Data Matters II

A quick follow-up to my Data Matters post:  Progress has been made but the jury is still out.

Last month, the DNC approved the creation of the Democratic Data Exchange (DDEx), a legally separate entity that looks a whole lot like the RNC’s successful data trust.  They also named Howard Dean to chair the effort.  Dean, a former DNC chair who is well-respected by state parties, seems to be an inspired choice to allay fears on both sides.  This is all good news.

The devil, as always, will be in the details – and there’s little to no public information about the underlying mechanics and capabilities of the DDEx.  Two random commentaries I’ve seen do concern me, though.

First, the agreement apparently allows state parties to withhold “certain data” which they can independently sell to campaigns.  While I understand the economics from a state party’s perspective, this could be a huge loophole.  If state parties decide to withhold a ton of their data, the DDEx will be useless.

Second, beyond a DNC press release saying that “campaigns will … become the beneficiaries of cutting-edge investments in voter-contact strategies,” there’s no public definition of what the DDEx will do with the data they gather.  If the DNC sees the DDEx as a shared digital Rolodex, they’ve entirely missed the point.  I would certainly hope that the DNC just doesn’t want to show their hand.  However, given that their historic approach to data is uninspiring at best, I’d really prefer to see a card or two.

Standardized augmentation of voter data is likely something best done in the shared repository.  A massive amount of raw data can be gathered about individuals by mining numerous online government databases, social networks, online media, dating apps, etc.  Additional commercial data can be purchased from numerous vendors.  That raw data can be run through numerous algorithms (e.g. an AI-based psychometric analysis) to build a remarkably accurate (and downright creepy) voter profile.  These profiles can be used for direct voter outreach and can be aggregated to build national, state, district, and neighborhood profiles.  In turn, these multi-level profiles can be fed into shared analytics to inform a campaign’s fund-raising strategies, advertising content and buys, state/district/voter targeting, campaign prioritization, inter-campaign coordination, etc.

None of this is easy, implementation time is short, and the DNC/DDEx hasn’t yet released anything that implies they’ve started searches for technical talent nor built a game plan beyond a press release.  I guess I just have to hope that Dean is on top of it.

While the 2020 elections won’t be won entirely with data analytics, they can certainly be lost without them.