In the last essay, which I highly suggest reading before this, I mentioned how University Challenge is widely known as the world’s hardest quiz. I may as well tell you now how you should go about trying to win University Challenge. 

Sure, the quiz is insanely hard, but here’s the thing: the questions themselves are not actually that hard. The quiz appears to be so hard because the questions appear to be random facts that you could only know about a topic if it’s your life’s work. And indeed, the questions appear to induce knowledge flows that are actually quite far away from each other. So, it appears that people answering multiple questions on the show must have insane brains that have several PhD-level careers!

Take a look at this diagram of knowledge structure within energy infrastructure. Most people think you need to know this entire knowledge stack to be able to answer a question about a fact at the bottom of the knowledge structure tree.

Here’s a better representation of the knowledge structure and what University Challenge is asking you for when asking such random and difficult questions:

Ok. So that’s how the questions are structured, but this doesn’t tell me how to answer them? Well, here’s the thing. If you’ve ever been lucky enough to get a question in your field, you’ll realise the questions are actually pretty simple. Something you might learn in your first year in university studying that topic. 

Here’s how you can win the whole game. Knowledge is all connected. If you break down the entirety of human knowledge about the world, you realise that there are only a few sets of generalisable principles from which nearly everything else stems. 

Interesting side note, a guy I know called Adam Bly is doing just this with his start-up, System. He is extremely smart, and when I had questions for a Covid project about the complexity of mapping Wikipedia networks, I went to him. He was also the person who created Spotify’s data engine which tells Spotify, with creepy accuracy, if you’re sad in bed, or frustrated driving to work. This system is the most sophisticated ad-targeting system that exists, a topic for a future essay.

Back to University Challenge. There are four people on a team; so, you can take that entire distance of knowledge about the world and reduce it into a handful of generalisable principles before dividing it up amongst four people, which is why you usually have someone from the arts, music, science and so forth. As a team, you could nearly cover all of the generalisable principles of knowledge.

This is the first of two tactics: the top-down strategy. At this stage, a single winning team can adequately understand (but not know) most of the ways in which the world and the knowledge therein functions. So how are these human geniuses able to answer the random facts that come their way?

First, they are not. Most of the time, these “facts” are actually not facts but “topics”, reducing the knowledge “distance” significantly. They just seem like they are “facts”, or highly specialised subsets of information, because you don’t know them! Secondly, there is another tactic used to win the game. And this is also the tactic used, knowingly or not, by the internet, Substack, and podcast celebrities whose fame is predicated on them being such unbelievable geniuses.

The 80% Rule

By now I’ve said it enough times: most of the people that you think are geniuses are not. They simply know different stuff from you. However, the people who make careers out of being supposed geniuses happen to be very systematic in the knowledge they acquire, and how it relates to other knowledge they (or their audiences) have.

Things brings me to University Challenge (and Substack writer specialty) tactic two: the bottom-up strategy. As my high school friend used to mention the history of West African politics in passing, telling people random facts makes them think that you know the entire knowledge structure for that fact. In fact, it goes further and makes us think “what the hell else do they know?” Thankfully, it does not mean this. It’s much more simple than that.

There are just some facts that are easy to learn, remember and recount that make you sound incredibly smart. It takes a lifetime to learn the entire knowledge stack. It takes as little as hours to learn the 80 per cent of the knowledge structure that makes you sound smart.

Imagine we were having dinner, and I tasted the wine and said “Not bad! Not quite a 1945 or a 1959 Romanee-Conti, but definitely not a 1983!”. You would be inclined to think: “Wow. She’s a wine connoisseur.” Now imagine we had dinner again and I said the same thing. You’d think, wow. She knows very little.

Then imagine we were having dessert and I told you about a recent trip to London. “The art exhibition was great! A real Kandinksy-meets-Basquiat vibe.” Again, you’d be inclined to think she knows her art. I have no idea what wine or what art she’s talking about, but she’s very cultured!

Well, I know people who have weird conversations like this. And at first, I couldn’t understand how they were so damn knowledgeable about everything. But guess what: they’re not.

It is easy to get to know 80 per cent of what you need to know to work your way around any particular topic of knowledge. And I really mean that. I’ve done it with loads of them. The hard part is figuring out which 80 per cent. However, this is becoming easier because more and more people are taking their work online to map out the knowledge structure and openly say “Hey! Here’s the stuff you need to know, ignore everything else”.

For example, if you want to walk into a room full of investors and scientists and make it sound like you are up-to-date on all things biotech, perhaps even an expert, yet you’ve never heard of the field until today, give yourself a week and read this.

As it turns out, there are a small number of writers (or Substacks) that produce quite interesting and novel work. And then there are a huge number of people who make a name for themselves as being “geniuses” for reproducing such ideas under their own name. I can nearly always break down the latest “opinion piece” of certain thought-leaders (I won’t name and shame) into the underlying authors whose original papers I’ve read. It’s quite pathetic.

So, finally, to win University Challenge, the bottom-up strategy is to pick the main question topics (like art, biology, music, physics etc) and learn the names of top people and theories of each main movement within those domains. You may have about 200 facts to memorize, which will serve you well for the rest of your life.

Combine this with the top-down strategy of the team understanding how the world generally works, which means that when questions arise for which you don’t have a memorized fact, you can typically take a very well-educated guess straight to the winning line.

Boom.

A quick note on experts

Simply knowing and understanding the downwards knowledge flow within a domain does not make anybody an expert in such things. And this is crucially important to understanding why most people are not geniuses. Just because I understand and have done a lot of research and work in econometrics, I am not an expert in that space. Far from it!

This is because becoming an expert in such things requires more than just bottom-up and top-down strategies. And it is not, as many people think, superior intelligence. The secret is this: hard fucking work.

There are a lot of differences between my life and yours, however, there is one thing that is absolutely the same, whether you’re homeless or the richest person in the world. The number of hours in a day. To become the world’s foremost engineer for thermal protection systems (TPS) for spacecraft takes a really, really long time. As you progress, the knowledge that you acquire and apply becomes deeper and more narrow. Acquiring knowledge also becomes much more difficult. You are no longer playing University Challenge, but solving real-life, previously unsolved problems. Patience, intelligence, and grit are side dishes to the main course: time. It is simply not possible to learn the entirety of many fields in one lifetime.

And yet this is what we tell ourselves that Elon Musk has done. We frequently hear that he is the best aerodynamicist at SpaceX; the best manufacturing operations engineer at Tesla; the best neuroscientist at Neuralink. This is simply not possible. By now you should have realized that the very people who say he is the best aerodynamicist in the world are not experts in aerodynamics. They believe such nonsense because their knowledge is different to that of Musk.

In short, generalised information is not expertise. There is absolutely a need for both in the world, but it is vital to realize that they are indeed very different.

What About Artificial Intelligence?

So, now that we know that most human intelligence is predicated on measuring the difference between what I know and what someone else knows, I want to talk in part three of this series about why we do the exact opposite when we think about artificial intelligence. Because for AI, we rate intelligence by measuring how close to a human it thinks.

In summary, this article shows that:

Perceived human intelligence = difference between knowledge of Human A and Human B

Where greater difference = perceived higher intelligence

In my next article, I will look at breaking down the following:

Perceived artificial intelligence = difference between knowledge of AI and Human

Where smaller difference = perceived higher intelligence