The Big Green: Data Mining Noise or Risk?
February 26, 2012 1:43 PM   Subscribe

What effect do my Ask MetaFilter searches on behalf of others have on data miners' sense of who I am?

Recent conversations about Google's privacy policies, data mining and sale of data related to online activities prompt the question of what impact my research in order to answer questions here has on whatever profiles of me are being generated. Do these searches add to the noise and help camouflage searches about topics that are personally relevant? Do they contribute to a risky-looking profile? (For example, I've done searches on medical topics like diets for cancer patients.) Both? Is there a particularly good way to anonymize searches done for HiveMind questions? Is there truly reason to?

I'm not familiar enough with data mining practices or profiling to ask a more informed question, so please add any aspects you think are missing.

Mods, if this would be better asked on the grey, please feel free to move it there.
posted by MonkeyToes to Computers & Internet (9 answers total) 3 users marked this as a favorite
 
Mod note: Pulled the link, but otherwise this looks like a generalizable question that may as well stay here.
posted by restless_nomad (staff) at 2:01 PM on February 26, 2012


Are you thinking about:
A. searches you perform with Google, looking only on this site?

B. searches you perform using this site's internal search (the search box in top right)?
posted by LobsterMitten at 2:09 PM on February 26, 2012


Response by poster: LobsterMitten, A.

To clarify: If I see a question on the green I think I can answer, I'll open a tab (same browser, same window), run a Google search, click to links from there. So not all of my Google searches are for *me*; I'm wondering how data mining companies/algorithms/data buyers interpret the wide range of subjects I click to in connection with answering questions here.
posted by MonkeyToes at 2:27 PM on February 26, 2012


I don't think this is unique to searches on behalf of AskMe questions.

I think that most people - or at least many people - do searches on topics that are not personally relevant. Many people generally curious about the world / things they read, and thus might look up a medical condition that they do not have, or a medical condition that a friend of a friend has.

So, although I don't know much about online privacy and what types of data is gathered, I would find it surprising if one or two searches was enough to seriously impact your profile. I would be very interested to hear if I was wrong, though!
posted by insectosaurus at 2:42 PM on February 26, 2012


"Data mining companies" are a bit of a red herring here. First, you have to restrict the number of data mining interests to those who have access to the data, which here appears to be Google, who I don't think really shares too much of what you're talking about.

Beyond that, if they were even trying to figure out why you search for so many different things, or who this weird person is with such a wide range of interests, I offer a couple of points:
  1. The last 10 years of US political spying on its citizens without being able to find e.g. Faisal Shazhad have told me that you're talking about a difficult problem in the world of datamining.
  2. If it wasn't a difficult problem, companies would be much more effective at getting me to buy a particular laundry detergent.
Your "profile," such that it may exist, is easily concluded to be mostly noise given current and foreseeable techniques. It's a lot to winnow! The best they seem to be able to do is what Target was written up as doing to try to augement pregnancy purchases, which if you think about it is very rudimentary in the world of advertising. The pregnancy market is probably one of the most predictable in the advertising and marketing industries, and Target is barely able to capitalize on it using datamining, even when focussing on it specifically.
posted by rhizome at 3:15 PM on February 26, 2012


Response by poster: rhizome, that's exactly the kind of thing I've been wondering about, especially because I saw that story within a short time of hearing this comment, from a recent interview on "Fresh Air":

"TUROW: Well, I've gotten ads recently based on my age, OK? Particularly for some reason having to do with social dating sites, possibly...

GROSS: You're married, wait a minute.

TUROW: Yes, I'm married, but I have a student - and this is where things get really weird. I have a student who's doing research on social dating sites, and I've probably gone online to check some of the stuff that he's been doing, to look at some of his work, and somehow they've made inferences about me.

And I don't know exactly how they've gotten my age, but they're converging data around that, and I've been getting a bunch of things about marriage - about dating, I should say."

And later:

"I am also concerned about an area that fewer people have brought up, though some have, and that has to do with social discrimination. And that is the notion that in an everyday world where companies are deciding whether I'm target and waste, and making up pictures about me moving forward, and I'm getting different ads and I'm getting different discounts and I'm getting different even maps of where I possibly will sit in an airplane based upon what they think about me, I would call that social discrimination. And the potential moving forward for us to be in what I call reputation silos, where we are thought of in a persistent way by advertisers as opposed to what our neighbors are thought of has a lot of ramifications of how we see ourselves and how we see other people."
posted by MonkeyToes at 3:21 PM on February 26, 2012


You can see some of what Google infers about you here.
posted by vasi at 10:28 PM on February 26, 2012


This is something that has never occurred to me and I thank you for bringing it up. I research things for older people and friends who are way too busy to spend hours digging through material to find an answer to a question they have. Very often it concerns a medical condition or a legal question and that could be problematic. Something to think about and take more seriously than I have so far.
posted by aryma at 10:53 PM on February 26, 2012


In that excerpt, Turow seems to be saying that even if they get the ad wrong by showing a married person a dating ad, they're still locking on to him. Never mind that he's selling books on the topic, I guess we really can't win against the adbots! I'm therefore skeptical, since he may just be satisfying an existing demographic abstraction, like, "marriage-minded people read Salon."
posted by rhizome at 12:04 PM on February 27, 2012


« Older Estate lawyer in GTA?   |   A clever name for a kids geology website? Newer »
This thread is closed to new comments.