Another message about Marshall's stuff from David Farber's IP list.
From: "David Farber" <dave@farber.net>
Reply-To: dave@farber.net
To: ip@v2.listbox.com
Subject: [IP] Economics vs. spam
Date: Tue, 17 May 2005 18:17:12 -0500
_______________ Forward Header _______________
Subject: Re: [IP] Economics vs. spam
Author: "James H. Morris" <james.morris@cmu.edu>
Date: 17th May 2005 5:12:24 pm
Advertising is Flirtation
I liked the van Alstyn and Loder paper on how to fight spam. It is actually=
=20
being used by a start-up, Vanquish.com. I've sampled their software and it=
=20
works as a challenge response system, at least.
Based on the premise that a message can have value to both the sender and=
=20
receiver, they present an elegant argument to show that even a perfect=20
filter is not a good idea. Economists say that these unilateral solutions=
=20
are bad because they decrease net social welfare, and they look for=20
mechanisms that promote the transmission and reading of those messages=20
where the sum of the value to sender and receiver is positive. To do so,=20
they recommend the obvious: exchange money to pay the partner who is=20
reluctant to communicate. This works fine when you're paying the sender,=20
but paying the receiver is more dubious.
Alladvantage.com tried this scheme in the 90's and failed miserably.
Apparently economists don't frequent singles bars -- notwithstanding the=20
bar scene in "A Beautiful Mind" in which John Nash discovers Nash=20
equilibriums. Paying someone to read an ad doesn't work anymore than=20
offering a woman $5 to talk with you does. You can pay for her drink, but=
=20
don't offer to pay her. Either she is offended; or, worse, she takes the=20
money and blows you off. Van Alstyn embellishes this scheme by suggesting=
=20
you escrow the $5 and she take it only if she doesn't like your line. This=
=20
seems unproductive, too.
The point is: advertisements are not one-time "information transactions";=
=20
they are inducements to learn more and eventually pay for something far=20
more expensive than the ad. They are flirtations; and, as any frequenter of=
=20
singles scenes knows, where and how you flirt is very important because you=
=20
are signaling about the kind of relationship you seek. Thus, it is much=20
better to simply buy her a drink, include a quarter in a junk mailing, or=
=20
throw money away in some other visible way. It makes you look generous,=20
encouraging interest.
Advertisers don't want to force their material on unwilling readers. Nobody=
=20
wants to show ads the reader no interest in. So if he know that an item was=
=20
ultimately not interesting, he wouldn't send it.
The essence of communication is that A tells B something B didn't know.=20
Thus the sender/receiver relationship is very non-symmetric. First, the=20
sender knows precisely what the information is and why he would like=20
someone to read it. He also is economically motivated, either to take money=
=20
for information or pay for advertising. The receiver, of course, doesn't=20
know what the information is. She might pay for it if she expects it to be=
=20
valuable, but is reluctant to read information for payment. More precisely,=
=20
if she does read it simply for payment; the sender is unlikely to receive=
=20
what he ultimately wants. Economists called this adverse selection.
But let's not ignore the economists' main point: the arms race between=20
spammers and browsers is wasteful because it leaves social welfare on the=
=20
table. We need to find ways to get more of those messages through. Bundling=
=20
is a better approach than direct payments.
Magazines are a way of bundling information in a way that accomplishes=20
this. The magazine pays authors for content that might interest the readers=
=20
and it charges advertisers. The reader pays the magazine, presumably for=20
the content; but maybe also for the ads that come along.
What is the the value of an item in a magazine to the writer and reader?=20
The crucial measure is reading time, how long the reader spends looking at=
=20
the item. It reflects the cost to the reader, her time, but also the value=
=20
to the sender, how much attention he's getting. Of course, the ultimate=20
value to an advertiser is the profit on something the reader buys; but=20
reading time should be correlate
The new idea here is that the value to both the sender and receiver=20
increases with reading time.
Whether something is content, advertising, editorial, op-ed, or a letter to=
=20
the editor, is really just a question of labelling the author properly so=
=20
that the reader can judge the content. The value of an article to the=20
sender rises with the level of rhetoric; i.e. the degree to which the=20
reader is persuaded to do something. But it is not a zero-sum game. Maybe=
=20
I'd be happier with a penis enlargement. :-)
James H. Morris
Professor of Computer Science
Dean, Carnegie Mellon West
412 609-5000
http://www.cs.cmu.edu/~jhm=20
-------------------------------------
You are subscribed as radev@umich.edu
To manage your subscription, go to
http://v2.listbox.com/member/?listname=3Dip
Archives at: http://www.interesting-people.org/archives/interesting-people/
-- Dragomir R. Radev radev@umich.edu Assistant Professor of Information, Electrical Engineering and Computer Science, and Linguistics, the University of Michigan, Ann Arbor Phone: 734-615-5225 Fax: 734-764-2475 http://www.si.umich.edu/~radev
This archive was generated by hypermail 2b30 : Tue Jun 09 2009 - 05:00:10 EDT