What I saw during the Superbowl


I had fun watching the Superbowl, and it was interesting to see a great defense contain a great offense. It was also fun to explain what was going on (on the field) to some attendees of our CIS2010 workshop who were more familiar with the other football. I hope they enjoyed it too!  But of course the Superbowl is not (only?) about football: there is the half-time show, and the ads. The most striking thing about the halftime show was that The Who are still sort of functional as a musical group. Who would have thunk it?

Some of the ads, however, have me a bit worried. In particular, there were two — the Audi and the Google — that triggered the latent George Orwell in me. Is it really a good idea (no matter how tongue in cheek) that the government have the power to coerce individuals’ behavior as shown in the Audi ad? While I am all for recycling, the mere premise that recycling should be motivated by threat rather than incentive strikes me as both perverse and subversive of our rights.  I guess I am not the only one with a negative reaction: Jeffery Goldberg calls it Gorewellian, while an eco-energy blog laments the Nazi allusions and the disservice to the green cause.

The Google ad was more insidious, depicting a boy-meets-girl scenario through a sequence of queries. Two things jumped out at me  in this ad: first, a lot of the queries were reasonably long (more than just a couple of words), suggesting that internet search is in fact evolving. The second was just how much could be learned about a person from the queries he or she writes. The Google ad was intended for a general audience and needed a positive message, what with Valentine’s day coming up, but one could just as easily analyze a person’s query stream to find “how to cure VD?” some time after “tu es très mignon.”

The point is that our queries can reveal a lot about ourselves, and the ability to aggregate them is not without danger. The danger comes in two forms: from the search engine company itself that mines data for purposes you may not like, and from the government receiving that information on demand. It’s not even clear whether rotating search engines will protect anonymity, since over time that is just as likely to leave actionable traces in more than one company’s logs.

It’s time to think seriously about how to protect individual privacy in our interactions with search engines. I can think of several broad approaches to improve the situation:

  1. Coercive: Constrain search companies’ ability to use queries for data mining purposes through legislation. Sort of HIPAA for search. This to me seems unworkable given the impossibility of auditing the use of such information.
  2. Subversive: Employ portals/gateways to obfuscate identity, depriving the search engine of useful identity cues. This is much better, but may make it possible for the search engine to block traffic from such a portal, assuming the number of people using it is small.
  3. Persuasive: Convince a major search provider that there is an opportunity to gain market share by attracting people interested in better privacy protection. While this may not seem particularly important right now, I am sure that at some point data privacy will become more important. Another possibility is to pay not to have your data collected.
  4. Evasive: Use off-beat search engines with much less coverage and data mining capability for most searches. An index 20% of the size of Google’s should still cover 80% or more of what most people look for.

None of this is likely news to those who are concerned about search privacy, but the airing of these Superbowl ads seems like a good catalyst for conversation.

Share on: 


  1. Yes, it is true that the level of inference that can be gained from a personalised “web history” is quite worrying. Then again, I still think I would be more conformable about a singular web-history, vs. something like Facebook that is amassing an entire profile. The idea that I can obfuscate searches is a lot more appealing than OCR which may be able to pick out place names/locations that I have not revealed in text from a picture. Maybe I am a little to suspicious though.

  2. Have you not seen the parody of that Google commercial? Check it out:



  3. The dark side of search history might be one of the unstated points of the terrific Slate parody of the Google ad: http://tinyurl.com/yhf6zgp . Certainly if you chanced to look at the ill-advisedly published AOL data (must be 3+ years ago now), most real query logs drove the privacy point home. (Yes, yes, I shouldn’t have looked, but it was hard not to–one’s eyes are drawn to just this sort of accident.) An example of the AOL data extracted from my blog in August 2006:

    “AOL user number xxxxxx [redacted for privacy] not only searched for Elvis (‘Elvis has not only left the building he is dead’), but also for ‘banana pudding with cool whip and sour cream’ and ‘younger man older woman.’ … Although I’m sorry to hear about those ‘tiny bumps after tanning bed use’ and ‘signs of depression.’ Not to mention the ‘can’t get an erection’ and ‘how to firm up saggy legs.’ You can learn more about Elvis’s fans than you can about the King himself.”

    For most of us, what we search for reveals way too much about us (and leads right back to us). We are not nearly horrified enough about what can be learned (and what is already mined) from our queries.

  4. Social comments and analytics for this post…

    This post was mentioned on Twitter by HCIR_GeneG: Posted “What I saw during the Superbowl” http://palblog.fxpal.com/?p=2965

  5. @James, I don’t use Facebook for that reason — too much aggregation, too little control, too wishy-washy with the privacy policy.

  6. @jeremy hey, I came up with the VD line all by myself!

  7. […] social media. People leave extensive traces of their online activity on social sites (and on search engines in general), and a range of Social Network Analysis algorithms originally developed by sociologists to analyze […]

  8. […] while ago I wrote about the general threats to one’s privacy posed by search engine histories. It appears that […]

Comments are closed.