Asdfbla

On the topic of algorithms helping you discover your own 'taste' (whatever that may be), some time ago when the topic came up on HN someone linked this interesting paper: "Recommender Systems for Self-Actualization" [1], which suggests some methods how filter bubbles could be avoided. I really liked it and its suggestions like having extra recommendation lists for things the user will likely hate, items that are very polarizing in the user base or things no one ever rated before.

[1] https://sci-hub.tw/https://dl.acm.org/citation.cfm?id=295918...

Understandably, since that press release is devoid of any real information. Seems quite reasonable to disregard it.

That's one thing that bothers me about security in general, that it's so super political and it's a bit hard to just be interested in the technical aspects of it without being sucked into political bickering to some degree (can't follow Twitter accounts of most security people for that reason). Of course, I understand that it comes with the field since many adversaries simply have political motives. Still annoying though.

Either we make life on Earth sustainable before that or we won't have enough time to develop the technology to settle on Mars anyway, so thinking of Mars as a backup Earth does not seem like a particularly good idea.

I guess you could argue that the nodes can reject the hard fork. Decentralization doesn't preclude nodes coming to a consensus to change (or not change) the protocol.

Yeah, I also never liked that kind of advice. For me personally just doing something that roughly seems somewhat interesting was usually enough - if you get good at something, the motivation can often follow.

If I had spent more time thinking about what my passion is in life then I'd probably still sit there pondering without an idea of what to do.

I find it pretty hard to curate even interesting lists of researchers on Twitter without running into too much fluff or political tweets. I don't blame them for using the medium as it might be intended and of course they are free to share their personal opinions, but in my experience even just a handful of people who tweet a bit too often can pollute your stream enough to make it annoying to follow along.

Maybe Twitter should just let you filter posts based on content. (Maybe it's already possible, I'm not a big Twitter user.)

I suspect some people are guessing that a multipolar world will be less interventionist (since the great powers might block each other from interfering too much) and therefore adventures like the Iraq war are less likely to happen again. This might arguably make the world more peaceful, though it might allow civil wars to continue for longer.

No idea if it's true or will work out that way, but it's one hypothesis I think is out there.

Sure, the downside of China's rise is that authoritarian states have more legitimacy, but it's not like the western powers did much about those states back when they were the hegemons, so I'm not sure if the argument carries much weight either.

The problem is that reinforcement learning is far from solved and doesn't work all that well yet, so these toy problems are probably what researchers will stick with for some time to come.

There's certainly enough opportunity to work for a more peaceful world instead - scenarios where two large military powers clash should probably better be avoided anyway, so it's probably even a more productive use of one's time than optimizing for yet another variant of mutually assured destruction.

While that may be the reality of things, it surely can't be a justification in any way. If anything, I'd be afraid to revert back to those dark times where technological advancements were driven mainly by military needs.

Didn't record labels complain some time ago that Spotify had supposedly commissioned music for its instrumental playlists (presumably to save money on royalties)?

There's probably a danger that large record labels will threaten to revoke their licensing agreements if they think that Spotify may start competing with them on that front. Not sure if that will be a problem, though, I'm not familiar with the whole streaming market.

Yeah, the figure[1] on the side in the article seems to show something different, because there the response goes over the recursive resolver again. Maybe the text is wrong?

All seems to hinge on the recursive resolver acting as a proxy, so a direct response from ODNS server to the client stub doesn't seem safe.

[1] https://s3.amazonaws.com/ftt-uploads/wp-content/uploads/2018...

While I like the distributed approach technically, it solves none of the social problems that social networks bring.

Filter bubbles, propaganda and competition for users' attention with clickbait and other cheap stories might even become worse, since a centralized platform could at least tweak their recommendations algorithms to break the bubbles (which Facebook doesn't seem to do, but it could).

Basically, I'm pessimistic because a distributed network only solves the (comparatively) easy technical challenge without addressing the problems of social networks.

Thanks a lot, that was what I looking for. Seems most realistic to configure DNS-over-TLS on the OS level then.

>supports encrypted DNS as well as DNS over HTTPS

Are encrypted DNS requests used by default? Does 1.1.1.1 somehow advertise to your client (whether it's a browser, the OS or a router) that encryption is possible? Do I have to configure my endpoint, which may expect to be able to send normal plaintext DNS requests, for it?

I guess DNS over HTTPS will surely not be supported by normal routers, but I don't know what other protocol Cloudflare refers to as "encrypted DNS", so maybe that will work.

Does someone have more information about how they implemented serializable in such a way that, as they claim, performance isn't negatively impacted? Seems pretty hard to achieve that.

As I understood it, the author didn't really claim that working at such a company makes you a god or super hacker, just that it might give you such a reputation among certain circles (not you, but others perhaps) - which many people will derive confidence from. Seems like a reasonable statement.

The article seems to be less about what some objective truth about good jobs is but more about how people perceive themselves, others and their goals in life.

> You can't remove it with a software bandage

Though some people might argue that the biases software reproduces will in turn reinforce biases held by the users at large. Conversely, you may reduce human bias if you algorithmically reduce the bias in the data (which is possible). Not saying whether Google should, that's a very delicate debate, but it's thinkable.

I wouldn't want to be in the place of the person at Google trying to make guidelines for that, but it's definitely doable. For instance, you could probably easily argue that the phrase "three black teenagers" (an example from the article) is pretty neutral in itself and definitely shouldn't be interpreted as a call to produce arrest pictures. In general the term "three [ethnicity] teenagers" could probably be normalized with regards to the setting the pictures show, for instance.

It's a difficult discussion because if you look at the literature regarding algorithmic fairness, there really is no measure of equality that satisfies all notions of fairness. If you force representation of seemingly underrepresented patterns in the data then you implicitly engage in social engineering.

This might be legitimate and reasonable, but you probably should be transparent about the fact that you "unbiased" your data by externally imposing a certain view of what the "fair" data should look like.

Ultimately it's a discussion for social scientists or laywers concerned with discrimination, it's not really in the realm of being fixable by engineering or computer science imho.

I probably expressed that wrong, I was more wondering if the packet sniffing had any beneficial impact on the performance in the sense of QoS or congestion control or something like that. After all, they have to do it for a reason.

But your second points mentions it's for billing and such, so I guess that's my answer.

As someone only superficially familiar with networking, I think it would be interesting to know what impact the current packet inspection practices by the middleboxes have on network performance. After all, the article mentions that those middleboxes want to control the message flow and use that as justification. The assumption behind QUIC seems to be that there's no "global" benefit in that specific kind of network management and that it's mostly selfish interests of local network operators that motivate messing with the packets?

Basically, what are the incentives of the middleboxes to inspect packages and what do they really stand to lose?

I love the concept of differential privacy, but it seems hard to incentivize the "data hoarders" to actually use it, even if you ignore the challenges of building real-world differentially private systems. Google and Apple use it for some things, but in general it doesn't seem like something the market will use by itself.

Also doesn't help that differential privacy itself is maybe too arcane and subtle for the public to talk about and demand, unlike for example encryption which people probably generally at least understand to mean something along the lines of hiding their data in some sense.

Did banks and other network operators who require monitoring their traffic just deem the MITM proxies too expensive or complex, or what was the reason for their protest?

Machine learning models leaking private information is a real problem though, unless mechanisms to ensure something like differential privacy are applied. Those approaches may still be too theoretical and hard to reason about for many machine learning practitioners. Will be interesting how that plays out.

Seems like the publishers could relatively easily figure out whose credentials are being used by searching for presumably uncached files, which will then trigger an access to the file using the credentials Sci-Hub is using, no?

But I suppose it's no use since the authentication credentials to scientific portals aren't that exclusive anyway, so playing whack-a-mole would be pretty ineffective.

This seems to be one of the less surprising false-positives of Youtube's copyright detection algorithm. You match millions of hours of copyrighted material vs. noise, seems like you have a multiple testing problem and at least one thing will inadvertently match somewhere in the 10 hours noise video

Wonder what Youtube can do here, except speed up dispute resolution. With datasets of their size, false-positives seem inevitable.

Do copyright-owners just set the system to auto-monetization or -takedown? If so, maybe Youtube should change their policy to require more manual action from copyright owners.

Fair enough, I wasn't aware of the original meaning since I only came across it multiple times on the web in this drawn, comic-like form.

I never got Zines. Beyond their cutesy art style, they generally don't seem particularly effective at conveying information in a somewhat concise manner. And if you don't want to be concise, why use such a very visual medium then and not just write an article?

Even though I'm also more of a cryptocurrency-skeptic (though mostly in regards to their utility vs. the environmental cost of proof-of-work), one thing I appreciate about Ethereum is that its design decisions and the building blocks of the system seem to be well-documented on their Wiki and have at least some academic rigor to them. This already differentiates them from stuff like IOTA.

How long does a computation step in a quantum computer (which I guess is technically a measurement) actually take? I'm (at a superficial level) aware of Grover's and Shor's algorithms, but only in the sense that they provide asymptotic speedups in the number of steps required for the computation. In classical computers it's easy to envision some model where a step is just some constant number of cycles or whatever, but how long do the measurements take in quantum computers?

I guess it depends on the specific experimental setup of your quantum computer somehow, but I'm just curious about the real-world speeds. A hypothetical computer that does few steps but takes a minute for each measurement wouldn't be so useful.