At no point in the 2020 Presidential Election process were things uncertain or unclear, and at no point were things too close to call (or close in almost any sense). But we’re convinced (I have a hard time avoiding this confusion myself) otherwise.
Today, we’re going to talk about confusion generally and misinterpreting information specifically. Then I’ll point to one way to solve our problems. Okay – some of them. We have quite a few competing for our attention right now. This one is a technical problem, though, and not a philosophical one, so it will be easy to solve in a bipartisan manner.
First, the problem. People this week – Election week – have been confused. They’ve been watching the news, following Twitter, and checking electoral projection maps constantly. As a result, they’ve gotten mistaken impressions along the way. They’re reading the results as they’re being announced as a horse race – watching this one pull head here, then that one pull ahead there – then a change in this direction, and again a change in that direction. It is more exciting than if everything were all clear at the same moment, and then, we don’t have a system that allows for this, either. So people were confused, and changing numbers added to their confusion. When things become clearer, their confusion doubled rather than decreased – because they thought they understood where things stood yesterday, and today looks different than yesterday.
Here’s what caused the problem.
When the polls closed on Tuesday and we began counting votes in 50 states + Washington, D.C., the result at that point was determined. We just didn’t know it. The winner was decided, and we had no idea who had been selected. To get that information, we had to get the voting tallies. And then we were strung along. News agencies reported the numbers as they got them: Trump ahead in Florida, Biden ahead elsewhere. Trump ahead in Pennsylvania – now Biden ahead in that commonwealth. Things appeared to shift and change. Doubt was introduced (largely by Trump). The thing is, there was room for doubt. People are not good reasoners. They didn’t think that the result was already determined and that we were revealing more and more of the result as the hours (and then days) wore on. They believed that a thing was true, and then the truth was somehow changing before their eyes. And believing this was not reaosnable, but almost everyone fell for it, whether they were supporting Trump or Biden. Most people were watching the maps change and interpretaitng the reporting as an actual change in result, which was not what was happening.
The results weren’t changing – only our understanding of the result.
Let’s imagine a simpler (by comparison) way of putting things.
Imagine that we run an experiment. I put 1,000 red and blue Skittles in an opaque box on a table. The candies are split at a ratio of 45 to 55, so there are 450 red ones and 550 blue ones. I invite a nonpartisan third party to come count the skittles, asking them to separate and to count them according to color. Before the person even begins counting, the result and tallies are already certain in so far as they are true facts about the world. There are more blue skittles than red ones. 100 more of them, in fact – it’s not even particularly close. But let’s imagine the person has to count them one at a time. They begin drawing the Skittles out of the box and tallying them.
Incidentally, the way I dumped the candies into the container was a bit skewed in that they didn’t land at the bottom of the container randomly assorted. Most of the blue Skittles are at the bottom of the box because I dumped those first, and some red and blue mixed together as I dumped the red ones in second. The nonpartisan counter begins counting Skittles. A pattern appears to emerge…
Let’s say we stop counting at 250 Skittles, a quarter of the way through the full amount. Red is outpacing blue at a rate of nearly 2 to 1! The nonpartisan counter, unknowingly offering fodder for a misinformation campaign, announces that, “Red is so far in the lead!” This announcement makes it seem as though things aren’t yet determined, but they’re being determined as we count, and furthermore supports the false impression that red is in the lead in the final count – perhaps even expected to win. But because things are already determined, and red is in fact outnumbered by blue, the pattern cannot hold. This is because the sample pulled at the start (or at any given time during this count) is not representative of the whole box of candies. At no point is blue or read “ahead” in the sense that a person hearing the announcement can’t help but to take it. In point of fact, since I dumped candies into the opaque box, there are and have always been 100 more blue than red candies in that box. However… if we announce the count at 20%, 40%, 60%, and 80%, it will seem as though the reality being reported is changing – when in actuality, it is the reporting that is changing, and we are only becoming more clear about the actual blue-to-red ratio of Skittles.
This whole process is (predictably) confusing to a person hearing the updates along the way. It seems as though a pattern were established early, and they made some assumptions in predicting that the pattern would continue – and they’d have been right had the sample somehow been representative (which is what good polling does).
This explains people’s confusion about the vote tallies this week.
A similar failure in reasoning has occurred repeatedly throughout the pandemic as new information and improved understandings have become available and have been reported more or less effectively. At first, we didn’t know how the virus worked, how long it lived on surfaces, how strong a viral load would be needed to transmit the virus (and whether a stronger viral load would produce more or less severe symptoms), what conditions were ideal for propagating or halting the spread of the virus, and we didn’t have much an idea of how to treat it (e.g. we now intubate later than we did at first and in so doing increase survivability of the virus’s effects). And we’re learning a lot about transmission, survivability, and treatment of the virus. We’re improving how we help people suffering from it (one reason we want to slow down the flow of people into hospitals during the peaks of waves of COVID spread). But some people observed the changing understanding of the virus – our revised hypotheses in light of collected data – and reacted with suspicion. “First they told us not to wear masks – now they tell us we have to wear them!” Most reasoners aren’t reasoning well, here, obviously.
A third example comes from my own family, but I’ll bet you’ve got a card like my grandmother in your family, too. My grandmother’s understanding of science is confused in the same way many people’s understandings of virology and epidemiology has been confused through 2020. Nana will say things like, “First they say one thing, then they say the opposite – they must know nothing at all! I was supposed to eat butter, now I’m supposed to avoid it. That’s crazy. I’m just going to eat whatever the hell I want to eat.” The confusion there is grounded in a misunderstanding of our how we add to our understanding of the world we inhabit, which is done by reinforcing or correcting, falsifying or confirming, our hypotheses. In truth, we adopt and live by hypotheses that we’re always testing and revising, and the smart play is to accept new hypotheses when the changes we hear about are supported by good data. We find the good data by listening to trusted sources (read: not Dr. Oz). What my nana wants is for the first hypothesis she hears to be the correct one. Instead of being a source of comfort for her, any revision of an hypothesis is a source of angst and doubt. I’ve heard something like this called the pessimistic induction before in philosophy of science. (The pessimistic induction holds that because we’ve been wrong before, we’re probably wrong now. It’s definitely not identical to grandma’s position – but it’s not far off, I think.)
This inability to incorporate new information into our webs of belief (thanks W.V. Quine) is a result of forming webs of understanding and belief that are not elastic enough to deal with the nuances and changes in information available in a real, messy world of facts. The truth about the world is that it is messy, and to manage in the world, we have to adopt (provisionally only) hypotheses about what is true. Then we have to be willing to modify our web of understanding in light of dissonance between what we understand about the world and whatever new data we’re met with.
But we are not elastic.
So we’ve got to be real. And the confusion this week has led some in the country to believe that the election is being stolen right in front of them. How could they not be upset at this understanding? I would be, too, if I believed it. So how do we change the way we count and announce vote tallies to avoid creating the conditions that are inviting of existential dread and potentially violent upheaval?
We can start with a modification to when we announce the races.
One of the best things about how we work through the democratic elements in our society (how we conduct votes) is that the system is decentralized. It’s a lot tougher to steal or to influence an election when there’s no one place to go to mess with things. We do not currently have a federal election system – we have 51 individual election systems. And along with the feature of preventing fraud comes a bug. Namely, that the fact we have 51 different systems means that the results are reported differently, recorded differently, announced by the states differently, etc. for those 50 states. It happens at different times and under different circumstances. Then we collect all of the information at the end of this long process and announce it to voters. Except that we don’t just do that. We also announce at every (apparent but not real) twist and turn, every shift in (unreal) momentum, etc.. So here’s what the federal government can and should do. We need some regulations that are federally mandated in our federal elections. Specifically, imagine we hadn’t announced the presidential race until every ballot for a state were tabulated and checked. And imagine further that all of them were announced in succession or simultaneously on the same day. We could make a patriotic and unifying event of it, even! And there’d be no confusion about an apparent change in the trends of vote tabulations. There’s no room for illogic in this modified approach. At least, there’s a lot less room for it. We may hereby prevent upheaval, property damage, and danger to actual human life.
Creating less doubt about the process will strengthen the integrity of our system. Let’s get on with it.