When 16 year old chess grandmaster Wei Yi defeated Bruzón Batista with a brilliant sequence involving two piece sacrifices and a precise follow-up with a non-forcing Queen move, it was quickly hailed as one of the greatest chess games of all time and an early candidate for best game of the 21st century.

The game is an example of where an interesting and speculative idea ended up working in real life. If you haven’t seen the game, I recommend watching the recap.

As humans, we’re naturally drawn to ideas that are “cool” or seem interesting. On one hand, this makes a lot of sense — ideas which hang together in an aesthetically pleasing way can often provide a valuable sense of where to look. Elegance can often be a proxy for simplicity or Occam’s Razor style arguments. On the other hand, it’s much more fun to win the chess game of the century than to just win in a normal, boring manner. Even though the boring wins still count as wins!

This is the same idea behind the Puskas award which awards extraordinary goals in football each year. Some goals are particularly beautiful or aesthetically pleasing, in spite of the fact that they all contribute the same amount of points to the scoreboard!

When Peter Svidler played against Wei Yi in the quarter-finals of the chess World Cup later that year he had an interesting strategy. The quote below is paraphrased from my memory of the post-match interview^[1]

He’s a player that loves exciting, tactical positions that can’t stand to be in boring positions. So I resolved to make the game as boring as possible in the hope that he would over-extend and press for a win.

Svidler went on to win the match.

I think the implication here is pretty interesting. Svidler is exploiting the fact Wei Yi’s enterprising style by purposefully going into a boring position. I think part of the reason this works against a player like Wei Yi is that despite the objective of chess being to win chess games, there’s a secondary objective which Wei Yi is optimising for which is win chess games in an interesting or exciting way. This manifests itself in two ways:

1. Believing interesting ideas more readily.
2. Over investing in interesting ideas relative to their actual probability.

In some instances, the interesting ideas work out! In the game against Bruzón, Wei Yi spent time calculating a beautiful sequence which ended up working and, as a result, played one of the best games of the century. The problem is, ideas that are speculative or beautiful don’t always work. The universe doesn’t care about our personal aesthetic sense and there’s an opportunity cost associated with calculating these beautiful, cool ideas and they can plausibly take time away from the real objective of forming true beliefs (or winning chess games.) If we’re optimising too hard for the secondary objective of win an interesting way rather than just win we might not win at all.

I catch myself all the time thinking about ideas, partly because they might be true or plausibly help explain something about the world, but partly also because they’re interesting or cool to think about. The point is not that cool ideas are more likely to be wrong, but I’m much more likely to invest time in trying to understand ideas that are cool, interesting or aesthetically pleasing.

Some examples I’ve identified in my own thinking below:

• Consciousness of future AI systems — There are many interesting reasons to believe future AI systems will be conscious. For one, it’s parsimonious and avoids drawing arbitrary lines between biological and silicon systems. However, when I really honestly examine my thought process for why I’ve spent so much time thinking about this idea, at least part of it is because it’s cool to think about.
• Many Worlds interpretation — again, I think there are many independent reasons to think MWI could be true. It’s parsimonious and takes the fundamental equations of physics seriously without positing an ad-hoc collapse. But it’s also really cool to think there are genuinely real versions of you in causally separate universes.
• Russellian Monism — I wrote a recent post outlining why I lean towards Russellian monism as a theory of mind. Again, there are lots of independent reasons to accept it, but one of the implications of the view is panpsychism which, I have to admit, is kind’ve cool to think about.
• Life in the universe — I think it’s likely that life exists elsewhere in the universe. Again, there are plenty of good reasons to believe this, but it would also be really cool.

The point about the views above is not that they’re necessarily wrong. Some of them like MWI or life in the universe are practically scientific mainstream. The point is that we have a fundamental bias as humans to spend more time thinking about views that would be cool if they were true. And there’s a potential bias here if my primary goal is forming true beliefs — I’m really optimising for the secondary sub-objective of form true beliefs about interesting ideas which might lead me to spend more time thinking about ideas which could be wrong. For example, it’s plausible that future AI systems can’t possibly be conscious, but if this turned out to be true I probably wouldn’t be very interested in thinking about it or investigating it.

Sometimes the universe does, indeed, turn out to be cool. For example, quantum mechanics is undeniably counter-intuitive and cool to think about. But we also need to be aware that the answers to the fundamental questions might turn out to have really boring and mundane answers. If the goal is truth-seeking we need to be careful if our goal is to form true beliefs to account for this bias in our thinking.

1. ^{^}

   I can’t find the original interview anywhere. If anyone can find the original interview or has a more accurate quote from Svidler please let me know!