Enjoying this article? Subscribe to our weekly C&E Newsletter & bi-weekly CampaignTech Newsletter by clicking here.

Craig “Campaign Doc” Varoga is a longtime Democratic strategist and manager. Questions on strategy, general consulting, or anything campaign-related? Send questions using LinkedIn, Facebook, Twitter @CVaroga or CVaroga@Varoga.US and he’ll answer them right here. 

Q: With all the technology changes happening, how much, if any, of my campaign should I automate or outsource to AI?

A: Varoga’s Rule: Half of all predictions are accurate, we just don’t know which half. But, in the spirit of political hacks immemorial, let’s go for it.

Artificial intelligence (AI) is already a factor in politics, from Google alerts to analytics-driven media buys, which is why the Centre for Public Impact (among others) is studying this buzzy future “in which everything from voter intelligence to voter targeting and conversational engagement can be automated.”

“Hope,” for example, is an app that can be used on various devices, using digital stories and conversations to give actionable options for political activation to friends, family and followers, presumably including debate-watch parties, canvasses, pushbacks against negative ads, et cetera. AI also has a future role in day-to-day scheduling, fundraising follow-up, data crunching field reports and sorting news stories and media impressions.

And to go totally sci-fi: Imagine the political ramifications of AI-enhanced brains, i.e., “No handwritten notes allowed in debates? No problem, the candidate’s cortex is hooked up to the press secretary’s quantum computer.” Imagine what that would do to the debate regarding candidates and the release (mandatory or not) of their medical records.

Imagine, too, a prominent Twitter account driven by either AI or an AI-enhanced cortex. Just when you (we) thought the information environment couldn’t get any more glutted.

Q: Campaigns get a lot of data from social media sites and digital firms. How can we be sure those numbers don’t include bots and fake accounts?

A: In 2016, 9 percent of desktop display and 22 percent of video spending was fraudulent. In 2017, by some estimates, 22 percent of all dollars spent on digital video execution will be consumed by fraudulent entities, resulting in $6.5 billion in pilfered funds. According to Harman Sodhi of Videology, “Digital advertising is rife with fraud.”

There are protocols to mitigate this fraud, according to the Association of National Advertisers (ANA) and White Ops, a cybersecurity company that protects digital advertisers and web app owners from automated threats, ad fraud, account stuffing and fake engagement.

Just two (there are more) of these “ANA/White Ops” protections include: One, requiring publishers to identify all third-party traffic sources, and two, including specific language that buyers will pay only for non-bot impressions. When the transparency in the first example does not occur, buyers should reconsider relationships.

Videology’s Sodhi counsels to avoid relying on post-delivery reports, since this approach answers the fraud question after the election, at which point it is too late to change either your persuasion or GOTV budget. Sodhi recommends, instead, working with both your vendors and cybersecurity experts on the front end to eliminate fraudulent activity before it occurs.

Q: One of our consultants complained before our recent retreat about the silliness of pollsters and their questions. Do you agree or disagree?

A: Let’s stipulate the obvious:  Pollsters (like all humans) exhibit the full spectrum of behavior, from the heights of wisdom and intellectual acuity to the lows of idiocy and CYA foolishness when they get something wrong.

Wisdom: A long-time pollster, who will go unnamed on this page, realized that an election was in jeopardy, despite happy poll numbers, because he understood the underlying structure of his own survey and because (imagine this!) he understood the state from an historical, non-polling perspective and believed (correctly) that late-breaking trends weren’t going to be captured in a snap-shot tracking poll.

Idiocy: Asking voters if they have a "strong" opinion about issues on which they are under-informed. Nonsense-in/nonsense-out is exacerbated by electorates that hate politics and are segregated according to biased news sources.

A strategist friend mocks this phenomenon with a fake Q&A: Do you agree or disagree that modern physics relies too much on the string theory? "Ahh, I guess I agree." Strongly? "Yeah." 

Substitute poorly worded descriptions of tax reform, deficit spending or any controversial issue, and it is inevitable that the “scientific” poll results will include an embarrassing profusion of false positives and worthless negatives.

So yes, pollsters (like all people) can be silly, but the ultimate failure is not theirs, it is managers and general consultants who lack the ability to force consulting teams to avoid blind spots and dumb mistakes.

Craig Varoga consults on local, state, national and international campaigns and is a regular political analyst in numerous news media. Send questions using LinkedIn, Facebook, Twitter @CVaroga or CVaroga@Varoga.US.