GroupLens Research

An article on Read/WriteWeb says that Google Maps are now editable by anyone with a google login (which surely must be everyone by now). The idea is that you can change the address, location, and link for any of the designated places on the map. This is a very cool step forward in crowdsourcing. It will be interesting to see how well they control vandalism.

One of the interesting questions is what methods Google is using for watching for vandalism.Many crowdsourcing sites seem to be using a "security through obscurity" approach. I wonder if it will prove over time that open approaches that publish effective security methods work as well? The ESP Game seems like one good example: the rules are published, but by working to ensure the players are strangers, most attempts to hack the system are not effective.  (The slashdot attack is a good counter-example.)

What do you think? Excited? Worried?

John

Anyone with an entrepreneurial spirit and a good idea for a recommender system should take note.  Strands will invest $100,000 in the best recommender system start up.  Finalists to present, and winner to be announced at RecSys08.

There's a fun article in the Winter 2007 AI Magazine about "Machine Ethics". The basic argument is that as machines get more and more in control (e.g., planned army robots that would fire weapons), it is more and more important (to humans) that they behave in an ethical manner.

The article argues that there is a fundamental difference between implicit and explicit ethics. Implicit ethics would be programmed into a machine by its designer, much as Asimov's imagined three laws of robotics. Explicit ethics would also be programmed in by a designer, but at a more fundamental level: the robot would be able to compute the ethics of new situations based on a fundamental understanding of ethics. The authors argue that explicit ethics are necessary for several reasons:

1) So it could explain why a particular action is right or wrong by appealing to an ethical principle.
2) Because otherwise it would be lacking something essential to being accepted as an ethical agent. (Kant admired agents that work consciously from ethical principles more than those that work slavishly from rules.)
3) So it could adjust to new situations, evolving the appropriate ethics.

(1) is a red herring: explanation systems often appeal to principles they don't understand in any sort of principled way. For instance, in our work on explanations for recommender systems some of the most effective (for humans) explanations were only loosely connected to the operation of the recommender.

(2) is in contradiction with an argument the authors make later in the paper. They argue that even though computers won't be conscious in the near term, they should be accepted as ethical agents if they act ethically. Agreed! So, then, all we need is that they act ethically.

(3) is intriguing. On the one hand, it would be remarkable if an AI agent could evolve new ethical patterns for situations it has never seen, based on core ethical principles.On the other hand, the results of that evolution might be very surprising. For instance, if a military robot were to decide, based on ethical principles, that it ought to prevent an attack on Iran that a general wishes to carry out, how would that be perceived, by the military, by the loyal opposition, by the anti-war effort? What if the robot assassinates the general to prevent the war? Overall, given our track record in predicting the performance of complicated software systems, I have some doubts about this approach.

I liked a later quote in the article, which says that ethical relativists cannot say that anything is absolutely good, even tolerance.

John

I just set up an account for my daughter with audible.com, and downloaded a book for her to listen to on the bus. The good news is that is appears to be all set up now, and ready to download to her iTouch. The bad news is ... everything else.We spent nearly an hour buying a single audio "book", and getting it copied down to her computer. The problems were nearly all related to digital rights management, though I'd class them in two groups: fundamental, and incompetent.The fundamental problem is that DRM makes downloading and using media much more difficult. It restricts which programs and devices you can use it with. Further, is it any surprise that downloading a program whose fundamental purpose is to prevent proscribed uses of a media file makes it more difficult to successfully use that media file? In the case of audible, we had to download a program to my daughter's laptop that insinuated itself into firefox and itunes in unspecified ways, so that she could download the Audible files she had paid for to her laptop, and thence to her iPod. This program failed to install itself properly the first time -- apparently it doesn't check to see whether itunes is running, but fails mysteriously if it is. When we tried to download the book we had paid for to her computer, we kept getting mysterious error messages. These went away once we reinstalled the software.The problems of incompetence were mostly caused by a user interface that tries to pretend that the challenge is easier than it actually is. The Web site makes a big thing out of the four simple steps required to get going with Audible. Step 1 is "Pick a plan". We didn't want to sign up for a plan, so it took us a while to figure out that you can buy books without a plan. Step 2 is "Download Audible software". In the description it says "You can also use ITunes to download audio ...". We decided to go that route initially, before figuring out that apparently the audible.com software is required in addition to iTunes. It didn't help that the iPod Touch is not listed in the "supported devices" list, so we had to guess which software we need. Step 3 is "Purchase and download". Our problems with this step are described in the previous paragraph. Step 4 is "Transfer your audio to your AudibleReady device". Here the solution was easy: we just had to figure out that Audible had created a new sort of "playlist" in iTunes, and that we had to tell iTunes to sync that playlist with the iPod Touch. A common step in iTunes -- but it would have been nice for Audible to walk us through that step.A very frustrating hour later, my daughter is pretty happy with having her book ready for the bus. I'm much less happy. Audible seems like a company that is going to fail if they don't figure out these user interface issues. What, then, will happen to the DRM that requires a "phone home" to install the book on a different device? (Yes, even the iPod Touch will one day seem outdated.) Even though I'm eager to listen to "books on pod" while I exercise, I refuse to buy these DRM-crippled alternatives. Yes, convenience is worth a lot, but more important to me is the principle that media that I buy must be usable for me into the murky future, independent of the survival of any one company, format, device, or business model.What do you think?John

Newsweek has an article that argues that Web 3.0 is going to be all about injecting the experts back into the information production and dissemination process. I think they've gotten the big picture badly wrong, but the saddest quote in the article is about why one of the 'experts' they interview thinks this change will come about:

Fueling all this podium worship is the potential for premium audiences—and advertising revenue. "The more trusted an environment, the more you can charge for it," says Mahalo founder Jason Calacanis, a former AOL executive who was previously involved with several Web start-ups. It's also easier to woo advertisers with the promise of controlled content than with hit-and-miss blog blather. "Nobody wants to advertise next to crap," says Andrew Keen, author of "The Cult of the Amateur," a jeremiad against the ills of the unregulated Web.

Pretty amazing that the argument is that advertisers are going to fight to prevent the amateurs from taking over information processes so they can protect their advertising revenue. (Newsweek is, of course, heavily supported by advertising.)

It's also interesting that none of the examples they give in the article -- from Google's wikipedia killer to the Maholo search engine -- have any real traction in the marketplace. I think we're seeing a fantasy here. People whose business depends on the elites managing who reads what where and when are arguing that we have to return to that model to make sure "good" information gets out.

I was speculating the other day about how different the world would be if there had been some way that radio and television could have been supported through a fee-based model, rather than the advertising-based model that we have today ...

John

Wired magazine recently published an interesting article on the Netflix Prize:

This Psychologist Might Outsmart the Math Brains Competing for the Netflix Prize

The article is a fun read. It provides some perspective on the importance of tuning algorithms and the potential for combining many algorithms for one prediction task. It also makes it clear that the prize-seeking community is very open to sharing results and techniques. Cool.

I would have been interested in reading more about why the researchers think going from 8% RMSE improvement to 10% improvement will be so hard. Is is because they've (finally) bumped up against individuals' abilities to accurately represent their own movie preferences on the 1-5 scale? I ask, because I had thought we were already there before this contest! How much room is there for algorithms to get better at predicting our individual rating idiosyncrasies and inconsistencies?

Max

Read/WriteWeb has a look at a very interesting new Web site called Traackr.  The idea is that you tell Traackr all of your media accounts (YouTube, Flickr, MySpace, etc.), and Traackr logs in every day to see if you have new content, and to see who has been looking at your content.  Traackr then generates cool graphics showing how much visibility your different content items are getting across the entire Web.  After all, what could be more interesting than seeing how interesting you are!The idea of a site that brings together all of your accounts around the web is not new, of course, but it could be very valuable.  Traackr is less interesting than some efforts in that it is focussed on showing you what is happening on all of those sites, rather than on helping you remember all your usernames, passwords, and how you actually interact with each of them ... but it's at least a step towards bringing together information about these accounts.John

I just got back from the Collective Intelligence FOO Camp that O'Reilly organized at Google.  The meeting was great, the people were great, and overall the experience was great. 

One issue that popped up is what exactly people mean by Collective Intelligence.  At a high level, it was clear that everyone meant basically the same thing:

agent -> work
agent -> work
agent -> work                                                                 (some require superlinear)   
agent -> work              ------ combining function  ------> outcome that would
agent -> work                                                                   be harder to produce
agent -> work                                                                   with any individual agent

Interestingly, a number of participants were only interested in examples in which the outcome was superlinear in the number of participants.  I'm not sure why this would be.  Several participants were speculating about what a "complexity theory" of collective intelligence would be like: could we identify problems that are demonstrably more difficult for a collective intelligence to solve than other problems?  

I'm personally more of a "big tent" CI guy.  I think that as long as the result is intelligent, I'm okay with situations in which the individuals agents are providing the real intelligence, and the combining function is simple.  If we want to taxonomize, I can see at least three interesting types of CI: 

 Types of Collective Intelligence

1. parallel intelligence: many independent agents (e.g., Wikipedia, reCaptcha)
2. aggregate intelligence: independent agents + combining function that joins the results (e.g., recommender system)
3. emergent intelligence: the result is intelligent, even if the individuals are not (e.g., ants foraging, leaving scent trails)

Overall, the experience was fun.  I did find it intriguing that we had no tools for applying collective intelligence to the process of creating a "unconference".  For that, we used white boards and markers, lots of sticky notes, and pieces of paper to cover up events that were cancelled.  It would seem easy to do better: people could propose ideas, which would show up on people's laptops. They could say which ones they would like to attend, which would cause them to be scheduled so most people did not have conflicts, and so they would be in rooms of approximately the right size.  If noone wanted to come, they could be cancelled or merged.  It would be fun to see CI in action at a Foo Camp of the future ..

John

ManagedQ is a very cool search engine interface.  It runs on top of google, and presents a visual view that aggreggates pages according to people, places, and things.  Kind of fun to play with ... but the fact that it requires the Shockwave Player means I can't use it on some of my favorite platforms :(.  I very much like the higher-level view on top of the vanilla Google interface.  Check it out!John

Read/WriteWeb reports on Tiinker, a news site that uses machine learning to figure out what sort of articles you're interested in, without messing around with all that social information that reddit, digg, etc. are based on.I think this is a huge step backwards: machine learning is fine at figuring out topic, but lousy at *quality*.  Further, how can Tiinker know your interest in something new that you haven't read about yet?  I'd be much more enthusiastic about an approach that would combine the social news of a reddit or digg with machine learning to create smart, social news.  In a system like this, people would read news that they're interested in, based on what other people like them have been reading, and their own automatically learned profile.What do you think?JohnPowered by ScribeFire.