Last week, my friend Jay Van Bavel posted this on LinkedIn:

I constantly see people post about the power of tiny gains (image on the left) showing that if simply improve 1% a day you’ll be 38X better in a year(!?!)

While this is technically true, it’s almost never how learning and performance actually change. So it acts as a form of misinformation from productivity gurus and growth hackers.

In fact, if anyone promises anything that will make you 38X better in a year you should instantly call bullshit. It’s impossible on it’s face.

For over 100 years, we have know that the normal learning curve has a rapid rise at the beginning followed by a period of slower learning and then eventually flattens (image on the right).

I wish we would be more honest with people about psychology and performance. Otherwise we set people up for failure and disappointment when they fail to get 38 times better at a task! They’ll be lucky if they get 38% better (3800% better is an unattainable fantasy).

My response:

I think that learning curve graph is also overly simplistic: improvement at anything is always the result of a bunch of smaller things coming together, making it a dynamic process: https://kstarr.com/improvement-is-dynamic/

The learning curve graph only applies to a very small section of skills, and of those skills, it doesn’t represent any person’s actual trajectory. I think this graph is just as misleading as Tiny Gains!

There is to graph to summarize improvements—not the Tiny Gains one, and not the learning curve one.

Every improvement is bumpy and uneven.

In one way, Jay is right: the learning curve graph (fast gains followed by a slow rise) is a better way to explain overall improvement: real life is not about constant compound interest.

And yet:

I’m not so sure. you should see the other comments below—apparently the learning curve is extremely robust across contexts. Your blog post only focuses on CrossFit, from what I can tell. But if you have alternative data showing it’s only in a few small contexts I’d love to see it.

Yes, I have data. It’s called real life.

I have data showing that one graph does not represent everything.

I have data showing that one smooth line sloping up does not represent all of humanity’s improvements.

Even decent graphs can only show averages. And your life is not average. Your life will never look like a smooth rise.