Scott Adams' Trump argument, and improving democracy

Scott Adams – the author of Dilbert – has written a series of astute blog posts explaining Trump's "seemingly inexplicable" popularity, which keeps catching analysts struggling to explain it. The series starts with Clown Genius, and follows with nearly each post since then. So far I have read one month's worth, though I have been reading two days.

I must say, Scott has me convinced. If Trump gets the Republican nomination, he is likely to win the US presidency. If not, he has promised to split the Republican vote, in which case the Democrats win.

You can read Scott's argument for yourself. It is an amusing read. If I were to summarize it, I would put it this way: Scott thinks Trump is the only person in the race who truly understands the mentality of the average voter, and possesses the skills to influence it. He has failed in his presidential attempts twice, but has improved each time. This time, no one else has his skill, and the chessboard is as laid out for him.

Scott thinks a majority of voters can be divided in roughly two blocs:
  1. One large bloc are rational voters. They vote for politicians based on their policies. They think it's the duty of each voter to educate themselves about political issues, and vote sensibly based on that. These voters don't understand how others won't educate themselves.
  2. Another large bloc are emotional voters. They vote for a politician based on how they make them feel. Many of us who pride ourselves on being rational actually also vote emotionally.
Based on his own supposed experience as a hypnotist, Scott's observation is that Trump is a master persuader who knows exactly how to capture the emotional votes on both sides, with effectiveness such that he's going to cash in a landslide win.

Scott is placing a high confidence on this claim, but is betting no money on it. His is a prophet's bet. If Trump wins, against the forecasts of other "experts", Scott gets the spotlight as one who confidently predicted it. Maybe he writes a bestseller or two. And if he's wrong – what did you expect? He's just an entertainer, and he puts a big disclaimer on everything.

But I think Scott's reasoning is correct. Trump is likely to win. At the very least, he has the skill to capture and rally a large proportion of emotional voters. Maybe you are an emotional voter, and are currently passionately against him. But that's just because he isn't yet rallying you. Currently, he's busy capturing the right. When he comes around to the center, chances are he will persuade you as well. It's his skill.

It seems to me the real questions are:
  1. Is the Republican party going to nominate him? They have a choice in it. They could self-destruct, just so that Trump doesn't win. By promising to run independent if he doesn't get the nomination, Trump is playing the dictator game. It may seem self-harming to not cooperate. But long-term, refusing to reward extortion is the only way to disincentivize it.
  2. Is the emotional voting bloc large enough? My intuition says it is. But on the other hand, Sanders is the anti-thesis of Trump, and has also had surging support behind him.
If it comes down to Sanders vs. Trump, it will be an interesting test of worldviews. Based on all I have heard about him, Sanders represents everything sensible. More than any other politician in the race, he is an honest person, not in the pocket of any special interests, and on the reasonable side about everything. As far as a rational candidate goes, he seems ideal. But, he does not have The Donald's showmanship.

If it comes down to Mr. Rationality vs. Mr. Showmanship, we may find out what the verdict for democracy is. If Sanders wins, it will be clear that the average voter is more sensible than I give them credit. It would bolster my faith in humanity. But if Trump wins that fight – at least some of the folks that are currently in denial might agree that democracy needs an update.

Graph-based democracy

In my opinion, we need to redesign democracy to not only count everyone's vote; but to do so in a way that accounts for each individual's (non-) expertise. I currently know of only one design capable of this.

In graph-based democracy, instead of voting on issues directly, everyone makes a list of people who's judgment they trust. These lists make a graph, and this graph can be analyzed with an algorithm such as PageRank. Out of this graph, we extract 1% of the most central nodes, representing people who's judgment is most trusted by others. In the US, this might be around 2 million people. This would probably include most celebrities, but those would be a small part of the whole.

It is these people, then, who would vote about actual issues. In this way, no one is disenfranchised – everyone's list of people they trust contributes to making the graph. But the average person's shortcomings are kept out of the political process, and the system is not easily gamed.

I've been trying to convince folks this is needed for the last 15 years. Unfortunately, few seem to understand the need. Folks do not seem to agree how hopeless the average voter is.

Perhaps I will be proven wrong about the average. Perhaps rationality ends up trumping showmanship. But if not – I hope awareness of the need to improve democracy will greatly expand if Trump wins.

Unless, of course, you think it's okay for a master of persuasion to just play everyone for a fool, and win the presidency.

Common questions and concerns

#1: "The people who I see as the most likely to end up in the top 1% are TV pundits, radio hosts, various con artists like motivational speakers and pastors of megachurches, existing politicians, and of course - people like Trump."

Yes, of course. All of them might be in there. But it doesn't matter, because there aren't 2 million of them. These self-promoters would now be 1% out of the 2 million most competent people that get to actually vote, and this is a group that's much harder to fool.

#2: "What makes you think that people who don't care to learn about politics and instead vote based on emotions would select someone informed and rational to cast the vote for them?"

The Dunning-Kruger effect tells us we can't know how much better people are than us, but we can tell who is worse. If we rate voters' informedness on a scale of 1-10, a voter at rank 1 will have a hard time distinguishing between 2 and 9, but he will be able to make a list of people that are at least 1-2+, as far as he knows. Those people, in turn, will make a list that is at least 2-3+, and so on. Note also that most voters are rank 5, not rank 1. They will recommend 5-6+.

We are all idiots of some rank; but we can recognize an idiot bigger than ourselves. People won't include on their lists people that they know are bigger idiots, so the lists will be biased towards competence.

Follow these links long enough, rank them with an algorithm, and you discover the most competent people that should be top rank, or close. Better than rank 5 average, which is how flat democracy votes.

Flat democracy relies on us selecting out of our midst someone competent. The Dunning-Kruger effect tells us this cannot be done. What we can do, however, is know who's stupider than us. Graph-based democracy is how we harness this knowledge.

#3: But how do we ensure a fair graph analysis?

The graph dataset would be public, and so would the algorithm. The algorithm would be agreed in advance, based on e.g. PageRank. Everyone could download the dataset, and check the results.

#4: How do we protect privacy?

There may be ways that the graph can be cryptographically anonymized, so that a boss cannot tell whether you've listed them; while still allowing each individual to verify that their contribution to the dataset has been counted. Intuitively, this seems doable.

#5: What if people just list everyone they know?

That is useful. If everyone maintains a list of everyone they know and respect; but excluding people they deem incompetent; then the lists will be biased toward competence. Following the lists long enough should find the close-to-most competent people.

Comments

Popular posts from this blog

"Unreachable" beauty standards

Is the internet ready for DMARC with p=reject?

When monospace fonts aren't: The Unicode character width nightmare