The US polls were wrong (again), but we still need them
Share article:
Share article:

The US polls were wrong (again), but we still need them

Tuesday, 17 November 2020
This is an opinion article by an external contributor. The views belong to the writer.

One thing that is becoming certain after almost every US election is how uncertain the pre-election polls are.

This year was not an exception. Having predicted a ‘blue-wave’ that would allegedly see Joe Biden win by a landslide, the pollsters got it wrong again. Not only Donald Trump got more votes than he did in 2016, but the margins in the states that Biden flipped into blue turned out to be nail-bitingly slim.

As to be expected, the electoral polling business is now soul searching: either saying that this year was unusual because of Covid-19, or that the Trump voter doesn’t like to confess to pollsters etc. In short, every time the polls get it wrong, they say ‘we didn’t know it this time, but we’ll know better next time.’ As if in reply to that, Nietzsche once wrote “It is too bad! Always the old story! When a man has finished building his house, he finds that he has learnt unaware something which he OUGHT absolutely to have known before he began to build. The eternal, fatal “Too late!”

The fact that polls were wrong seems to somehow spoil the mood even this year, which didn’t see a shocker defeat by the Democrats compared to what happened in 2016.

The New Republic writes “if the polling industry is as broken as it seems, then we simply don’t have a good sense about what Americans actually think — a serious problem for a democracy.” To a certain extent it is, because polling generally is a necessary way to gauge national sentiment on crucial issues facing society. But many pre-election polls  — especially in the US, where they are sometimes updated every single minute — serve to satisfy public impatience and keep it guessing while it waits for the election.

And that is not a trivial role for polls to play as far as democracy is concerned. In fact, isn’t it actually quite democratic NOT to know exactly what Americans think before they cast their vote? It is only in totalitarian societies that everyone knows all too well in advance who will be the winner (with or without fraud).

The suspense is a sign that democracy works. What is worrying however is the lack of suspense, such as hearing Donald Trump saying “we were getting ready to win this election. Frankly, we did win this election.” Trump’s refusal to concede the presidency was a predictable move (by some) even before the election, and such predictability is a bad sign for a democracy.

But, as pollsters continuously err as they did in the past elections, they risk being replaced by AI technology which increasingly fares much better in sensing the public mood. But we have to be careful what we wish for.

Ecology of ignorance

Ignorance of the future is not a curse, but a necessary evil says German sociologist Niklas Luhmann. In fact, the task for a democracy is to be comfortable with ignorance, as “the other side of knowledge”. Luhmann calls it ‘the ecology of ignorance.’

Societies create institutions and organizations not to provide them with certainties about the future, but rather to absorb their uncertainties. Their aim is not goal-realizing but rather goal-seeking.

According to Luhmann the fact “that the future is unknowable is expressed in the present as communication. Society is irritated but has only one way to react to its irritation, in its own manner of operations: communication.”

Some, more autocratic systems choose the goal-realizing path of assuming control over communications and public opinion, as they claim to already know what is good for the public. By contrast, democracies resist final answers and continuously seek for them through communication. This is why they are essentially ignorant.

But democracies are also generally more peaceful than totalitarian societies because the irritation and the energy wrought by the anxiety of the unknown future is dissolved into communication and not into other, more radical forms. Hence, the free press and free institutions — as goal-seeking instruments — are essential for sustaining peace. Churchill famously said “to jaw-jaw is always better than to war-war.”

Communication essentially means sustaining ignorance while seeking for knowledge and truth (but stopping short from actually attaining it). This is a lesson from Plato’s Republic, where Socrates kept on asking Athenian citizens and statemen what the notions of ‘good’ or ‘justice’ mean, but no one could give him a definite answer.

In a way, Socrates was the world’s first public opinion pollster. But even his wisdom did not lead to any definite conclusion about what the public actually thinks. It would have even been dangerous if he had provided us with final answers, because the goal-seeking methods through communication that democracies inherited would have ended right there.

Thankfully, that didn’t happen, and the wise Socrates eventually concluded “I know that I know nothing.” Ever since the Socratic paradox, sustained ignorance through continuous communication became an essential part of any true republic.

By means of communication ‘truth’ always remains somewhere in the middle and is not claimed by either of the extremes. Eventually, communication itself becomes the truth. “Only communication can communicate,” says Luhmann. Or, as Martin Heidegger famously said: it is not humans that speak, but “language speaks.” In other words, truth and all the right answers belong to communication itself, but not to us. Our only task is to continuously seek truth by ways of free communication.

A numbers game

But what Socrates didn’t have at his disposal is AI technology. Today we live in times when communication is no longer in oral or even written, but in digital form. This means that it can easily be quantifiable and computable with mere numbers (of likes, views, retweets, hashtags and so on).

After the Cambridge Analytica scandal we learned that with just 10 Facebook likes the AI machine could evaluate a person’s traits more accurately than their co-workers. And with 70 likes it could do better than that person’s close friends.

When it comes to pre-election polling business, AI is also increasingly doing better than pollsters and pundits. In 2016, a numerical study by the Brussels-based think tank Bruegel found that analyzing the #Brexit hashtag was a very reliable way to predict the outcome of the EU-UK referendum.

In fact, AI defied the pollsters and pundits with an almost diametrically opposite conclusion, and it won. Surprised by the finding, Bruegel conducted a second study on the Italian constitutional referendum the same year, and got the correct result once again.

This year some AI algorithms out-performed pollsters as well, some even correctly predicting the outcomes of all the 11 battleground US states.

With pollsters doing badly every year, and AI’s performance constantly improving, the job that Socrates started might soon be over. Perhaps the day will come when we will replace the complex aggregate (human) polling systems based on communication through emails and phone calls with a simpler but much more accurate social media analytics, and we will finally know the outcome of the election before it even occurs.

But that might not be a good day as far as ‘the ecology of ignorance’ is concerned. Pollsters should improve their methods of listening better to the public mood today in order to prevent AI from creeping into their territory tomorrow.

At least they should not err by as much as they did in the past two elections. After all, the world yielded to goal-line technology to correct the mistakes by human referees in football. The same could happen with the referees of public opinion.

Serghei Sadohin