Don Marti

Sat 29 Nov 2014 09:59:05 PM PST

Unpacking privacy

Maybe the word "privacy" has something in common with the "freedom" in "free software". Privacy is a big heavy word, with too many meanings to be be a good part of a business message. Some free software people handled their version of the problem by coming up with the open source brand, to help close deals without having to have a big conversation about freedom. Maybe what we need today is something similar, a name for a subset of privacy that's worth money.

The big place to cash in is display advertising on the web. The money in advertising is in signaling, not direct response. And an ad medium can optimize for response rate or for signaling power, but not both. So there's clearly a small part of "privacy" that has cash value: for publishers and brands on the web, the quality of having an audience rather than a set of database records, the chance at making web ads work like magazine ads, not like the "windshield flyers" they are today.

Before the emergence of the "open source" brand, people kept having "software freedom" vs. "commercial software" arguments. But the problem wasn't freedom on one side against business on the other. The framing around open source made it clear that some kinds of commerce work better when market participants have some kinds of freedom.

Today, "people want privacy" sounds to me like "people want freedom" and "people want data-driven services" sounds like "people want software functionality." That's a recipe for wasting a lot of carpal tunnels on having two different arguments, threaded together. We need a new word for the economically helpful aspects of privacy, so that we don't have to argue about a word that's just as complicated as "freedom" every time we want to implement a subset of it.

(We do need to keep talking about freedom and privacy sometimes, even though they're hard words. But just as we can have better Internet freedom conversations when we can show examples of corporate-supported free software, I mean open source, we'll also be in a better position to talk privacy when we can point to whatever the new word for a subset of privacy is projects that work in the interests of publishers and brands.)

Publishers and brands can both use whatever the new word is. Publishers first. When ad networks can track the same user from expensive sites to cheap ones, and agencies buy impressions based on who the ad networks say the user is, then high-value sites (the ones that invest in original content) are stuck in the business of selling the same impressions as lower-value sites.

Once the ad networks have a user labeled as a "car intender", then some low-end site can show him a cheap cat GIF and get paid to run a (relatively) high-value car ad on it. Makes it harder for the sites that actually review cars. Content sites lose, and intermediaries win.

The question then is, why do high-value sites participate in user tracking at all? Why not just run only first-party ads? There's some research on that. The problem is that if the medium is targetable, then the best strategy for an individual site is to do targeting, even if (because of the signaling value of its content) the site would do better in a system where no user could be targeted. When we stop thinking about privacy as a big, complicated, hard concept, and try to break out some kind of Minimum Viable Privacy, just enough to protect that "car intender" from site to site tracking, then ways out of the race to the bottom start to present themselves.

For example, high-quality sites could be encouraging users to install anti-tracking tools, to make those users less targetable anywhere. This would reduce revenue in the short term for the high-quality sites (by making inventory disappear) but have a much more dramatic effect on the lower-quality sites that are only viable because of targeting. For brands, the case for helping and encouraging customers and prospects to protect a subset of "privacy" is even stronger. Just need a word for it.