Why Stats Fail in Sports, and Why AI Will Fail Us

I should not even have to write this.

But I will.

Using stats to run a team in basketball or baseball does not work.

It does not work because players are not islands to themselves. Their individual stats are products of the players around them.

To give a quick example, Michael Jordan was a product of Scottie Pippin (and others). And visa versa.

If you take Michael Jordan out of that equation and place him somewhere else, the result will not be the same.

Teams particularly the Philadelphia 76ers think that you can manufacture a team by statistics.

They think you can take a James Harden from here, a Joel Embiid, from there a Tyrese Maxey from still over there and come up with a great team.

Well how did that work out for the past two years?

Not well.

Can you paint a masterpiece using paint by numbers?

Did Picasso paint that way?

Can you analyze all the great painters, look for commonalities and then derive the ideal painter and the manufacture him a la Frankenstein,

That’s what Dr. Frankenstein did. He assembled a monster out of parts.

He placed together parts from different people, and these parts had no organic connection to each other.

This is what the Philly Sixers have been doing for several years now.

And the results are in.

It doesn’t work.

What the Sixers have assembled is a collection of very good players who do not complement each other.

A good team fits together like Gleason and Carney, Abbott and Costello. You can separate them but the synergy goes far beyond their individual talents.

Statistics can’t account for this.

It can not because the very essence of statistics is to isolate items for study while removing other variables.

We want to know about Jordan. If we include Pippin, we are not knowing Jordan. We are knowing Jordan and Pippin.

The problem with that is that other teams aren’t hiring Jordan and Pippin; they just want Jordan.

But if a team did want Jordan and Pippin, they would then be isolated on just that combo.

They are still losing synergy between the Jordan-Pippin combo and the other members of the team.

And so it keeps going.

***

AI will fail.

AI will fail because inherently AI is based upon mathematical regularity, orderliness, linearity and of course algorithms.

The basis of this ultimately is the impregnated binary structure of the computer.

AI can not and will not take the intuitive loop to think outside what is provided to it.

Intuition is not based upon what is logical. Intuition is based upon what is not logical.

Any perceived dramatic insights achieved by AI will be rehashed options thought to be undiscovered but in reality cold meals from a time ago.

The bases of AI are algorithms and mass data.

Mass data if it is to only include statistical data is the other problem for reasons explained above.

The algorithm though is the main problem. It forces one into a stultifying and often labyrinthine maze of structured choices that force one to consider a before b, then b before c, rather than think about all at the same time.

Moreover, the algorithm forces you to choose one path or the other. But what if true intelligence chooses both pathways?

This maze of structured choices is overcome by raw computing speed, a feat humans do not ordinarily achieve.

That’s why we use computers. We set them up so that we can do things that we can’t do. But that doesn’t mean that computers can do everything better than us. One of the reasons we are so slow as human beings is because we do not use structured algorithms when we think. Our brains do not work as computers.

If our brains didn’t work as computers, we wouldn’t be able to have achieved the intuitive insights that have allowed us to progress.

Let’s look at Darwin’s theory of evolution. That seems like a logical starting place.

Could AI in any of it forms arrive at the theory of evolution? Would AI be able to churn out millions of conceptual models of life, test them out, then arrive at the correct one.

Looking at the problem, one would think: Sure! It seems logical.

Well, of course it seems logical. That’s because we already accept it.

But could it do it?

It can not because intuition requires a leap outside the logic of the day.

It requires that you bridge two events that currently have no recognized logical connection.

There is no logical connection between the arm of a man and the fin of a fish.

It is only Darwin’s – and now our theory – that unites the two.

Hence, AI will not add new insights, only an improved version of what we have now.

Sincerely,

Archer Crosley

Copyright 2023 Archer Crosley All Rights Reserved