As some of my earlier posts have indicated I did a career transition from over 20 years in IT for a large corporation to a self-employed data scientist in 2015, and arranged my life so that I’d be able to participate in the Hillary campaign in a much more meaningful way than just donating money, due to lack of time, a demanding job, mortgage etc.
In April 2015 HRC gave her first racial justice speech and even though I’d not yet cut the cord on my old life, I did my best to try to get into the campaign at an early stage, using both official (eg, apply on website) and unofficial means (I’ve got friends, alumni, family etc fairly advanced in their careers and I was seeing if anybody knew someone who knew someone). Among other things I was skeptical about modern polling, but I was also an outsider to doing data work for campaigns and it would have been an amazing opportunity for me.
Could I have maybe spotted either the breaking of undecideds for Trump or the mobilization of conservative rural voters with just media and no ground game? I’ve thought long and hard about that alternate world, where I was in a fairly senior but subordinate role on the Brooklyn analytics team and my honest, best guess of that alternate world is “probably not”. If I spotted anything that triggered “that’s odd”, there would have been several people in the room with vastly more campaign experience than me to explain what I was seeing in ways that made sense. I wouldn’t have had the credibility as the outsider newbie, whatever my achievements and skills in my prior career.
During the primary I did my own private practice but watched the polling closely, and how it translated into state-by-state results, and admired the efforts of Benchmark to try to get it down to a precinct-by-precinct level instead of relying on statewide results. By Super Tuesday I’d noticed a pattern, even did a diary or two about it — HRC’s basic support in a primary tended to not change much from early polling. High name recognition etc. Bernie tended to hoover up all the undecideds, but couldn't cut into her core support at all. (note, consolidating the Not Hillary vote is something I consider to be a major political achievement, right alongside his fundraising prowess). So by early March I was pretty sure Bernie was toast and HRC would be our nominee (the rest of the primary went pretty much exactly the same way, although in a few of the later primaries HRC overperformed her early polling, as some of the Not Hillary support moved to the presumptive nominee).
In this period I'd caught up on the modern political theory of how to do effective GOTV, what analytics and ground games do for a campaign and what they can’t do etc. My application to work for HRC in 2016 was much stronger, and they did wave after wave of hiring where they said “your application is very strong, you have a good chance” but in the end, didn’t hire me (from April to late July).
If they had hired me then, could I have made a difference? It’s possible. I applied for a battleground position where I’d be the person who talked to the Brooklyn data team and to the state director organizing the data oriented volunteers for a battleground state, a position I thought a mature person with diverse background but who had data science chops would be especially attractive. I’d at least have the trust of the person I worked directly for, and if I had been in a midwest state or NC or FL maybe I might again have seen the raw data and gone “That’s odd” and had my state director and/or my peers in Brooklyn get interested and look for the same problem in other states. But what’s ironic is if I'd been a strong enough candidate to get hired, I’d likely have been assigned to a Western state (CO, NV or AZ, where the campaign pretty much got the electorate exactly right).
Again though I’d be fighting the perception that I’d be the newbie, learning how a campaign works, both in how they accept anything I found, and in my own head, if they had plausible explanations for what I saw that matched their experience.
In any event I didn’t get that opportunity. In many ways I’d probably feel even worse today if I’d landed either position and had been blindsided last night. Without the raw data, the odds I’d catch a signal and get it to the attention of someone who’d listen are essentially zero, so I don’t have the professional failure to face.
My plan B was to aggressively work to be a data volunteer, putting in as many hours as possible. This is something of a challenge in CA, as campaign data is mostly on paper and doesn’t leave campaign headquarters. I did some work in NV, but it wasn't data work, just things I did because I was working closely with the folks who also organized trips of volunteers to NV. I was able to find data work enough to keep me busy 3-4 days a week all summer, and from a professional standpoint, actually working with campaign data in its raw form, and with the data collection software doing data entry would be valuable for any future work in the sorts of jobs I applied for. I’m among other things very fast and accurate at data entry, and rather enjoy it in the way many folks enjoy playing Tetris or solitaire.
Meanwhile HRC’s campaign was doing everything campaigns were supposed to do. Strong issue positions, good convention, good debates, mobilizing a monster volunteer force and executing with them during GOTV. Polls showed a stubborn if small lead over Trump, even in the worst cycles and his paths to victory looked very slim.
The last 3 weeks of the election I was a data captain in a phonebank, where we ramped up from a twice-a-week, 6ish callers to doing over 300 volunteer shifts in the last 4 days of the election. Doing that engaged my “conference room pilot” skills, as a lot of the job is teaching volunteers how to interact with the software that connects them to voters. I know for a fact that from the time I switched to plan B, being a volunteer I did everything possible to elect HRC and defeat Trump, within my power. There are no lines of communication from that level of worker to those who do the analytics in a campaign. We get instructions on who to call, or where to walk/knock or where to set up for voter reg, and we execute — we are measured on volunteers who do shifts, and how many voters they contact or registrations collected.
I did pay close attention to the criticism of Nate’s site and wasn’t especially impressed with the trendline logic, thought the connected states idea was interesting but “similar” seemed overly broad the way it was applied, but the one that I did find convincing was the undecided vote, especially after someone brought to my attention that in 2004 they’d mostly broken for Bush.
So I spent some time looking for what everyone thought those voters might be. One of the major data misses in this election was that analysis. Essentially people thought that with a historically awful candidate (say what you will about HRC, she did all the stuff a good campaign does, and her favorables always exceeded Trumps), some of the undecided would be R voters that would go third party or stay home, or even cross-over. Well, they were R voters, but they all in the end decided on Trump. Even so, and even with HRC not polling above 50%, enough of the undecideds did break for HRC that she likely would have won except for the other big analytics miss. So I looked at the undecideds and still didn’t see how Trump would win enough states, even if they all behaved pretty much as they did.
This lead me to the conclusion that a narrow win was possible, but equally likely was a real blowout, if we got some crossover votes or even if any of them stayed home or did 3rd party protest votes. Which was conventional thinking, and I didn’t have any real anxiety going into the final night, indeed I was a calming presence at the phone banks with the more jittery volunteers.
But — the traditional GOTV combined with targeting our more unlikely and “invisible to poller” supporters seemed to be working. Early voting was strongly HRC and was exactly the precincts and demographics we’d worked so hard to reach in the GOTV efforts. We blew away records in the strongest Democratic counties in most of the states we lost.
But the rural conservative voters, who had been voting something like 70-30 vs Obama went 85-15ish vs HRC, with also record breaking turnout. Their women didn't vote for HRC or stay home, they voted with their men.
As the returns came in from NC, FL and Virginia I got physically ill, while still working last phone bank and trying to keep positive vor the other volunteers as we wrapped it up and started breaking down equipment. I often describe my primary talent as a data person as an instinct or feeling when something is off, and then, if I have access to the systems, I can often tease out the root cause. I’ve never had a reaction like this before, it wasn’t intellectual. Literally all my fight/flight stuff went off and I felt nauseous. Because the rural returns could change the equation in a way that the undecideds could not, and my data instincts said “this matters”.
It still wouldn’t matter if it was confined to the Atlantic Seaboard. So my intellectual brain tried to keep my emotional response under control long enough to get the computers secured, get home, warn my wife, cuddle with my dog and wait. But by then I knew the same thing was happening in the Midwest, although not so much in GA, TX, AZ and NV (and eventually CO, but that didn’t come out until later). I was pretty sure once it was being reported that she was struggling in her +5-10 margin states that Trump was the odds on favorite to win. A night like this, she wasn’t going to flip GA or TX, and while AZ was possible, it didn’t have enough EVs to cancel out losing a midwestern state, much less the two she eventually lost.
I felt like I did when my first dog was dying. Literally the same physical response, and an inability to distract myself in any way (all activities seemed pointless). So I just had to endure it until my body calmed down, I got some sleep and what I did during that period was just think about might have beens. I can’t imagine how someone with leverage to really make changes in how HRC ran her campaign must feel now. I personally would feel responsible for all the bad things that are going to happen now, as they go forward. If I was HRC, I’d be shattered. Two of the best people I’ve seen in my lifetime, her and Obama, get to watch our country destroy itself, even though they had the best chance to change its course.
So WTF happened? Well one thing that Bush learned in 2004 was that you could target your message to different demographics, and the key to getting them to actually vote was to engage their anger and fear. The thing is though, everyone thought that you had to actually have a campaign organize and target that message for it to work. Bush did something like what Trump did in the Kerry campaign, but on a much smaller scale. Enough to win, but not enough to shift the margins as far as Trump did.
So it turns out that ceding all media in rural areas to Fox and right win talk radio can be really fucking dangerous. Romney and McCain didn’t have a message that triggered those microtargeted anger/fear emotions that get people to self-motivate to vote. Trump did. Exit polls show universally three things about Trump’s support.
1. They feel they’re worse off than they were 8 years ago
2. They feel the future will be even worse
3. Therefore they need a change.
So first, economics — Yes, actually poor people voted for HRC. But the household that went from 100k/year to 70k/year seems to be voting for Trump. I don’t think that #1 in the above is fake, I suspect that’s actually hard reality. Yes, their “poor” current state is richer than the base of the Democratic party, but it’s still less than they had, and because #2 (RW media) gives them no hope of any improvement with the status quo and constantly insists things will get worse (so they’ll get another economic shock, any day now). They have a factual basis for their anxiety, but are in a media-based reality, not a fact-based reality, on what their future prospects hold or what is likely to improve them.
Second, social change — The communities that came out to vote are racially, culturally and religiously the same, at least within the circle these individuals know. People different from them seem extremely threatening, and again, RW media makes natural discomfort with “Other” much, much stronger. Yes, I do think most of them are culturally racist, sexist xenophobes. But without the external validation and reinforcement, normalization of these responses they’d be far more likely to react with mere discomfort, rather than anger and fear sufficient to GOTV like this. Also it is really, really easy to blame the economic situation on the social change. It obviously fucking works as well in 2016 as it did in 1932 and 1980. The social change makes you uncomfortable, the economic situation makes you scared. Connecting the two is a natural human response.
Finally — likely voter models. These guys were absolutely invisible, the same way non-english speaking people tend to be, recent immigrants etc. The HRC campaign was all over finding their natural invisible supporters, outside of likely voter models. Their internals didn’t replicate the NV polling miss that happened in 2010, 2012, 2014, we got the NV electorate right on the nose. But the rarely-voting white rural voter? Nope. The campaign’s sin was that they knew their own supporters and worked damn hard to bring them out, assuming in a high turnout campaign, when we vote, we win. Honestly, the other side wasn’t even trying to consciously identify and reach their equivalent low propensity voters in any way a campaign could recognize (donations, volunteers, voter contacts). They just had a message that stoked anxiety/fear in a very, very large demographic and a megaphone to get that message to every ear that they didn’t have to pay a dime for.
So in the end, from a “Can I live with myself” standpoint, “did I leave it all on the table and Trump won anyway”. I put myself in the position of being able to help, and was not granted the opportunity to do so. With the opportunities I had? I enabled a lot of work to organize the army of volunteers that brought those nice early voting results and delivered Nevada. I loaded data from events, captured NDA’s in electronic form, set up user accounts, trained both data and phonebank volunteers and I know that my city turned out hundreds more shifts of GOTV and tens of thousands more voter contacts than we would have if I hadn’t been available (I was one of about 3 people that made the effort work, the others had different skills. None of us could have done it without the others).
Had I landed an analytics position there’s a decent chance I’d be facing the really bleak moment where I might have noticed something and got it to the right ears to course correct — and failed to do so because I trusted the expertise of more experienced hands, or I didn’t have the credibility to convince anyone that what I saw was meaningful. Alternately I might have missed it entirely too. (The big undecideds, yeah, I’d have brought that up repeatedly in meetings, unless in my state they were relatively small. The invisible rural voter? I'd need a “that’s odd” moment of insight when messing with raw data).
So I sit here today going WTF can I do with my life now that will help the most people? I thought that with HRC elected I could take a breath, and focus on some things in the rest of my life. But we failed.
I still have all those skills I wanted to offer HRC. I’m still able to relocate to any state in the union, or even to any foreign country that would have middle-aged me and my disabled wife. I’m even got the privilege of being a straight white man who looks 10 years younger than my age and would fit in almost anywhere in America without causing bad reactions, even those invisible rural communities, if I was careful to not make it obvious that I’m a liberal athiest. Also the financial resources to survive income that isn’t steady, although the likely destruction of the ACA by 2017 is a scary thought. Some of our household income and expenses will also be affected if Medicare or SSDI gets changed for the worst, but I can probably ride it out.
I’m in a position to help the vulnerable, or to contribute to some effort to change the current equation in this country, to somehow break the cycle of government failure-enables-anger-fear-enables-the-folks-who-broke-things. But I don’t know how.
I welcome any ideas. My body is recovering from the shock, but I’m lost as to what to do next.