VIA Interaction

Tuesday, December 18, 2007

Two heroes together in one rest room

Two of my heroes, John Tierney of the New York Times and Donald Norman of Northwestern University, take on the "automatic" rest room faucet: Scientific Breakthrough: How to Wash Your Hands - TierneyLab New York Times Blog.

Anyone who makes things that people use or interact with should read Norman's Emotional Design. And I have added The Design of Future Things to my Christmas list.

posted by Tim Beidel at 12/18/2007 07:07:00 AM

---

 

Monday, December 03, 2007

Behavioral targeting, for better, or worse

About 16 or 17 years ago, as a technology writer for the Albany Times-Union, I wrote an article about the concerns for personal privacy around computerized data collection and storage.

At the time -- when the largest hard drives available held 210 megabytes and the personal computer at my office was not on a network and had its floppy disk drives disabled by a wary IT department -- the loss of privacy was more theoretical than actual.

Well, we're getting closer to that time that privacy advocates worried about.

The AP has a story today on behavioral targeting, the idea that Web sites show you advertisements tailored to what your Web browsing patterns say about you (see Ad Targeting Improves on Web Sites - New York Times).

"Based on the weather reports and restaurant listings you check out online, Yahoo Inc. has a good idea where you live. Based on searches you've done, the Web portal might also know where you want to go. Don't be surprised then to suddenly see an advertisement on flight deals between those two places. It's what United Airlines did with an ad on Yahoo earlier this year as people browsed for something completely unrelated to travel."

As a member of the advertising industry, this seems great to me. The Internet (and other media, soon, thanks to the digital revolution in audio and video delivery networks like cable and satellite systems) provide a cost effective way to "track' what people do and serve them ads in context.

As a human being, I find it a bit worrisome. I have come to appreciate seeing an ad for a cheap flight to somewhere I want to go. (That kind of contextuality is at the heart of the information architecture practice here at VIA.) I am sure, though, that I do not want someone selling data about where I surf that is attached to my name.

Right now there is a modicum of privacy built into the cookie model. Web sites can place information on your computer (a cookie -- a small text file that the Web browser manages) and they can read that information. That Web site can't read cookies placed there by other Web sites, so there is a security model that works.

The trouble comes with the cookies accompanying online ads. These come from the Web site that serves the advertisement, most likely not the Web site you are visiting. So if you visit a bunch of different Web sites, and the ads on those sites are served by the same server, that server can put two and two together based on the cookies and start to create a picture of you: My Web history this morning suggests that I like reading about religion and college football(or is that redundant).

This can all be defended, I think, when the picture created of you is anonymous. That is, the behavior can't be tied to me personally. Just a nameless construct of me for the purpose of selling someone like that construct something they would be interested in buying.

Unlike the days of sneaker net, though, tying you to your behavior is not only possible, it's a business model. If you are a registered Yahoo user, Yahoo has the technology to be able to tie your behavior to your specific identity. Yahoo will share information about you with itspartners under confidentiality agreements, according to its privacy policy.

The ability of advertisers to target people who may be especially interested in their products is either better (if you are an advertiser, or you would prefer to be exposed to ads for things you might actually find useful), or worse (if you believe that your behavior may say things to some people -- like insurers or potential employers -- that you believe is private, or misleading).

(As an aside, thinking about 1991, when the coolest thing in mass storage was the Bernoulli Box -- a portable hard/floppy drive hybrid cartridge that held 90 megabytes for $1,100. I bought a loss-leader 2 gigabyte USB drive last Christmas for $20. So in 15 years, the cost of portable data storage went from $12.22 per megabyte to $0.01 per megabyte. That's a 99.92 percent decrease.)

posted by Tim Beidel at 12/03/2007 11:26:00 AM

---