Victoria Stodden from Stanford Law is giving a Berkman lunch talk on the Internet and democracy. [As always, I’m paraphrasing, getting things wrong, omitting important issues, etc. You can always hear the whole thing at Media Berkman.]

She’s an idealistic empiricist. Her course was changed by Danny Hillis who told her that if she really wants to make a difference, she should create tools.

She points to three possible relationships between the Net and democracy:

1. Net as disseminator that increases the flow of info. The Net can help make us more autonomous in determing opinions, and can provide better info. But, who are the gatekeepers? And do we end up more polarized than before? (She cites Fishkin as an alternative to Sunstein’s view on this.)

2. Net as tool for implementing group decision-making processes and opinion formation. Estonia votes on line. The 2002 S. Korean elections were affected by activities organized online. Danny Hillis has a “Collective Reasoning Tool.” Local communities might lose out, however, as we move to topically-based groups; this might cause geographically-clustered racial groups to lose influence.

3. The Net can educate about democratic possibilities. But governments can block this if they fear it.

She raises some ideas that could be explored empirically: Map the pervasiveness of Net use and how it’s used; a map of the “attention backbone” with an eye toward the extent of polarization; differing regulatory structures; and the correlations among these aspects.

Now she switches topics and talks about a Computational Sciences Research License she’s working on. A new license is needed because people want to be able to release massive amounts of research, and not just the final results. The research might include code and media. Creative Commons doesn’t want people to put code under its licenses. [Very Everything Is Miscellaneous, if I may be allowed to reduce the entire world to my terms.] Her license would require attribution, but not much all, and would use CC for the media and maybe GPL for the code.

Q: [melanie] Yesterday Science Commons released an Open Data Protocol.
A: That’s exciting.

Q: [oliver] Is it time to rethink what “authoring a paper” means? Is there data authoring, for example?
A: I’m not with this questioning the norms. But if, according to the license, I use your data, I attribute it.
Q: But just citing the data doesn’t seem enough.
A: The scientists understand how valuable and hard data collection is and just want the attribution.
Q: If I were the scientist who did the data collection and aggregation, I’d like more than just a simple citation. Maybe we need some type of super citation. If the paper author had called the scientist, the scientist would probably be listed as a collaborator…
A: Authorship and citation varies from field to field.

Q: [oliver] I liked your conceptual model, but I’m still looking for a model about how it works. Is it a type of neuronet or what? You’re an economist, so what’s your model?
A: Ideally you’d like to have repeated instances, but that can be hard with social sciences. So I think we should start with the case studies. I’m not sure one model is going to work.

Q: [ethanz] The case studies end up being enormously controversial. You actually hit 3-4 topics that were big controversies around the Center. E.g., there’s controversy over whether in 2002 in Korea the Internet had anything to do with it. We’ve been trying to get beyond anecdote to data. Very hard. How do anecdotes like these turn into testable, statistical rigorous research? We’ve been struggling with this. And, by the way, there are only 16 anecdotes in this field. [laughter]
A: I am worried about the amount of data out there. N Korea isn’t going to turn over a lot of data.

Q: [me] It sounds like we’re waiting for history to happen. How do we do data collection and analysis when history hasn’t happened yet?
A: Yes, that’s the problem.

Q: A question about the taxonomy. Why did you pull education out of dissemination?
A: It’s all arguable. You can’t have democracy if people don’t know what it is, so I gave education its own area.

Q: [wendy] How do you attach rights to what copyright law doesn’t cover, such as the collection of data?
A: We treat the data as copyrighted.
Q: Well, in programs the functions aren’t copyrightable, and the data collection isn’t either.
A: The license doesn’t cover the data itself. The license intends to require attribution for the data but doesn’t protect the data otherwise.

Q: [oliver] Maybe you could make a cooperative…
A: You’re thinking of the neuroscience example.

Q: [terry] How can we do better, looking forward. We often know that a critical political event is about to happen. Is there some way we could enhance prospectively our data gathering capability so that after the fact we’d be better able to assess the relative effects of various factors, including the role of the Internet? Second, there’s been a lot fairly fundamental work in statistics, breaking away from old models of controlling for variables that distort your understanding of the impact of a variable on an outcome, and new techniques for isolating the impact of one variable. Do any of these new techniques of addressing Ethan’s speculation that cellphones are 10x more important than the Internet?
A: Maybe there are comparative studies we could do. Some of those new statistical techniques might be helpful. But there are probably going to be so many confounding factors…
Q: [ethanz] When we work from anecdotes, we’re working from extraordinary cases. When Suharto falls because of mobile phones (supposedly) there are lots of other dictators who didn’t fall even though there were plenty of mobile phones around. So I think there’s something to Terry’s comment. We’re not going to be able to get all the data we need. But if we were to say that Ghana’s ’08 election is going to be interesting, we could think about what data we wanted to collect. And we could collect similar data in surrounding countries.
A: Maybe collect data from random countries and see if we can anticipate hotspots.

Q: [jz] We’re hoping to ferret out which countries filter what and when. We would love to release this to the world. But some of us are reluctant because we’d just be giving playlists for one country to discover what sites it’s missed. How might we share it with the world and make it difficult to be useful for bad purposes?
A: Not really. How can we release data and make sure it’s not used by bad people. But maybe there wouldn’t be a race to the bottom of filtering.

Q: [rob] The independent variables are so co-linear that you’d need monster data sets to analyze them. The problem isn’t just statistical. The theories that link the individuals to mass movements are weak. To put it all together, you’d need both. [Tags: berkman democracy victoria_stodden ]

Categories: Uncategorized dw