Saturday, February 19, 2005

SBL RESOLUTION UPDATE: Jim West comments:
The Society of Biblical Literature survey results are published now and other bibliobloggers have offered their perspective. I think the results were skewed because a) anyone who got the link could participate- and unfortunately the link was made widely available to non members of the society. And b) intentional "padding" of the survey was performed by many who simply cleared their "cache" and took the survey as many times as they liked. In future, such surveys / resolutions should be sent, by regular mail, to current active members of the Society only and a signature should be required in order for the survey to count.

He copied this to Matthew Collins, who wrote back:
Hi Jim,

Actually the survey recorded via individual identifiers in the link who responded and who didn't. It only allowed one response per identifier - so even if the survey was forwarded or the cache was cleared, it would work only once for each member. The results were in fact not skewed, but accurately reflect the membership.

-Matthew

I'm afraid this doesn't solve the problem. The language here is not very specific, so it's a bit difficult to interpret. The issue was never clearing the cache, but rather that deleting the cookies allowed one to vote multiple times. If a cookie is the device for recording via individual identifiers in the link, then the concern I raised is not addressed at all. If Matthew means that they checked their server logs and weeded out multiple votes from the same IP address (which I suggested in my unanswered e-mail to him), that goes part way toward addressing the problem, but doesn't solve it. In the first place, it would not correct cases where the same person voted from two or more computers (home and office, for example). It would also incorrectly delete a vote from two SBL members (e.g., spouses in the same household) using the same computer. Second, dialup services sometimes assign a different IP address to the same user with each session, so someone could have voted, logged off, then redialed and voted again later.

In addition, Matthew's reply doesn't address the problem of votes from non-SBL members. The Host Names of IP addresses are rarely so specific that they pin down the actual name of a user, so nonmember votes remain generally undetectable even with the information from the server logs. Nor would it be possible to attach a single hostname to a specific member in most cases.

In a word, the survey mechanism was insecure and the results should not be considered accurate. In the future such online surveys should be password protected according to the individual member's SBL password.

All this not to speak of the whole question of whether the Society of Biblical Literature should be getting into this sort of political debate at all.

UPDATE: Ken Penner e-mails:
Re: your blog on the SBL survey.
I think Collins was referring to the digits after the ?A= in the link sent to each member as the mechanism by which each member could vote only once. The link I was given was as follows:
[I've deleted the link here. - JRD]
Is the link you were given the same or different?

Yes my link was different. It looks as though Ken is right. Sorry Matthew, I withdraw my objections regarding security.

No comments:

Post a Comment