11,865 karma · joined March 13, 2008
Research: https://www.cs.princeton.edu/~arvindn/
ETA: However there is one thing you say that is easily refuted -- having a DNT header without legislation to back it up is useless, because nobody has the slightest incentive to respect it.
"Opt-in" would mean that by default -- with no action on the user's part -- tracking would be prohibited. Clearly that is infeasible (for precisely the reasons you describe; it would kill the industry.)
While the proposal is about opting out of "third party web tracking", it is not obvious what a third party is. Is a YUI, Google Analytics, or Facebook Connect plugin which I chose to integrate into my site "third party code"? What if I have a business account with something like Crazyegg.com -- can I put their JS heatmap tracking in my webpage to figure out what buttons my users are getting confused by?
Yes, these are exactly the questions that me and my colleagues are working out. The goal is to avoid breaking the web as much as possible. For example, one way in which an exception to Do Not Track is triggered is if the user logs in to the 3rd party provider, so FB Connect will work smoothly regardless of DNT.
If this is complex for hackers, do you trust the government that thinks the internet is a series of tubes to get this right?
Great point, which is why things are structured the way they are -- the idea is for Congress to delegate authority to the FTC, which is a highly tech savvy organization (they just tapped Ed Felten, for example). They're also great at getting input from the technical community (I was myself on an FTC privacy panel earlier this year.) In short, hackers are quite involved with the process.
I'd be happy to answer any questions.
Cellular networks are centrally administered, enabling service providers and their governments to conduct system-wide monitoring and censorship of mobile communication. This paper presents HUMANETS, a fully decentralized, smartphone-to-smartphone (and hence human-to-human) message passing scheme that permits unmonitored message communication even when all cellular traffic is inspected.
HUMANET message routing protocols exploit human mobility patterns to significantly increase communication efficiency while limiting the exposure of messages to mobile service providers. Initial results from trace-driven simulation show that 85% of messages reach their intended destinations while using orders of magnitude less network capacity than naïve epidemic flooding techniques.
http://www.usenix.org/event/hotsec10/tech/full_papers/Aviv.p...
In the 1940’s, the paleontologist von Koenigswald was searching for early human remains on Java and decided to enlist the help of the locals in his search by offering them “ten cents for every piece of hominid bone they could come up with.” Unfortunately for von Koenigswald (and for his findings), he discovered too late that the locals “had been enthusiastically smashing large pieces into small ones to maximize their income.”
From http://freakonomics.blogs.nytimes.com/2009/10/20/when-youre-...
(The story is from the book A Short History of Nearly Everything, which is one of the best books I've ever read. The amazing thing about the book is that it is in fact a short history of nearly everything.)
Combine the difficulty of getting incentives right with the inherent problems of Government and a dangerous mix results. For example, every time a subsidy is created, a special-interest group sprouts up dedicated to preserving the subsidy in perpetuity, long after it has outlived its utility.
It's typically (rot13'd) bire n uhaqerq gubhfnaq.
(Surprised no one did this yet.)
I've been tracking the security/privacy problems with Instant Personalization for a while; my recent post might be relevant: http://33bits.org/2010/09/28/instant-personalization-privacy...
I'm also curious to see how things will turn out when a whole bunch of YC startups get Instant Personalization access, as YCRFS7 promises.
I'm thinking about topics that would make good study, reference, or entertainment material, like "weight training," "atomic physics," "Bay Area travel," "Paradoxes" (http://en.wikipedia.org/wiki/List_of_paradoxes) or "50 most interesting Wikipedia articles" (http://copybot.wordpress.com/2009/04/07/the-50-most-interest...). Needless to say, there are thousands of potential good topics.
I would happily pay a dollar or two to read these on the Kindle/iPad etc. As far as I can tell this is neither against the letter nor the spirit of the GFDL or CC-BY-SA. There are a few value adds here compared to reading Wikipedia directly: 0. the obvious one of figuring out which articles to include 1. ebook reader-specific formatting 2. for study topics, creating a more-or-less linear flow out of a web of articles on a topic 3. quality control -- checking for vandalism, etc.
I really hope someone will create the infrastructure for this: i.e., a web interface much like the one under discussion, but which also formats the books in AZW, ePub, and whatever other formats and lets you automatically push them to all the self-publishing stores. (I envision a revenue sharing arrangement with users who create books.) I will even offer to put together a dozen books for free to sweeten the deal :-)
Your point that moderation is the larger goal and reputation/identity is only a tool is certainly valid. I guess we disagree about how effective a tool it is. For a visceral, deeply depressing account of how different the same person's behavior can be depending on whether or not they're anonymous, check out the story of the harassment of two female Yale Law students on AutoAdmit: http://www.portfolio.com/news-markets/national-news/portfoli...
"This is a message to YCombinator companies and hopefuls: if you tell Facebook about your startup before you reach critical mass and you get found by Facebook in any way you're an idiot. They will steal your company's ideas and try to get you to take a small price for your company and vision and take a job at Facebook. Which by the way is going to be a job that sucks. Don't do it. It's a scam. It's a trap. It's a trap. It's a trap. Insert Star Wars quip here. It's a trap. IT'S. A. TRAP. End of message."
Semi-supervised learning is a good idea in this type of situation, given that unlabeled samples are far more abundant than labeled samples, but there are gotchas to watch out for. In general, SSL helps when your model of the data is correct and hurts when it is not.
Here's an example of what can go wrong in this particular application: let's say the word 'better' is mildly positive, but when it appears in high-confidence samples, it's usually because it appears together with the words 'business' and 'bureau', as in "I just reported Company X to the Better Business Bureau", i.e., strongly negative. This means that the new self-training samples containing the word better will all be negative, which will bias the corpus until eventually 'better' is treated as a strongly negative feature.
Occasional random human spot-checks of the high-confidence classifications would be useful :-) Also, self-training gives diminishing returns in accuracy, whereas the possibility for craziness remains, so turning it off after a while might be best.
A survey of semi-supervised learning: http://www.cs.wisc.edu/~jerryzhu/pub/ssl_survey.pdf
1. The motive here is fairly straightforward -- they are trying to silence the researcher. (Edit. There is of course also the fact that they want to find the identity of the source; I don't know which of these is the primary motivation.)
2. 10 years ago or probably even 5 years ago this would have absolutely worked. The Indian government just hasn't woken up to the fact that news travels fast these days due to the Internet, and they can't control it. Indeed, the Indian media aren't even covering this. That's probably what they were counting on.
3. It is also clear to me that it is not malice. Their point of view is that if these pesky security researchers didn't go around poking flaws, then no one will find them and exploit them. Regrettably, that is a surprisingly common view. For example, many commenters right here on hacker news criticized my research on those grounds (http://news.ycombinator.com/item?id=1193417)
4. Yes, most people are going to go on with their lives. What else would you expect -- there are dozens of these minor atrocities happening around the world every day, there simply isn't enough time to do something about all of them. A good many people in India and elsewhere are doing something about this, and I believe it is making a difference. Police brutality (let alone random arrests) used to be common in India; things have gotten dramatically better in the last decade due to activism.
You often throw out these terms from philosophy/logic; that doesn't make your arguments more logical. Just an observation.
Summary: "First, it completely contradicts historical legal trajectories where name changes have become increasingly more difficult. Second, it fails to account for the tensions between positive and negative reputation. Third, it would be so exceedingly ineffective as to be just outright absurd."
She provides strong evidence for all of these.
"Mainstream culture" collapsed about a decade ago -- did you know that Seinfeld was the last TV show watched regularly by 50% or more of the American population? As the article notes, memes are increasingly where we get our culture. And hoaxes are an integral part of the meme landscape.
Anyone who can find a way to modify the Internet (not the Internet itself, of course, but people's use of the Internet) to better resist the spread of hoaxes could significantly change society. Think not just hoaxes, but outrages like #amazonfail. What a huge social cost (on attention, reputation etc) these things impose. http://www.shirky.com/weblog/2009/04/the-failure-of-amazonfa...
Now that might not seem like a great startup idea -- users are just having fun spreading links around (or getting outraged), and if you can't even monetize content then how can you monetize crowdsourcing the truth?
But then I noticed that Fox News and other media outlets fell for this hoax. Surely there's a significant monetary cost to that, which provides an opportunity for a business model?
The old Snopes way of doing things isn't working so well any more -- these things spread in minutes, so the fact checking mechanism needs to use channels similar to the ones that the hoaxes themselves do.
One possible idea is a prediction market (that pays out real money). If it is successful, it might become a standard part of media fact checking policy to check the prices there. A news organization probably be willing to actually wager a few thousand dollars -- a fraction of what the story is worth to them. Coming back to this example, that would give a tremendous incentive to one of the many people involved in creating this hoax to make a decent pile of money through the prediction site (anonymously, in fact). In effect, it would be a way for anonymous tipsters to leak out the word (in addition to aggregating third-party analysis and intelligence), except it would be far more streamlined and effective because it's a market.
Anyway, just a thought.
But then Aaronson responded by saying "If I were a Bayesian rationalist, I’m sure I’d agree with you!" and instead claims that the rationale for the $200k is that "If P≠NP has indeed been proved, my life will change so dramatically that having to pay $200,000 will be the least of it." and that "If P≠NP is proved, then to whatever extent theoretical computer science continues to exist at all, it will have a very different character."
That almost sounds like the justification for a reversed insurance bet, except of course that if the proof if wrong then Aaronson doesn't stand to gain anything. So yeah, confusing.
Bottom line, I would caution against interpreting this to mean Aaronson is betting against the proof.
Now you changed your question to whether they can solve each position optimally or not. Fine.
Why don't you re-read the article once again? They state that they can solve random positions optimally at the rate of 0.36/sec, in the same table where they say 3,900/sec for solving it in 20 moves or less.
Search techniques that can quickly solve any cube with a small (near-optimal) number of moves have been known for a while, mainly due to the work of Kociemba: http://www.jaapsch.net/puzzles/compcube.htm#kocal The techniques used are standard AI tree/graph search algorithms with lots of Rubik's cube-specific optimizations.
A few years ago, these methods became good enough to solve almost any cube quickly within 20 moves (which was conjectured to be God's number.) So the algorithm as well as a fast implementation already existed. Here's Kociemba's page http://kociemba.org/cube.htm and here's an iPhone app with a neat twist: you can photograph your physical cube to solve it http://www.wired.com/epicenter/2009/01/iphone-app-solv/
What these guys did was to make further optimizations and run it on a cluster to search through all possible position sets. As they say, they can solve about 4000 positions/s (in 20 moves or less) on a single machine.
* As far as I know this paper wasn't circulated for informal peer review before being made public; I heard no talk on the grapevine. (Edit: apparently it was circulated and someone other than the author made it public.)
* Therefore a proper assessment is going to take a while. Until then we can only speculate :-)
* While the crank attempts at P =? NP are statistically much more common (http://news.ycombinator.com/item?id=347295), this isn't one of them. The author is a legit computer scientist: http://www.hpl.hp.com/personal/Vinay_Deolalikar/
* On the other hand he hasn't published much on complexity theory and isn't known in that community. Which is weird but not necessarily a red flag.
* Looking at his papers, it's possible he's been working on this for about 5+ years -- he has two threads of research, basic and industrial, and the former line of publications dried up around 2004.
* On the other hand I don't think anyone knew he was working on this. The only known serious effort was by Ketan Mulmuley at U Chicago.
* It has been known that the straightforward combinatorial approaches to P =? NP aren't going to work, and therefore something out of left field was required (http://web.cs.wpi.edu/~gsarkozy/3133/p78-fortnow.pdf). Mulmuley's plan of attack involved algebraic geometry.
* This paper uses statistical physics. This approach doesn't seem to have been talked about much in the community; I found only one blog comment http://rjlipton.wordpress.com/2009/04/27/how-to-solve-pnp/#c... which mentions the survey propagation algorithm. (Deolalikar's paper also talks about it tangentially.)
* If the statistical physics method used here is powerful enough to resolve P != NP, then there's a good chance it is powerful enough to have led to many smaller results before the author was able to nail the big one. It's a little weird we haven't heard anything about that earlier.
* Finally, since the author is using physics-based methods, there's the possibility that he is using something that's a "theorem" in physics even though it is technically only a conjecture and hasn't actually been proven. Physicists are notorious for brushing technicalities under the rug. It would be very unlikely that the author didn't realize that, but still worth mentioning.
* If that is indeed what happened here, but the rest of the proof holds up, then we would be left with a reduction from P != NP to a physics conjecture, which could be very interesting but not ground breaking.
Conclusion: overall, it certainly looks superficially legit. But in non peer reviewed solutions of open problems there's always a high chance that there's a bug, which might or might not be fixable. Even Andrew Wiles's first attempt at FLT had one. So I wouldn't get too excited yet.
There are cultures in the world that are completely unaffected by this illusion - http://goldmark.org/jeff/papers/ridley/html/img1.gif
If something as basic as visual perception is so hugely affected by culture, the paper asks, is there any aspect of psychology that is not? How, then, can we trust the conclusions of studies whose subjects are all drawn from the same, highly unusual cultural group?
I'm glad I stuck with it until I got to this point. Now I have to read the rest of it.
You might claim that the hardness guarantees of PRNGs are not enough for you -- they are based on some assumptions after all -- but that is a specious objection, because those are the same assumptions that underlie all of digital security and e-commerce.
Perhaps you're using it in a metaphorical sense. But I've never seen anyone else use it that way.
There have been at least half a dozen articles on Hacker News recently with exaggerated or incorrect claims about Facebook and privacy. I wrote about this "mob behavior" here: http://33bits.org/2010/05/10/facebook-privacy-public-opinion...
It is interesting that the discontent about Facebook has reached such levels that anything at all can kick up an outrage. It is also depressing that this is the top story on Hacker News. It is what I would have expected from Digg or Reddit.
I voted it up anyway, that's how much I like this article. After reading this comment thread http://news.ycombinator.com/item?id=481557 on that page, I decided to read Moneyball, and it was one of the most riveting books I've read.
Thanks, HN.