Monday, June 26, 2006
Of the 150 or so e-mails that I receive every day, my guess is that about 10 percent of them are spam -- for example, offers to take my fair share of $30 million that has been won in an African lottery and must, must be moved to a safe and secure bank account in the United States. Like you, I have tried some sophisticated ways of getting rid of spam, such as filters and rules. None of these is entirely satisfactory. The University of Illinois's spam filter gets a few egregious messages, but it lets most things through. The College of Law spam filter is overly zealous, tending to classify as a "Junk Suspect" any message without a subject heading but letting through any message, however slimy, with a subject heading. In the last couple of weeks the COL filter has proposed deleting messages from some of my oldest and dearest friends. When replying to them, I have to be careful to delete the "Junk Suspect" that the filter inserts in the subject heading. So, because neither of these filters is particularly effective, I have to watch what I delete carefully and must read the first couple of lines of most e-mails, even those that appear to be spam. There ought to be a better way.
Perhaps there is. Over the weekend I was in London. Tim Harford, the Undercover Economist, had an interesting column in The Financial Times Magazine ("Let's Get Personal," June 24.25, 20006) in which he proposes a market solution to the problem of spam. The gist of it is to enlist the help of third parties to screen our mail for us. He suggests that we "compile a dossier about ourselves and our families, including birthdays and anniversaries, favorite authors and music, need for loans or mortgages, and what big purchases are under consideration. We would own that information and could give it or even sell it to companies who wanted our business." In fact, he proposes that "information agents" would be willing to pay us for that information. In turn,they would then sell information from our dossiers to others -- specifically only to those businesses in which we might plausibly have an interest. The hope is that if we hadn't expressed an interest in helping African lottery winners to hide their winnings in the U.S., we would never hear from those fraudsters again.
To a very limited degree, something like Harford's scheme already exists. For instance, Amazon.com and Barnesandnoble.com, with both of whom I do a lot of business, send me unsolicited e-mails announcing new books by authors whose books I have bought in the past or that say something to the effect that "We notice that those who have purchased music by Waxen Wayne have also purchased music by Eban Flo." (There is a website called AuthorTracker that will automatically send you an announcement of new works by some of your favorite authors, so long as they are published by a certain publishing house.)
Actually, I like those messages. Several times these notes have lead me to good new authors or musicians, new books by trusted authors, or new musicians. But these are different from what Harford proposes. These are sellers who are plausibly using information you have already shared with them. Presumably one can opt out of this system, telling Amazon.com or Barnesandnoble.com that you do not want any further such notices.
I'm a little skeptical of the practicality of Harford's proposal to enlist third parties to serve as two-sided agents for us and for retailers (and others). It's not so much that this wouldn't be better than the current situation. We might get better tailoring of the junk e-mail that we receive under Harford's proposal. But I'm not sure that it would deter the purveyors of drugs, lottery schemes, and fraudsters.
I've always liked Ayres' and Nalebuff's proposal (available in a short radio-show format here) that all incoming e-mail have a "Penalize me" button. If it was unwanted e-mail, the recipient could hit the button; so doing would assess a five-cent penalty on the sender. (There would have to be a method, which Ayres and Nalebuff explain, for enforcing the fines.) Presumably this system would cause senders to think twice about mass mailings. Rather, they would be well-advised to tailor their mailing to receptive (and unlikely to penalize) audiences. So, the end result might be the same as Harford's, but I suspect that the Ayres-Nalebuff would be more practicable.
This "Penalize me" scheme needn't apply just to spammers. It could apply to ever sender. I have had some messages from colleagues for which I'd speedily have hit a "Penalize me" button. To avoid overinclusiveness, one could presumably set up and update a list of senders for whom one would be willing never to hit a "Penalize me" button. We can already do that for those who send graphical content -- a "Safe senders list." I'd love to be in the faculty meeting at which we debate whether all of our colleagues should presumptively be on that list of senders not to penalize.