GroupLens Research

As a die-hard political junkie, I've been looking for something to fill the void left by an abnormally tidy presidential race.  Luckily I live in Minnesota, and get minute-by-minute updates about the Great Minnesota Recount. Projecting the winner for the recount is difficult: Were challenged ballots type I errors or type II errors?  Exactly how many challenged ballots were withdrawn by each candidate?  And would Al Franken or Norm Coleman be a better representative of the political views of The Lizard People.

The Star Tribune has enabled readers to vote on the outcomes for challenged ballots.  I assumed that the reader votes were for entertainment purposes, but the Star Tribune has cleverly analyzed two million reader votes to project the final outcomes for over 6,000 challenged ballots.

It is easy to imagine that these online votes are biased.  Online users trend democratic, and Democrats may award more ballots to Franken.  However, some anecdotal evidence hints that the Star Tribune's projections beat those from political experts.  During the past two days, the Strib's projection has hovered around a 75 vote lead for Franken.  Meanwhile, projections from Nate Silver of FiveThirtyEight, a highly respected voting analyst, have slowly been converging to the Strib's.

Maybe if we had crowdsourced the original counting of the ballots we wouldn't be in this mess!

The drama here in Minnesota these days is the recount of the U.S. Senate race -- and it's dramatic: a difference of a couple of hundred votes out of 2.5 million cast, a hand recount, going to court, etc. Kind of like Florida in 2000, but more competent. (Minnesota Public Radio has good background coverage.)

But the most interesting part is the "challenged" ballots: these are ballots where the election judge doing the recounting wanted to classify the ballot one way, but a campaign observer objected. There are a few thousand of these ballots; most of them are fairly clearly frivolous, being done for PR reasons, and now the number of challenges is fluctuating day by day as the campaigns withdraw and unwithdraw challenges.

These ballots are now before the Canvassing Board, a 5-member panel convened by the Minnesota Secretary of State to determine the final disposition of each. You can watch the proceedings, live during the next few days or archived. Here's a direct link to one 90-minute segment of today's proceedings. (I was only able to make the link work on Internet Explorer. However, don't use Explorer until you've patched the latest nasty zero-day vulnerability!)

I find the process fascinating. Each board member, and a lawyer for each campaign, has a binder containing photocopies of each challenged ballot. For each ballot, the committee chair (MNSOS Mark Ritchie) announces which ballot is being considered, and then makes a motion with a proposed disposition ("I move to reject the challenge and allocate this ballot to Coleman"). Simultaneously, the board members and lawyers examine their photocopies, and an aide passes the original ballot down the line of board members. The process takes a few tens of seconds for a ballot where there's immediate agreement with the chair's motion, and maybe a couple of minutes if there's discussion.

Having the original ballots (Minnesota uses optical scan voting) is critical; some of the ballots which are clearly one way when viewing scans online are clearly something else when the original is viewed. (Example: The pen ran out of ink. The bubbles are blank in the scan, but on the original you can see the indentations in the paper where the voter tried to mark -- clear voter intent, which means the vote must be counted under Minnesota law.)

Also, if the decision takes more than a few tens of seconds, the camera switches to a laptop feed, where there is someone manipulating a PDF viewer to show the ballot. You can see the mouse cursor and UI manipulations and everything.

It's a very human process, with synchronization all accomplished by verbal announcement, at least one synchronization error ("Mr. Secretary, I missed the decision on ballot Minneapolis foo"), and occasional dry humor on the part of the committee members.

Bottom line, I think there's some HCI research with a great story here. I think the basic premise is a case study of the human processes going into this laborious and monotonous task, with some participants having extreme vested interests, some but limited computerization, the need for high public visibility, and a very high-value outcome.

Who's gonna do it?

(Crossposted to my personal blog.)

My friend Jamie Thingelstad has a blog about a new technology for RFID in the home, from a company called Violet.  The basic idea is that you get a small USB device for your computer and a collection of RFID tags.  You can then associate RFID tags with actions on the Violet web site.  The hope is that eventually people will come up with cool uses for this technology in the home.

The video is mostly underwhelming: very few of the apps are things that I can imagine using, mostly because they are no more useful or fun that existing alternatives.  One that might have some legs is the ability to put RFID stamps on postcards, so the recipient gets taken to a page with related content (e.g., pictures from your trip).  This app, however, suffers from a network effect: it's only useful once lots of people have the readers, which will only happen once there are lots of useful apps.

It will be interesting to see where RFID technology for the home goes.  Nice to see the tools getting delivered ... now it's time to come up with the apps.

Any app ideas?

John

The Times Online has a terrific article about Nicholas Taleb's reaction to the market crash.  Taleb is the author of The Black Swan, a very fun read about his theories of unpredictable events.  (It's interesting that the ideas in the book come across as much more attractive than their author, who presents himself as unconscionably arrogant for someone whose strongest arguments are about the limitations of human knowledge!)

I found several things enchanting about the article. First, it's amazing to an American that a professional athlete in Britain writes and thinks like Ed Smith does.  Our athletes mostly get in fights with their bodyguards in bars; they certainly don't write with insight about major economic events.  Second, there are a number of wonderful quotes that strike at the heart of Taleb's arguments.  We'll go through those individually.

So is Taleb really against expertise, or is he simply pitting his own against that of the experts? He got this call right. The fall was forecast-able and he forecast it. It was not really a black swan.

This problem grabbed my attention while reading the book.  Taleb keeps attacking the ability of "experts" to know what they're doing, all the while arguing that he himself knows exactly what he's doing, because of his analysis of the limits of knowledge ... but how can he -- or we -- know that he knows what he's doing?  What evidence can be trusted, given the limits of all evidence to predict the future.  Taleb comes back to this point when he says:

But Taleb's victory today is a pyrrhic one. “I wake up every morning at 2am, scared. I have made money on my bet that the financial world will go under. But now, if the banks go under, I can't cash my money. If I follow my logic all the way through, I get scared.”

This point is the critical one for his reader: how far ought we to follow the chain of logic in predicting that astonishing disasters will occur with (unpredictable) regularity?  Ultimately some of the disasters will be ones we can't recover from.  What about those?  It's impressive that Taleb made lots of money by betting the economy would crash, but how does one short sell an ecological disaster, for instance?

A phoney meritocracy of people who got massively lucky and think they did it all themselves is a recipe for social disaster.

[later]

There is surely, for Taleb, an uncomfortable irony. Much of his present notability is due to his having made one devastatingly accurate prediction. Had he got his forecasts for the fall of banking wrong, the error would only have strengthened his general theory of black swans. But it would have undermined his popular reputation, which has never been higher. Now that really is luck. I leave him hoping that it may prove contagious.

Lovely writing about lovelier thinking.  Taleb is powerfully convincing about the limitations of human knowledge to predict outside of their sphere of understanding.  However, his theories aren't powerful ones, like, say, chaos theory, that give us rich meta-theories from which to predict the overall structure of likely events.  Instead, they just tell us that very, very bad things are going to happen, and that their happening is completely unpredictable.  Fascinating ideas ... but what should we do about them?

John

Cute cartoon that reminds me of Nathan Good's paper with Joe Konstan (and others) on the timing of software license agreements.  It would be very nice if we could reliably change our mind, perhaps after the monster shows up.John

Read/WriteWeb has a list of 10 Micro Trends for your next startup.  Many seem plausible or likely, and Read/WriteWeb folk are usually spot on, but a few seem off-base to me.  Let's talk about those, since that's more fun than cheering the wins.4. Micro-trend Slopes replace Chasms.  Seems to me this mistakes what is happening in startups with what is happening in the world.  In the world large organizations matter (5!), large organizations are slow to move, and large organizations try to make decisions hierarchically (6!).  Crossing the chasm is tough enough even if you aren't pretending it's not there!5. Small is the new big (my cutesie phrase).  The problem is that there are economies of scale, and accumulations of power that are only easy in large organizations.  There are lots of benefits to an economy to having lots of small organizations be where the action is -- but lots of benefits to the organizations to be large.  Until there's more pressure put in place to prefer small, the dinosaurs are going to continue to roam.6. Self-organizing networks beat command and control structures.  Hard for a scientist not to like the "evolution beats intelligent design" argument ... but I think this technique works best when you have time to make millions of mistakes on the way.  Designed structures with lots of flexibility for cross-pollination seem the more likely winners in the short term.John

Lessig has yet another interesting article on Remix culture, this one adapted from his new book.  The key take-aways we've already heard.  In my view, the most important of them is the need to simplify the automatic generation of copyrights for the wide variety of remixes that have no commercial cost to the copyright holder.  Yes this would reduce the power of the copyright holder, but to the benefit of our society, which, after all, ought to be served by the copyrights.John

There's a very interesting story on Hot Hardware (of all places) about how two news robots interacted to help kickoff the (false!) rumor that UAL was going bankrupt, leading to a huge selloff in the stock.  Apparently the news bot for a newspaper moved the story to the front page because someone read it during a "low news time" -- during which even a single read was a lot.  Google News saw the story on the page, and picked the date up from the top of the page, since there was no date on the article.  This date was the current date, not the six years ago date of the original story.  An analyst saw the story on Google News, and a billion $ later, the rest was history.As a recommender researcher, the scariest part of the story is the inference drawn from that single view of the article, which looked at the time -- in the wee hours -- like a statistically significant indicator that the article was becoming interesting.  This problem reoccurs in recommenders all the time: if we're looking to make recommendations of items that are not popular, we'll often be recommending based on not very much data.  John

Read/WriteWeb has an article about a fascinating new browser plugin called SpinSpotter. The idea is that readers of journalism on the Web can mark it up as "spin", and propose alternative edits to the text that was in the original article.  For instance, if a journalist writes "Most observers believe that ...", the reader might change it to "Some observers believe that ...".  (If the reporter doesn't have a study to cite, it's spin for him to suggest that he knows it's "most".  He might be right, but journalism demands that he either cite a source or leave the opinion out.)  Other readers can then choose to see the original text, the original text with symbols marking the "spin", or the edited text.Two studies that would be fascinating to do: (1) Does the presence of the spin markers change the messages that a reader takes away from an article?  Are readers able to judge articles more carefully because of the markup?  (2) Do different users agree on the appropriate spin markers?  If not, should a recommender system be used to show users spin marks that they agree with?  What effects does that have on the messages readers take away from articles?John

Interesting article on Read/WriteWeb about how YouTube is thriving despite the fact that much of its content is copyrighted by others. The basic business model is simple: if a content owner finds its content on YouTube, it can ask for the content to be removed -- or it can choose what ads should be shown next to it, and presumably take a profit from the ads. Apparently many content owners are choosing to let YouTube keep their content live, because they benefit more by having it seen by the YouTube crowd than be preventing its borader sharing.This development is exciting for the file-sharing crowd, because it offers a potential path for broad sharing of information, in increasingly creative ways. However, there are important limits to the potential benefit. The simplest to understand is that mashups are likely to not thrive under the advertising business model. In most collaborations everyone thinks they're pulling more than their share of the load. Thus, the owners of the content in a mashup are likely to each individually want more than their share of the advertising profits. (This problem is exacerbated by the fact that profits in this model are mostly shared after the popularity of a video is known; as founders of startup companies know, doing a deal before you know how much value you're splitting is easier.)The deeper problem is that this approach to enabling creativity only works in an ad-supported content model. There are cultural risks to offering large companies the most access to people's attention. Over the long term, it might be healthier for people to directly pay for the media they wish to consume, but the YouTube approach is continuing us along the ad-supported path.John