Survivorship bias

Um, uh, uh, Warren, um, We- Warren.

Wayne, are there different types of, uh, survivorship bias? I mean, some things-

Yes

... seem to, you know, y- you know, you can, you ...

Is the same true for an investment case? You can use the history to prove anything.

Yeah, look, I mean, you could always find some statistics to prove your point, and that's why, you know, you still need, I suppose you could say, the human touch in investing.

But, you know, survivorship bias, as you mentioned there, Stephen, this is a massive thing where, you know, we once again [laughs] as we talking, went back and studied past data when I was still working, and we said, "Right, we know the top 10 asset managers in South Africa.

We know their 10-year performance record.

I actually wanna go all the way back to World War II.

And I wanna talk about this thing called survivorship bias.

If you're not familiar with survivorship bias, Back in World War II, the military, they really wanted to figure out where to add armor to its planes.

And so what they did is they examined all of the planes that were returning from combat, and they mapped where the bullet holes were in the planes so that that way they can add more armor to where the planes were getting hit.

But this statistician, Abraham Wald, was his name.

It's a hole in the entire process, right? Because the bullet holes are shown to planes that took the damage and they survived.

But there were many more planes that did not make it back, that the bullet holes may have been somewhere else.

And so this is where survivorship bias starts to come in.

Yeah.

And then hearing about two people and they were fine, that is survivor bias.

Yes, that's right.

Yeah.

You know, you mentioned survivor bi- bias.

I think there...

I think it's not, it's not quite the same as survivor bias, but there is a phenomenon where, uh, across the course of the 20th century, some extraordinary advances were made in the healthcare for, for women and babies and ar- around pregnancy and the, and the postpartum period.

And those advances were so successful that, uh, that some of the diseases that they were invented to address have practically disapp- disappeared from our-

No worries.

It's a survivor bias, keep that in mind.

For each one Mecha Chameleon, there's like thousands of failed games that you don't see.

Because it was what? Millions.

The issue is that's not the critique that would be levied against this data set.

The critique, which the authors of the study even acknowledge, is the possibility of survivorship bias.

Essentially, those that survive longer in spite of elevated cholesterol It's because they're paying special attention to their health as a result of knowing that they are at elevated risk.

Now there's multiple reasons to think that would be true here.

The man himself, Napoleon Hill.

The man, the myth, and the survivorship bias.

Because his biography, it's a mess.

It does not survive archival scrutiny at all.

We're just reading Hill's personal hunches.

Hunches that happen to work for some people, but yes, hunches.

And that leads straight into the danger of survivorship bias.

Which is arguably the biggest flaw in the whole philosophy.

One, it's that meme.

You've seen that meme, that survivorship bias meme with the plane where all of the bullet holes on the plane, it's like all those red dots and you see the plane.

Basically where that comes from is in World War II when planes would come back.

The army would look at the planes and they would see, okay, where are planes most likely to get shot? All the planes that would come back.

And so then they would put armor on those parts.

But what they weren't realizing is that all of the planes that were actually shot down were shot somewhere critical where they weren't putting armor.

That's the issue of survivorship bias.

It's the same issue that we run into crypto.

It's, you know, it's interesting.

You've talked a lot, like survivorship bias is kind of one of your signature arguments there, right? It shows the winners and it hides the losers.

And it seems like there's a lot of like big brand moves, like people just want to copy each other.

When they see something work in the industry, right? Suddenly you have a whole bunch of copycats trying to do a very similar thing.

I do think right now there's...

I can't say there's an over-adoption of AI, but there is certainly, I think, a bloated excitement around it and what it can and will actually do.

I also think alongside of that, the naysayers are going to use confirmation bias, which is the evil sister of survivorship bias, to confirm their point of view.

So the minute there's a hallucination of, let's say, an AI drive-through voice, Well, that's...

It's higher RPEs.

And survivability bias or survivorship bias-

Mm-hmm

... means that the only ones that do make it to the end, by virtue of the fact they've made it to the end, have unbelievable progress.