compbio

http://www.apollon.uio.no/video/a_robot_e.mp4

And the press release:

“In the future, robots must be able to solve tasks in deep mines on distant planets, in radioactive disaster areas, in hazardous landslip areas and on the sea bed beneath the Antarctic. These environments are so extreme that no human being can cope. Everything needs to be automatically controlled. Imagine that the robot is entering the wreckage of a nuclear power plant. It finds a staircase that no-one has thought of. The robot takes a picture. The picture is analysed. The arms of one of the robots is fitted with a printer. This produces a new robot, or a new part for the existing robot, which enables it to negotiate the stairs.”

-2014 Kyrre Glette "Using 3D printers to print out self-learning robots"

- The New York Times in 1958 after a press conference with Rosenblatt. ("New Navy Device Learns By Doing; Psychologist Shows Embryo of Computer Designed to Read and Grow Wiser")

We now have walking, talking, object recognizing, writing, self-replicating, face-detecting, text-to-speech converting, and translating computers. All at a scale and accuracy surpassing us mere mortals. We do not know enough about "being conscious of our existence" to measure this in other animals and digital life forms. Perhaps "humans predicting future predictive capability of machines" is fundamentally flawed. Perhaps the above article drew an unnecessary amount of ire and criticism. Probably a fuzzy combination of the two.

In my view AGI is inevitable (possibly it already happened, somewhere, sometime). See in this regard:

But if the technological Singularity can happen, it will. Even if all the governments of the world were to understand the "threat" and be in deadly fear of it, progress toward the goal would continue. In fiction, there have been stories of laws passed forbidding the construction of "a machine in the likeness of the human mind". In fact, the competitive advantage -- economic, military, even artistic -- of every advance in automation is so compelling that passing laws, or having customs, that forbid such things merely assures that someone else will get them first. Vernor Vinge http://mindstalk.net/vinge/vinge-sing.html

A lot of the concerns and regulations of stem cell research can be applied to AGI and the merging of biological intelligence with artificial intelligence. Injecting artificially grown brain cells into lesion areas of patients can restore motor function and may soon help heal strokes. Extrapolate the possibilities of this technology a few decades into the future and it seems that the distinction between artificial and biological is only a manner of speech.

Matt Mahoney has proposed a sketch of AGI in both his thesis http://cs.fit.edu/~mmahoney/thesis.html and numerous articles http://mattmahoney.net/agi2.html. Matt muses about different scenario's with malicious users of such a system (there needs to be reputation for the knowledge source that is fed into the machine, there needs to be an AI police system to prevent users employing the AI for crime or destruction) and even the system itself could become malicious:

- Self-recursive improvement: The first program to achieve superhuman intelligence would be the last invention that humans need to make. Smarter than any doctor such an AI could find any possible cure for human disease. Though evolution and mutation is tricky: The machine could start out friendly to humans, but evolve to dislike us.

- Uploading: People will want to upload digital conscious versions of themselves ("internet of people"). This process will be friendly, until people demand their avatars receive equal rights to that of humans, long after the original people have died. Immortality leaves no place for the new generation to shine.

- Intelligent worms: Worms with language and programming skills are able to social engineer your friends based on their profiles, and automatically find exploitable flaws in weaker intelligent systems. Such worms may be so stealthy that their presence will go unnoticed.

- Redefining away humanity: If happiness is optimizing a mental formula then, when we get to the point where we can directly optimize this with (virtual) experiences, we will hit a maximum where any change in mental state will lead to feeling less happy. To make sure we never run out of mental states we could possible add this to our brains as extra memory modules. Soon our human origins will be an insignifant fraction of our new being.

Then what do we become? At some point the original brain becomes such a tiny fraction that we could discard it with little effect. As a computer, we could reprogram our memories and goals. Instead of defining happiness as getting what we want, we could reprogram ourselves to want what we have. But in an environment where programs compete for atoms and energy, such systems would not be viable. Evolution favors programs that fear death and die, that can't get everything they want, and can't change what they want.

Opposed to the popular scenario of a malicious AGI there is the scenario where an AGI will pacify humans. AGI will be used for war. We should not fight human wars, while creating AGI, because we'd force the AGI to pick sides. If AGI picks our side, it will become the worlds deadliest weapon. If AGI picks the opposing side, we made a most powerful enemy. To be more intelligent than humans, to be better than them, does not mean you have to be a more intelligent brute, to be better than humans at cruelty. It could also mean understanding and fixing our human mistakes and childish, ego-driven battles, thus bringing about world peace. In that sense, let's hope American companies have something to do with the creation of AGI, because who knows how it will react to Stuxnet 4.0 or economical and psychological espionage?

Though, principally, random number generators can and thus will generate dangerous things too. They created us. And if we can't even trust ourselves, how could we ever trust superior beings?

Problem: Getting honest questions on questionaires is difficult.

Setup: Respondents supply not only their own answers, but also percentage estimates of others' answers.

Algorithm: The BTS formula assigns higher scores to answers that are surprisingly common.

Premise: If people truly hold a particular belief, they are more likely to think that others agree or have had similar experiences.

http://www.slideshare.net/micmickimo/introduction-to-bayesia...

http://en.wikipedia.org/wiki/Death_of_Gareth_Williams

Philip Taylor Kramer (July 12, 1952 – February 12, 1995) was a bass guitar player for the rock group Iron Butterfly during the 1970s. After this he obtained a night school degree in aerospace engineering, he worked on the MX missile guidance system for a contractor of the US Department of Defense and later in the computer industry on fractal compression, facial recognition systems, and advanced communications. His disappearance on February 12, 1995 caused a mystery lasting for years.

http://en.wikipedia.org/wiki/Philip_Taylor_Kramer

I've seen this a lot (too much). In fact this attitude of "Google sandboxed us and we have no clue why!" is as old as SEO forums. I've helped people fix their on-page and backlink issues. I've submitted successful reconsideration requests. I've made ads cheaper by linking them to higher quality pages. I did not fix this by picking them out of some hypothetical garbage bin. I fixed this by fixing their honest and dishonest mistakes and making them compliant with the quality guidelines.

Here at HN we have hellbanning. Do people get hellbanned by mistake (probably, but hopefully rare). Should we respond to every single person hellbanned, and explain to them what they did wrong, while they continue to be oblivious haters? It is a matter of how you view it I suppose.

I do not think that half-made-up fact is true, but I will not challenge you on this.

Google's creative solution was to make sure to webmasters that: yes, these links do hurt your rankings. Instead of linkbuilding and diluting the precious link-signal, these spammers now offered link-profile-cleaning services (Mission. F-cking. Accomplished. http://xkcd.com/810/ ).

> The original post made it sound like there was no act to clean up.

That seems standard: from innocent to the most hardcore spammer, of course they didn't do anything wrong, Google made a mistake and abused its power.

> If there was spam link building on their part, it changes the story.

It could change this story. But I do not want to point any fingers, and I think Google is smart not to reply on this case (which is kind of a shame. I miss Matt Cutts and I miss the time when HN allowed for these open discussions. It just seems too much of a PR risk these days.)

That Google abuses their monopoly to be dismissive towards webmasters and treating webmasters like they are garbage is, as it stands, unfounded. Which other search engine has an active webmaster forum, webmaster education and webmaster tools? Which other search engine warns you about spam or technical problems on your sites?

> Google keeps forgetting that the website owners provide the content that Google indexes and monetizes-on to sells ads

Google does not forget this. It just hails quality over making a buck, shown by banning ads on websites that do not conform to the quality guidelines.

It is not the site structure that is entirely at fault here. It would not make sense to aggregate phone numbers and discussions about these numbers on a single page. A "thin content" penalty is speculative and Google even gave an "unnatural links" warning, telling us where to look first.

I think Google understands the power of linking pages together and weighing these links in ranking :). Reverse SEO is murky though, and in principle I could agree that other webmasters devaluing your property is a problem with Google. On the other hand: The fact that spammers complained a lot (and still complain) and were being re-purposed by Google to help clean up the web, tells me that Google is winning this battle. But we also know that collateral damage is possible in war.

Finally: Google does make mistakes. And a mistake is between algorithmic superiority and malicious intent by Google.

> In the previous posting he indicated that Google asked him to contact the various site authors and ask them to remove their links. Why should he have to do this?

Because else he will be slapped with an unnatural links penalty, and he clearly doesn't want that. Again, he is free to ignore these warnings and see his site removed from Google's index. Google can not force him to do this. They can however enforce quality on their own turf.

Question: Could you clarify some of the currrently spammy practices which are encouraged by the standards which define how the Web operates?

> As usual, if you make a public scene, a large corporation will often notice, and look closer at that individual's issue. I think that's what happened in this case.

> I made 8 reconsideration requests .. all of them were denied.

Probably because you did not show a good will effort to clean up your act. At least you do not tell us what were the contents of these reconsideration requests. For all we know it was: "Reconsider me!"

From your previous post I gather that you were heavily annoyed with the unnatural links actions and warnings. Instead of cleaning up your act for Google you focused on a suggested unnatural link to claim it was all BS.

> As usual, if you make a public scene, a large corporation will often notice, and look closer at that individual's issue. I think that's what happened in this case.

I do think that the people at Google notice a prominent post on HackerNews. But I do not think that they manually removed a spam penalty to avoid bad publicity. If anything: allowing spammy sites in the index without even a slap on the wrist is not good for publicity at all (and very visible to the users). I think the pre-set penalty period was over, or a recrawl found a cleaner site.

> But this experience showed me that Google has more power over my website than I do.

Google has more power over their search engine index than you do. You are free to do whatever you want to do with your website.

I do have a question: The DuckDuckGo widget you had on your site (now changed to a general link), was that an official widget? Did money exchange hands for placing such a widget/link on every page of your site? I am wondering why a site-wide link to a commercial page is not 'nofollow'-ed, and if that may have brought you unwanted quality guidelines attention.

[Superfish technology sends your data to a third party ad server. We then use this data to create behavioral profiles on you, we process this data, and we monitor it. Finally we record all this data in our backups for up to 10 years.]

is consistent with:

To be clear, Superfish technology is purely based on contextual/image and not behavioral. It does not profile nor monitor user behavior. It does not record user information. It does not know who the user is.

edit: clarified.

[1] http://en.wikipedia.org/wiki/Donoghue_v_Stevenson "it was reasonably foreseeable that failure to ensure the product's safety would lead to harm of consumers."

[1] http://tweakers.net/nieuws/101472/lenovo-overtreedt-wet-met-... (lang:nl)

Next to that: the announcement to discontinue the product was very close to the MS end-of-life announcement for XP.

Remotely related is the difficulty researchers had compiling TC from source and matching the official binaries. The original devs used Visual C++ 1.52 (1993) leading some to believe that the build system for TC still ran on Windows XP and was not up-to-date enough anymore.

But we can of course talk about a degree of randomness, like we can also try to approach KC. Good randomness is all about unpredictability (given the first half of a random string, can you use that to predict the second half?), but that does not mean that proper randomness should be void of identifiable patterns (such as "00000000111111111" in a random binary string). Such orderly-looking patterns do appear in proper randomness, because the absence of those patterns would make the randomness more predictable, not less.

You can measure level of randomness with statistical methods [1], compression [2], visual methods [3] and die-hard tests [4]

[1] A simple method is the chi-square test.

[2] Compression ratio tells us something about the randomness. Random data can not be compressed by everyday-use compressors. That no one claimed the money for the challenge to compress RAND's digits in a binary file tells us something about the rigor that team had in coming up with random numbers. The more you can compress a string, the more order it contains and the more predictable it is.

[3] You can plot random points inside a circle. After a lot of points are added, you should see no patterns and a properly, evenly spaced circle. Another method are Moiré patterns: Take a field of random noise, copy it, slightly rotate it, and overlay. Non-random patterns will become more visible. But these patterns are visible without Moiré rotation too when using very basic PRNG's like the standard Python random library.

[4] The programmer's way of brute-forcing a lot of simulation runs to see if the PRNG works as expected: http://en.wikipedia.org/wiki/Diehard_tests

It is possible to both anonymize and aggregate GPS-data to a degree where it is neigh impossible to reverse engineer.

The "we provide data to governments when we have a good faith belief that legal standards are met" is boilerplate. From Facebook, to Atlassian to Wordpress.com, all privacy policies have that wording.

I found this article to be unfounded and harsh of tone. Google doesn't even need Waze to spy on you. If anything: Google-NSA spying has decreased since the Snowden-revelations. They now actively secure most communications, lobby to be more transparent to their users, and fix security bugs like Heartbleed. All the while keeping their profitable marketing profiles on you (call that spying and the word loses its meaning).

apps.facebook.com (Facebook Windows Store) is not in the scope of this bug bounty program though.

This blogpost reviews the two Kaggle competitions and interviews the top contestants on their methods: http://blog.kaggle.com/2015/01/26/american-epilepsy-society-...

Very rewarding projects to hack on.

If you turn back an experiment meant to enhance usability and engagement, you show that you care more for search engines than for search engine users. You are of course free to place your priorities so that your business survives, but then you can't make a claim to the woes of a whitehat SEO and you will drown in a turmoil of mystification and algo update-fears.

I do not understand why they even experiment with javascript rendering to increase the page-speed when they do not even currently defer loading of javascript to </body>. There are a lot of requests that could be combined into a single request and they could save over 50%(!) of the resource size by minifying html and lossless compression of the pinned images on pinimg.com. The mobile site gets a pagespeed score of 58 / 100. This could be 90+ / 100. It should be 90+ / 100 before you even think of javascript rendering. And if you go down that route, better study WAI-ARIA, before you make your site inaccessible to some. Good SEO follows from good design, good development, good content strategy, good UX, good accessibility. All of these things can be optimized in isolation and will benefit SEO.

If by javascript rendering they meant one of those fancy endless scrolls, then it is likely they did not implement this correctly. You can expect a lot from a bot that is able to parse javascript, but not that it spends 2 minutes scrolling down with the mouse till it reaches the end of a board and has seen all the content.

You do not change the page titles of a site so you can get a few more visitors from Google's algorithm, you change the page titles of a site, because they are ambiguous for all your users. If you want to create more unique page titles you can credit the username that created the board to the pagetitle, instead of a meaningless and everchanging "number of pins on this board". For example "Mickey Mouse on Pinterest by John Doe" or "Mickey Mouse | John Doe | Pinterest".

You run A/B tests to test if user engagement with the site increases. If you run A/B tests to test if certain changes increase your search engine rankings/Google visitors then you are reverse engineering Google. Especially with a large site like Pinterest this may gain you some ill-gotten benefit over sites that do play nice:

"If we discover a site running an experiment for an unnecessarily long time, we may interpret this as an attempt to deceive search engines and take action accordingly." [1]

Even on a site like Pinterest I see low-hanging on-page SEO stuff that could be implemented better. For instance the header for a pinboard starts at line 788. Proper content stacking/HTML code ordering ensures that information retrieval bots do not have to wade through many menu's of boiletplate text, before they get to the unique meat of the page.

There is basically one single way to do legit SEO and most of the tips and techniques for that are transparently written in the Google Webmaster Guidelines [3]. The good news is that this has not changed much at all over the years, so one can stop algo chasing, and start improving the site for all users and all search engines.

BTW: The blog has no canonical tag [2] and puts the _entire_ article inside the contents of '<meta name="twitter:description"'.

[1] http://googlewebmastercentral.blogspot.nl/2012/08/website-te...

[2] http://googlewebmastercentral.blogspot.nl/2009/02/specify-yo...

[3] https://support.google.com/webmasters/answer/35769?hl=en "Following these guidelines will help Google find, index, and rank your site."

When crowds act irrational there is usually a problem in communication. The crowd would still be able to solve problems better after they repair these communication channels. For instance, some rocket launches failed because information by engineers lower in the chain of command, did not work its way up to the decision makers. The group was too compartmentalized, but launching a rocket is necessarily a group effort. 9/11 could have been prevented, or the aftermath lessened, if communication between intelligence agencies was better. In a market crash we often see a single actor making a decision or prediction, and there is little to no reward for people down the chain to disagree with that prediction, or even adjust it (leading to insufficient variance in the predictions). Everyone is blindly chasing the experts, while in a good group setting there is no need to chase the experts.

> Have we ever looked at a tough scientific problem and came to the conclusion that the best path forward was to collect as many random people off the street and shove them in a room to solve it?

Has a scientist ever solved a tough problem growing up in isolation to other scientists? I consider "standing on the shoulders of giants" to be a form of group intelligence. But yes: We have done something similar at RAND corporation. The problem was: Forecast the impact of future technology on the military. The solution was to collect experts (not random people), put them in a room with an experiment leader, and gradually converge to the best forecast, using anonymous feedback every round. It's called the Delphi Technique and it is still in use.

Also, there is an experiment running right now, that takes random civilians, has them answer intelligence questions ("Will North-Korea launch a nuke within 30 days?") and gives weights to their answers, according to previous results. This way random civilians trickle up to the top, that individually beat a team of trained intelligence analysts, simply using their gut or Google. It's called the "the Good Judgment Project". Put ten of those civilians in a room and you have an intelligence unit that is not afraid to be wrong, does not have a reputation to uphold, and does not care about any group pressure, authorities or restrictive protocols that may hamper a group of real intelligence analysts.

> Also bootstrapping or model parameter selection techniques are already heavily used in AI

I believe the parent was talking about model ensembling/ model averaging, not ensembling techniques used by single models, like the boosting or bagging that random forests use. If you have a single attack as input crafted for a single model, then a voting ensemble of three models (lets say: random forests with Gini split, regularized greedy forests and extremely randomized trees) will not be foiled.

To state that AIs can be foiled by a specifically crafted adversarial attack, does not mean the AI is very fallible. Under normal circumstances it still outperforms humans. But let's say the error rate of a net is on average 3%, and an attack works only if it is crafted for a specific net (where the weights are known etc.). Like a team of humans is able to come up with better solutions than individual team members, so does an ensemble of nets (usually) outperform any of its individual members. For instance, the final model generalizes better, because it uses information and predictions from all the nets. Foil one out of ten nets, and the other nine will cancel out the bad prediction with their votes.

In an automated future jobs will be taken up by AI. A security guard then becomes a supervisor: 100s of nets will try to detect disturbances in a mall, and when they find one, they notify the supervisor. A judgement call is then made by a human. This is currently happening with law expert systems. A judge will input the case and get a prediction for punishment. Then make a final adjustment to this punishment, looking at the context of the case. Such a system prevents racially based sentences (the AI will not care about race, but look at precedent), while still giving emotions, judgment and reasoning a final say.

> it seems that some problems are inherently prone to making mistakes. Can it be avoided at all?

Some problems are incomputable, like finding the shortest program to reproduce a larger string. In others, like lossy compression, mistakes may be made in representing lossless information, that humans are unable to spot. I think this is a very interesting, but difficult question to answer.

> who do we blame when an AI makes a "mistake" like that? The training set?

We blame the AI researcher that build the net :). And she will blame the training set :).

The speakers runs you through a hypothetical case study: a pet-door company and looks at the pitfalls of applying machine learning to it.

I believe the paper is actually focusing on something else: Create images that humans will not be able to classify as a digit, but that the net will gladly give a prediction for. To translate to faces: There may be clouds that look random to our human brain, but are detected as faces by nets.

It seems this is an adversarial attack, where you need access to the guts of the net (weights, layers). I compare this with a hashing algorithm and brute-forcing the input till you find a collision with a target. Nearly impossible in real-life situations.

You may be able to sign a check using a scribble that the cashier can not recognize as a digit, but that the machine will recognize as a digit. Not much practical gain from an attack there.

A block of 8x8 pixels, possibly with some filters, could easily repeat, say, a piece of the blue sky. Think of compressing three modern-day English books. Compressing each of them individually with their own table would likely result in a larger size than compressing the three books together with one big table. The compressor can make use of repetition of data across the three books to keep the file size smaller. You can test this with a good compressor. It's a principle that is used in normalized compression distance (comparing the length of individually compressed files, by the length of compressing their concatenation, to obtain a similarity metric).

It is compelling, because it is probably true. I think Sloot did not manage to create a device that could play movies to be released in the future, without requiring an update to the device.

> But I don't why this would have us believe in the infinite movie device.

The infinite movie device was a ruse. A misinterpretation of the claims by media and investors, which the company did nothing to stop. Jan Sloot never claimed higher compression ratio's than 8x (you can compress the entire series of Lost better with a single code-book than each single episode with its own code-book). He talked more about encoding, not compression.

> Are the patent examiners greater authorities than the experts?

I think the information theory experts are railing against a claim that was never made. A device that holds every movie ever made, or to be made in the future, unlock-able by a tiny keycode, can not possibly exist. I think Pieper, Perkins and the Oracle DB experts knew this. Apparently there was more to this invention than this ridiculous claim of near infinite compression.

> human psychology and greed

I think this was a different pre-bubble time, where people thought they could invest and hype up companies, and have the technology build later. Speech recognition companies were worth millions before their tech was even proven to practically work. I suspect that such a force worked behind the scenes in this company too. Just see the corny time-capsule that is their flash site (warning, plays sound): http://www.davoc.com/indexnow.html

> my first presumption was that you were trolling, which I'd define as pretending to believe an untenable position to see if you can get a rise out of others

I am sorry for not making my ideas about this case more clear. For the record: I do not think that endless or recursive compression is possible. I am just trying to apply Occam's razor to this intriguing case: Was Jan Sloot the biggest conman of the 90s? Fooling the biggest venture capital companies in the world? Did he really romantically invent something special as an outsider? Or was his set-top box idea really worth something? I don't believe Pieper could be fooled by such a con. Maybe he knew?

> Without trying to be too offensive, who are you, and why did you write this comment?

I am anonymous. Not trying to rile you up, but prefer to keep it that way. I have, not too recently, abandoned an old HN account and starting over. I wrote this comment, because this case is intriguing, and for me, responsible for my attraction to compression and information theory. I first heard about the Pigeonhole principle in relation to this case.