"Too close to call."

It was David Brinkley's election night epitaph to the 1960 Kennedy-Nixon matchup; as it was 40 years later for Peter Jennings during the long Florida night that left the Bush-Gore contest in dispute.

With results within the margin of error for manual, mechanical and digital vote counts, television networks reworked their outdated predictive models, and Congress -- through the Help America Vote Act -- set a timetable for introducing electronic voting machines. Out in limited release this year, full rollout is anticipated in 2006.

Direct recording electronic voting machines are the new hanging chads of American politics, sparking a debate over disenfranchisement amid concerns about validating that every vote is counted as it was cast.

If the e-voting debate is over which votes are counted, the implementation of customer relationship management (CRM) in presidential politics raises equally important questions about which votes are cast. Complex and highly partisan CRM is being deployed by both major parties to help tip election results in their favor, district by district, mobilizing their respective bases and wooing fickle swing voters.

The Democratic National Committee built "Demzilla" with demographic, geographic and psychographic data on 158 million Americans. The Republican National Committee locked in its "Voter Vault" the same kinds of data on 165 million Americans. Given the information's sensitivity, it belies otherwise sophisticated political apparatuses that both systems have names that are at once, sophomoric and Orwellian. What's more, after limited use in local and state races, Demzilla and Voter Vault go head to head in their first presidential throw down next month.

The number of names is less important than the contextual data wrapped around each name.

"We have a numeric coding system," said Washington state Republican Chairman Chris Vance in an interview about the Vault. "One is a hard Republican. Two is a soft Republican. Three is an independent. Four is a soft Democrat. Five is a hard Democrat. Six is someone we reached, but refused to answer our questions. A zero is someone we have never been able to reach, we know nothing about."

The first five categories bring a certain scientific precision to the art of mobilizing the base, but the political prize is in converting zeros to partisans by election night. That puts political CRM in the cross hairs of the same groups targeting e-voting as a threat to democracy.

At issue are the inferences drawn from manipulating previously discrete data elements, including the usual stuff about who we are and how to reach us, and inferences gleaned from our reading habits and organizational affiliations. Layer on whether we vote and make political contributions (derived from secondary use of public records) and our views on war, gun ownership and abortion (which we may volunteer to the earnest, PDA-touting campaign volunteer at our door) and we end up with targeted messages that serve up "my president, my way."

Apparently we are unaware that the candidates look different to people placed in the other buckets.

One academic observer went so far as to condemn the parties' segmentation strategies because he claims they are not just correlated to, but the cause of, a precipitous fall in voter participation. Curiously that complaint doesn't appear to have been extended to the legion of advocacy groups using the same methods, punctuated by media campaigns and even clothing lines, to convert nonvoters to political participants.

This could all be a hideously bad idea. Or it could be a defining characteristic of a new civic engagement that solves some old problems while creating new ones.

Paul W. Taylor  |  Contributing Writer