Recently, TOPP Labs embarked on a joint venture with Transportation Alternatives to build a “Candidate Survey” website, showing New York users how their local political candidates responded to a TA survey.
TA is going to build the public-facing site, and TOPP is going to build a back-end service that does geographic lookup of candidates based on the user’s address. Our service will be publicly available for other uses too. (If you’re paying attention to such things, you might wonder why we need a new service and don’t just use the Votesmart API or the Mobile Commons Legislative Lookup API. Well, MobileCommons only provides lookup of districts, not the candidates within those districts; and Votesmart has some info on local officials, but not unelected local candidates.)
The big sticky issue that’s come up is crowdsourcing the data. Some people at TOPP really want to make the candidate data editable by the public.
We’re not doing anything as critical as counting votes on this site, but recent elections have shown that disinformation is still a problem, whether perpetrated by pranksters or partisans. For example, there were several cases of bogus fliers just before the 2008 election.
I think a totally open system is too risky. Our demo of the back end service at http://pvoter.org/ is currently totally open; anonymous users have free reign to edit or delete information about candidates, or even add fictitious candidates (if you can figure out how; it’s not hard). It would be easy to deface TA’s website or anything else that used the service.
What’s the solution? Restricting access to some “trusted” users? Open editing with moderation by some “trusted” users? No user editing at all?
Comments welcome.
Purple voter seems to be working! Making the candidate profiles editable like a wiki-candidate thing is a really interesting idea, but it would be almost impossible to get any political candidate to participate in it. They might do so behind the scenes–but probably not in an official way.
That said, there’s a lot of value in getting the public’s reaction to a candidate and a collection of information that’s not published by the campaign. For instance, voters might have been interested to know that Pedro Espada doesn’t live in his district…
I am not sure what the final solution should be, but as a start I with the following:
1. To edit, one needs to be a registered member for 48 hours.
2. To register one needs a valid email address and to fill out a simple captcha.
3. Keep a revision history so that admins can easily revert spam
I don’t edit on Wikipedia so I don’t know their exact process, but it looks similar to the one described above, and it seems to work pretty well.
In addition to some kind of user registration/verification, what about requiring a citation and using moderation. Edited entries could include some metadata about whether they include a citation and if they’ve been verified by a moderator. Consumers of this data source could then decide if they wanted to filter the data by those factors or show disclaimers. Wikipedia displays a prominent disclaimer when someone flags an entry for lacking citation or not having a neutral point of view, but the content is usually still left in place.
I agree with the above two options. We should allow for contribution with with some user validation. Anonymous mark-making on the web is something that has had its day. I think people who want to contribute to a conversation, collaboration and knowledge, are ready to leave a digital trail of their thoughts. Again as mentioned above Wikipedia seems to be getting a lot of it right.
Here is an interesting read from 2006 on pattern language around moderation strategies http://radar.oreilly.com/2006/03/etech-clay-shirky.html
Thanks for the comments. There are some interesting possibilities here, which will need to be explored if/when pvoter is going to be useful beyond NYC.
For now, I’m going to punt and on my branch of PurpleVoter I’ll simply make all data editing restricted to authenticated users. Anonymous users will have read-only access.