite the linker, or sticking with header files, can be seen through this lens. I think those choices were made with an eye towards what would make the language "succeed in the marketplace", not what would make for the best programs or the best programming experience.

Obviously some of the times those things are correlated, but they're not always (e.g. the linker situation).

Now, you might think that being popular is a good thing. But the upshot of this is that in the marketplace of computer languages, computer languages that are designed to be popular will beat out computer languages that are actually good. (C.f. 'Worse is Better', etc.)

I don't think that's a good thing.

If they have 40 customers and they grow by 60 customers, they'll have grown by 150%.

To grow by 60%, they need to grow by 0.6*40 customers. That would be the same as 0.4*60 customers; in other words, they need 40% of the 60 customers remaining, not 100% of the 60 customers remaining.

In other words, to grow by 60% they need only 40% of the market they're talking about. That's why the grandparent was critizing their math.

However, comparing the prominent S-curves in the foreground reveals a significant difference in perspective/foreshortening that makes it clear that the color photo is taken from a higher elevation. The distant shapes seem to match pretty well so I don't think it's an aspect-ratio fuck-up, although that would be all too common in this modern world where nobody seems able to notice that effect either.

A mathematician poses a problem about a father with two children, mentioning that one child is known to be of a certain age and born on a certain day, and asks what the probability is that the other child is the same gender.

The answer to this question (assuming 50:50 birth rates) is 50%.

This is because I'm accumulating probability over sets of questions & fathers & children without omitting any cases, much as the non-50% answer accumulates probability over sets of children and omits some cases.

Long version:

If you try to understand probability without trying to count over large sets of examples (i.e. statistics), you will almost surely make mistakes.

Let's step back to just the Two Children Problem and, I'm going to try to point out what I hope is a slightly deeper intuition about the two answers.

The naive answer of 1/2 just assumes the child gender is independent. Woohoo, rock on.

The "clever" answer imagines that we have thousands of fathers with two children. Some of those fathers are the fathers of two girls. The rest are the fathers of at least one boy. Only the latter fathers can pose this problem; therefore, if we randomly select a father from that set of fathers who can pose this problem, we have only a 1/3rd chance of drawing a father who is the father of two boys.

But we're linking the fathers to the problem in an overly specific way. Suppose we draw a father randomly from the set of all fathers with two children. That father may have two boys, two girls, or one of each. If we draw a father with at least one boy, we can proffer him to the reader and ask what the odds are his other child is a boy. If we draw a father with at least one girl, we can proffer him to the reader and ask what the odds are his other child is a girl.

So, if you actually had to do this--pick one person, ask the question, and you're done--what would you do? Well, it's dumb to draw a random father and then go "oh crap, we can't actually ask a question, this whole thing was a waste of time". So you'd probably ask the "at least one boy" question if they had two boys, the "at least one girl" question if they had two girls, or... one or the other question in the other case. Maybe you always ask the boys question. Maybe you randomly pick. If your tie-breaker is to just always ask "at least one boy" if it's possible, then the answer to "at least one boy" is 1/3 chance the other is a boy, but the answer to "at least one girl" is a 100% chance the other is a girl!

If you choose "fairly", then when you draw a father with both a boy and a girl, then half the time you ask "at least one boy" and half the time you ask "at least one girl".

If you choose that way, and happen to draw a father with the "at least one boy" question, then the probability the other child of his is a boy is 50%. (That is, given 100 fathers, the three cases are 2 girls: 25; 2 boys 25; 1 each 50; you split the 1 each case between the 2 questions, so you have 50 cases where you ask "at least one girl"--25 of those are they have 2 girls, 25 of those they have one of each; and 50 cases where you ask "at least one boy" with the same odds.)

Thus a complete analysis of all the cases under a more plausible situation for where you might be asking this question returns to the naive solution: the other child will be a boy 50% of the time.

So, as other people have said it depends on context. If the context is that it's a mathematician posing the question once about a hypothetical non-real situation, and the mathematician is sexist or otherwise favors mentioning boys, then maybe it's 1/3rd. But if he's not--if he chose the question randomly based on the gender choices available to him--then it's 50%.

Moreover, even if you choose a sexist bias for how you split the middle cases, if you measure over asking ALL questions, both "at least one boy" and "at least one girl", no matter how you split up the cases, the average odds that the other child is the same gender as the listed gender is 50%.

The same analysis applies to the tuesday problem; if you pick a random two-child-parent and ask some similar question--rather than simply failing to ask a question if the desired question doesn't apply--the average over all such answers will be 50%.