Asdfbla

Wonder how their surveillance infrastructure works. Do they duplicate all data and pipe it into some government datacenter? Or do they have some surveillance boxes installed on-site that just scan the data for certain keywords (like the NSA supposedly often did it)? Maybe the latter seems more likely since filtering the data first makes the overhead more manageable.

That statement doesn't preclude some political activity if you consider it to be something that you can actually "impact and improve", in the author's words. Of course I can only speculate about the author's original intent, but I think I agree insofar that most daily news out there is irrelevant for most individuals. Even if you have a political cause you are interested in, you have limited time and energy, so it might be better to just stay informed in those areas that concern you and ignore the rest of the daily news.

You can't fix the world as an individual, so it's pointless to concern yourself with things you can't impact. Of course that still means that everyone should probably find a little thing where they can try to make a change.

How is bitcoin a good model for humanity? If anything, it strives to do away with the little trust there is left in society and replace it with an antagonistic system where everyone is out to sabotage their peers, only being held back by a (intentionally wasteful) proof-of-work mechanism that enforces compliance with the protocol. That's a dystopic vision for humanity if I've ever seen one. Unnecessary too, since trust-based systems have more or less worked so far.

Not to mention that proof-of-work isn't ecologically sustainable on a larger scale anyway.

Not that I would strive for it myself, but I think at least it was honest that the job posting was upfront about the requirements instead of being silent about it and then peer-pressuring the employee into 70 hour work weeks anyway because everyone there does it.

Who knows, maybe being upfront about it causes the applicants for the job to self-select for those who are confident that they can productively work 70 hours? Maybe the research only applies to normal people like us and the mythical super engineer they are looking for can do it (pure speculation, I don't know the research - I guess even super programmers have biological limits).

Then again, it's probably right to be wary of such job postings so that they don't create a general atmosphere in the industry that such working hours are necessary, whether you want it or not.

Were there no patent problems? I'm glad about it, of course, but it just somewhat surprised me that so many companies can use the same basic principle and there didn't seem to be too much hassle with people suing each other. Then again, I might just not have heard of any high-profile cases.

The basic working principle is probably also basic enough to have lots of prior art I guess.

Another angle of criticism would be that solving the Byzantine generals problem with the threat model that cryptocurrencies target (i.e. anyone can join and leave the network somewhat anonymously at any time) is simply not necessary in the real world unless you are a die hard anarchist. Therefore, for any application that doesn't have those requirements it is a huge waste of resources.

Doesn't seem like Youtube was particularly good at copying established business models in the past (streaming, video on demand, etc.). But you knows, at least for the video-producing crowd (what fraction of Patreon users is that even?) raising money through Youtube donations might be an alternative.

Well, they include an error metric in the paper - basically average vertex distance of predicted face and actual face model (in interocular distance units). They seem to show that they are much better than state of the art with respect to that error, but of course if you want to interpret what that actually means then it's probably easier to interpret the error with your suggestion of actually visualizing it.

It always surprises me that those intelligence agencies obviously think this is worth it, considering they certainly must be aware of all the potential drawbacks that occur when you fuel war and instability like that.

They must operate with totally different views of what constitute "good" and "bad" outcomes of the whole conflict. I guess you can justify all of that by just being cynical enough, but I'd like it if the military (at least in democratic countries) had to make their rationale for heating up conflicts transparent to the civilian world.

Very interesting result, somewhat encouraging to me, because it seems to cautiously indicate that you can actually do something about those echo chambers that may start as jokes but then actively normalize hate and discrimination, possibly pulling formerly neutral users in too. (And keep in mind that those hate echo chambers are not places of honest discussion, since they regularly ban any opposing viewpoints and only strive to keep the hate train going.)

Though the authors clarified that only the hate speech specific to the banned subreddits has disappeared. I think that's still a strong result, but maybe the users might have gone on to spread hate about different topics. (I think that may be unlikely though, since they probably aren't just indiscriminately hateful but focus on a certain topic.)

Isn't this a result you should immediately be skeptical of because the results are so significant? I mean, the idea that you can detect sexual orientation from facial images seems somewhat plausible, but no way I would immediately buy that a classifier can discriminate just from the facial-structure with 81% accuracy as the title may imply, for instance. To be fair, in their abstract the authors mention grooming style and so on, but I feel that the title alone is slightly misleading.

Also: man, that preprint formatting is annoying to read.

Are they still using microarrays? Curious how far those can get you these days. Enough to detect common SNPs, but there's such a wide variety of other important genomic regulators that require whole genome sequencing to detect.

I guess they know what they are doing though, I was just curious if they have extended their technology. Good that the FDA at least prevented those pseudoscientific predictions and they focus on research now.

They can force Facebook etc. to hand over the data anyway, so it does not seem like much would change in terms of the threat posed by governments.

Numpy is such an essential library for any type of scientific computing in Python that ignoring it would be missing the point, if anything. The library infrastructure is part of the appeal of a programming language and Numpy is the default for anything compute-heavy in Python.

Seems hard to judge which of his statements are caused by his condition and which by him just being a vocal and opinionated person in general - but I guess that's a point in his favor because if that distinction is hard to make, it's probably better to err on the side of free speech instead of trying to figure out what's typical and what's atypical behavior which gets a free pass.

And is it a bad thing to let him know it is offensive? If he can't comprehend it naturally, maybe he can at least learn.

I guess you could make a case that people who are diagnosed to lie on the autistic spectrum should get a bit more leeway from the administration with regard to disciplinary action (because it's not really their fault), but what else are you gonna do? Give them a free pass? I don't think that even has much to do with the feared political correctness anymore.

Is the hypothesis that seems important for the article - that civilization was mostly advanced by lone neurodiverse geniuses - actually true? Sounds dubious, but who knows. At least modern science seems to be a very collaborative affair.

Was there ever a poll or study to find out if university faculty as a whole actually is scared to offend people or threatened by some hypothetical PC police? I would imagine that the debate is only relevant for the social sciences anyway, since you shouldn't have opportunities to offend people in more technical subjects anyway.

I know programming within the constraints of mobile devices isn't easy and I want to take nothing away from the release, but when you are still used to Desktop computing it's kind of funny to see how seemingly trivial nice to have features like picture-in-picture or notification dots get all hyped up in the press release.

>Subsequently she fired Victoria Taylor, an incredibly popular moderator at /r/ama.

She didn't. Alexis Ohanian did: https://bits.blogs.nytimes.com/2015/07/13/details-emerge-abo...

She actually did very few of the things the Reddit witchhunt accused her of. The whole Pao affair was just another dark period that showed how insane online communities can really be.

You can defend her if you can resist your urge to praise Hitler and deny the Holocaust yourself in court. Again, that's all transparent and clearly defined in German law, so it literally had nothing to do with the original slippery slope discussion.

I have nothing to say really about your first point where you do nothing but speculate about police motives, but the second one doesn't prove anything either.

The second case is not a slippery slope because it does precisely what's codified in German law, nothing more and nothing less. The lawyer herself denied the holocaust and that's punishable in Germany. So the law was correctly applied. That has absolutely nothing to do with the slippery slope discussion.

Many European countries (like Germany etc.) have operated with free speech restrictions since the end of WW2 and the slippery slope that is always brought up never materialized.

Slippery slope arguments are only valid if you believe that your jurisdiction doesn't have proper rule of law. Otherwise experience, at least in European countries, showed that courts are very well capable of recognizing the importance of free speech even for tasteless and hateful speech.

I wonder what the input for the bot was. Just a screengrab like in other reinforcement learning examples (like Pong or GTA etc.) or did they use some Dota API to make their lives a bit more easy?

Impressive nonetheless, like others said, knowing the reward function would be interesting.

Seems like a pedantic article, but I liked his remark that the 'effectively constant' argument can of course be extended to any algorithm where the input size is bounded and therefore, shouldn't really be used to describe your algorithmic complexities, even in real life implementations (except when you maybe use the inverse Ackermann function or something).

>Musk has been able to take two or three other previously considered impossible tasks and bring them to completion

Such as? SpaceX and their reusable rockets might be the closest thing I guess. Tesla and his solar company use pretty conventional technology, on the other hand. He is willing to invest a lot of money, that's true. But when talking about the Hyperloop, it's hardly just his money that's at stake if government is supposed to be involved.

>What's that network you're using to blast this tripe straight to my eyeballs right now?

A network that evolved from humble origins decades ago. Maybe if Musk financed a hyperloop prototype in the desert for a decade and showed its economic viability, then you can make that comparison. But who knows, maybe that's what he plans to do with the government approval he's supposedly gotten - good for him then, hopefully it works.

>You could have made the same arguments about smartphones 10 years ago

That true, but the lesson from that is that no criticism of speculative sci-fi projects is allowed anymore or what?

Besides, smartphones didn't come from nothing. Prototypes and similar things had been around since the 90s. And they are arguably a much easier problem than infrastructure.

>The primary problem to solve is discrimination against non-engineered people.

You can't be naive enough to think that this problem wouldn't occur. On the contrary, if genetic engineering on such a level was realized (which would require a shift in our current biomedical ethics to begin with), discrimination against the less-optimized would be a foregone conclusion.

In such a scenario, we could only hope that everyone would be optimized equally and the suboptimal humans wouldn't be born into a world where they are second class citizens.

Extreme libertarian views like yours have been rejected almost everywhere in the world. Maybe it's your opinion that allowing scammers to manipulate people is their basic right, but no one seems to share that sentiment.

Musk maybe oversells the state of AI a bit when he says it's potentially an existential threat to mankind - but on the other hand, the advances of the recent years alone are enough to disrupt the daily lives of almost everyone. In that sense, some alarmism might be appropriate, especially considering control of the best machine learning systems will likely be concentrated in the hands of few powerful players.