It's worth asking which problems fall into the set of those that can eventually benefit of exponential progress, and why. This reminds me of Taleb's distinction

It's worth asking which problems fall into the set of those that can eventually benefit of exponential progress, and why. This reminds me of Taleb's distinction between mediocristan and extremistan, or non-scalable randomness vs scalable randomness.