Law in the Internet Society

Houdini: Magic Election Data


The 2008 Presidential election cycle will be remembered as one where technology played a critical role at several junctures, and without the use of which the United States would not have elected President Obama. Specifically, a key part of his day of election efforts centered on the Houdini Project where information about voters who had already voted disappeared off call lists as if by magic. The rationale for the investment in technology was that it increased the efficiency of Get out the Vote (GOTV) activities. This short analysis seeks to explore the presence of significantly more information about voters entering and staying in the electoral process.

GOTV and data

The traditional method of getting out the vote was achieved by fielding local residents in the weeks prior to the election who would obtain promises of votes. Those names would be typed or printed out on paper with several carbon copies and hung on a wall. Representatives of the candidate would identify voters at polling stations, pass this information to office with the lists on the wall and as people were identified as having voted they were crossed off. Every four hours or so one copy of the list would be torn off and runners sent to houses to identify and encourage those who had not yet voted to do so. At the end of the day the paper was scrapped and the only data available to the candidate in the next cycle would be that in the minds of the repeat volunteers. This worked well in areas where a candidate had a high proportion of voters and a large number of volunteers. Where the volunteers were thin on the ground and didn't know the area it didn't work so well. Moreover, bringing in outside volunteers was hampered by their lack of geographic familiarity of the area.

These methods have developed considerably over the last 8 years. Nowadays it would be uncommon in the United States for a list to come from one of two databases most of which cover the whole country (VAN, andVOTER vault, Catalyst.) Moreover, as these databases can be accessed remotely, and voters will confirm voting intentions by phone, promises can be obtained in neighborhoods never before canvassed. Specifically in the Houdini project those who recorded the voters as having voted no longer have to physically transfer the record but can phone the result in to a phone response system or upload via a web interface.

Superficially, the change seems small - lists can be generated more than four times a day and they are likely to be more up-to-date. Practically, the process is significantly different – the tasks assigned to volunteers are more modular and the movement of data across the net and into the database no longer requires people to travel (sometimes long distances.) Skill sets of volunteers can be narrower and fewer people have to know the geography of the area. Fundamentally, this makes the deployment of resources more efficient by a significant factor. There is less unnecessary traveling between locations (already aided by the ever present Google map), and more effective deployment of resources with scarce skills (such as personal relationships).

Much more importantly though is that the data is persistent, granular, and linkable. Persistent inasmuch as it exists and can be used in subsequent election cycles. Granular in that the sharing by voters of the issues most important to them rather than just a pledge to vote provides significantly more insight about a voter. Most importantly it is linkable with data regarding magazine subscriptions, credit ratings etc. This change doesn't seem particularly important until you consider the possibilities - political donation requests can be centered around your most important issue, calibrated with your credit rating, and aimed at generating a donation from the high school friend (identified because you are friends on facebook) who very likely has a higher income than you do since they subscribe to Conde Nast's travel magazine.

Possible solutions

The absence of anonymity online, tied to the ability to accumulate data within a campaign has only just begun to be exploited - Only subsequent to the election was it reported that Obama victory sweatshirts prices were calibrated to the level of prior campaign donations. Moreover, though some of the data may become less accurate it is hard to see the Committee to re-elect Obama failing to use the data collected in 2008. Moreover, they will be building on technology not assembled on a tight deadline (as in 2008) but quietly over the prior 4 years. Not only will this result in an even better fundraising operation, but it also means that the the politicla campaign, an entity that was always transient and never had any assets bar its bank balance now has a valuable non cash assets. What will former President Obama do with the data after he has finished with using it? Participation in election cycles in America have not only become more costly but it has generated salable assets that don't have a trivial value.


Though the presence of this information cannot be magicked away it is possible to enact privacy rules and practices that will permit voters to share only data they want for only the period they want. A permission based approach where the data is stored by the user and only accessed by those permitted at a particular moment would partially solve the problem of data that is nowadays is consciously shared but access to which cannot currently be meaningfully revoked subsequently. For implicit data, data about voters harvested by cookies placed by campaign websites, this can be limited by those who wish to limit it through use of technologies that permit pseudonymity on the web such as TOR servers.

However, and in a way that Houdini would surely have appreciated, data is here to stay and though the voter disappears off the screen the information about them does not. The presence of this information in election cycles going forward has barely begun to be appreciated. Perhaps the saving grace for now is that the Houdini system broke down across many parts of the country in 2008. We still have a few years to enact technologies and rules that sustain some level of voting privacy.

-- TomGlaisyer - Dec 10 2008

Feel free to comment below



Webs Webs

r3 - 22 Dec 2008 - 05:58:22 - TomGlaisyer
This site is powered by the TWiki collaboration platform.
All material on this collaboration platform is the property of the contributing authors.
All material marked as authored by Eben Moglen is available under the license terms CC-BY-SA version 4.
Syndicate this site RSSATOM