My name is danah boyd and I'm a Principal Researcher at Microsoft Research and the founder/president of Data & Society. Buzzwords in my world include: privacy, context, youth culture, social media, big data. I use this blog to express random thoughts about whatever I'm thinking.

Relevant links:

Archive

Cobot and Data that Matters

[From OM]

In Implicit or creepy?, David is dead-on and i would like to expound on this.

First, if you aren’t familiar with the lessons of Cobot, you should be. Cobot was a nice friendly little bot that sat in LambdaMOO, collecting data for its masters. Members of the MOO were bothered by this and felt that Cobot should give back to the community it was observing, like any good social scientist. So it did. You could ask Cobot anything about the social patterns going on and the data it was collecting. People started asking ego-centric questions: “Who do i talk to the most?” and such. And then, people started asking who other people talked to the most. Trouble emerged from there. All of a sudden, human jealousy reared its head. People were irate that those who they spoke to the most did not speak to them the most. What did this say about reciprocal value? Gah!

Cobot’s willingness to provide social data created a social rupture because it was evaluating data, not its meaning. Yet, people who were accessing the data were deriving meaning. They were using coarse data about social relationships to imply something much deeper. Sound familiar?

I talk to Phil from the corner deli more frequently than my best friend or my mother simply because of proximity. Yet, they play a much more central emotional role in my life than Phil. Quantity and quality are often not correlated. Yet, if some system were to rank my relations and Phil came out above my mom, damn straight she’d be pissed.

The way that systems and users of systems interpret our data often affects how we interact with them. When Viegas and i were visualizing email data, we often joked that our systems motivated you to write more messages to the friends who had strong emotional connection but apparently not frequent email connection simply so that they played a more visible role.

In the case of David’s metadata, this is particularly true. How many of us can truly list our favorite books? We know that this will be publicly displayed. What we list is a performance where we try to select titles that convey something meaningful about us for the viewer. We count on that audience, on that interpretation in selecting our titles. We are performing for that human audience to interpret, not the system. Yet, if the system starts interpreting our data, we may shift our scope of audience. But then what is it that either the system or the humans are interpreting? Are they capturing essence? What happens when the system re-projects its interpretations back to a human audience? How do we then deal with this doubly-mediated projection of self to a human audience?

It is not simply creepy, it’s outright destabilizing.

Print Friendly, PDF & Email

5 comments to Cobot and Data that Matters

  • social and scholarly

    One of my greatest aha! moments learning about hypertext theory was when I at a conference naively commented, over drinks I expect, that hypertext requires links. A spatial hypertext expert (I think it must have been Frank Shipman, or maybe Gene Golovc…

  • The underlying issues – apophenia and jill/txt

    After reading today’s posts at both apophenia, jill/txt, and the posts to which they link, I’ve been thinking about the discussion of hard and soft social data. While the authors seem to place the bla…

  • The underlying issues – apophenia and jill/txt

    After reading today’s posts at both apophenia, jill/txt, and the posts to which they link, I’ve been thinking about the discussion of hard and soft social data. While the authors seem to place the blame for faulty research outcomes…

  • tony

    why not assign a value to emails,posts,im,etc. say a 10 for schedule/meetings/job related/family emergencies,1 for a “hi”. of course people will misuse it and assign 10’s for “hi, how ru?”

    I tend to notice my longer emails go to people I care about,my shortest to business relations.Could value be assigned by word count?
    just wondering

  • Dan

    “I tend to notice my longer emails go to people I care about,my shortest to business relations”

    This isn’t true, in my experience: quite the opposite. Emails to people I care about are often quite short: passing on news, arranging meeting times, etc. Business (in my case, academic) emails need to cover more areas and convey more facts. Perhaps that’s just because I tend not to deal with emotionally involved discussions by email – other people will vary.

    On the wider issue, I think you could draw one of two conclusions from Cobot:
    a) people just need to interpret data more wisely, and know the limits of statistics
    b) we need ambiguity in our social relations.