Aim For Convexity: Longevity in Training...and Life


"I have been wrong about most things most of my life...tell me something I don't know... the key is to have a convex distribution to returns when I am right vs I am wrong" - @nntaleb #m4conference @Uni_of_Nicosia @nyutandon

That’s a tweet-quote from Taleb’s keynote at the M4 conference this year. I read that at the same time that I was following a discussion forum fight about “optimal” training.

There’s no such thing as optimal training. Or optimal nutrition. Or optimal anything. Anyone who’s selling you “optimal” is a charlatan. Why? Well, no one knows what they’re doing - life is too complex and there are too many interactions to know how to be “optimal”. Even if you could be optimal in one dimension (training, health, longevitiy,…) you can’t be optimal in every dimension at once.

Rather than trying (or fighting ad naseum on the internet) about being optimal, it’s better to set your life up to be convex to non-optimality. In the long run, you’ll succeed - and even if you’re not perfect, you won’t fail. Attempting to be optimal will ensure ruin.

What does setting yourself up to be convex mean?

Life is non-linear. If you position yourself right, sometimes doing the right thing has more benefits than the negatives for doing the wrong thing. If the response to an action curves upward (holds water, to use the Calc I description), that’s a convex response - and that’s where you want to be.

Everything in life has a response to an input. Usually we draw that as increasing your input to the right and beneficial responses to the top.

Let’s say you start at the arrow in the picture above. The x-axis is training effort and the y-axis is strength, health, fitness, etc - you pick what you care about right now. Going to the right: If you put a little effort in, you can do better. if you put even more in, you’ll do even better. Going to the left: doing less training, or sitting on the couch, etc you’ll start to detrain. But here’s the cool thing about the human body, you’ll detrain less worse than you gain from training. Over time, if you hover around this spot, you’ll gradually improve. (This is how traders turn volatility into income, but it works for just about everything.)

But, of course, everyone doesn’t want to wait. They want to be “optimal”. They want to do the most to improve the best. And if you look at that plot, you can keep going to the right, training harder and harder. Doing two-a-days six days a week. You’ll keep getting better and better, right?


I didn’t draw the whole function before - it really looks like this. Life may be convex, but it’s not infinite. At some point things go badly, and they go bad quickly. Generally, elite athletes attempt to hover around the peak. They need to get as good as possible as quickly as possible.

But if you go just a little too far, the wheels fall off the bus. No one knows a healthy elite athlete.

We understand this intuitively in some places - jumping off of a small box can be a hormetic stressor that makes you stronger, jumping off the roof of a building will kill you. In the context of training, you’ll get better by training more…until you go too far and crash.

The trouble is that we don’t know where that point is - and it’s different for everyone and changes all the time depending on the context of your life.

So, to avoid ruin, we try to stay in the green region. We do light and easy training consistently over a long time. We may not be doing the most every day. You may not even feel like you’re working out hard. You definitely shouldn’t be tired, but since we’re staying in the convex region, we’ll be improving.


If we train too hard - everything will fall apart and we may not be able to recover. This is the place where injury can take you out. For example, chronic glycolytic over-training can cause mitochondrial damage or adrenal fatigue.

So, what if we really do know everything about training - like those internet marketers like to make you think they do? It’s it more efficient to be optimal? You’ll be the best you you can be, right?

You’re not an elite athlete. Performance isn’t the only thing that’s important in your life. I’m sure you care about being healthy and injury free.

I only drew the convex function in one dimension. The trouble is that life is multidimensional and changes in one dimension may have an effect elsewhere. Think about training volume as one dimension and health on another. If you’re perfectly healthy, you can train more than if something’s not going great. That is, being sick might pull the function to the left a bit.

Or maybe it’s the other way? Training heavy may pull the health function to the right a bit.

Think about that, maybe you figured out the most bestest way to ever train - but the act of training optimally will make you more fragile in the health dimension.

And what about fractal dimensions? The x-axis could be a generic “training volume” or it could be deadlift volume and bench press volume, and snatch volume, etc. Being optimal in one of those may impact the other.

We really don’t know what all of the interactions are, and that’s the whole point. Since we don’t (can’t) know everything, it’s better for your longevity to stay away from “optimal” because getting it wrong up there is so much worse than getting it wrong in the convex region.

Michael Deskevich