[kinkrtyavimoodh]

Isn't there an inevitable aspect of diminishing returns here? For example, you can't run Moore's Law forever since there are physical constraints to how close things on your chips can be.

It is obvious that any problem to which people have not turned their attention to probably has lots of low hanging fruits that will give you 'progress' very quickly and relatively easily.

[deathhand]

I would disagree. The IoT despite being the latest fad has true possibility to change the world. We don't see this change in English speaking countries though. Why not? We are comfortable. If you go to an advanced Asian country they are just now gaining traction into the world and will leap frog this innovation. (Think Europe in the 1890s-1935.)

[tedsanders]

I think there are reasons to expect diminishing returns in the very long run (such as exhausting the search space), but they are not inevitable, at least in the short run.

If you model technological invention as combinations of more fundamental technologies, then as your stock of technologies grows, the number of new combinations should grow exponentially.

The exponential growth predicted by this model is consistent with the increasing rates of growth from the 1700s through the 1900s.

That said, I agree it seems that technological innovation (or at least its impact) is slowing. The difference in life between 1870 and 1920 seems to me to be much bigger than the difference in life between 1970 and 2020. In our fifty years, we made tremendous progress in information technology. In their fifty years, they made tremendous progress with electricity, engines, plumbing, chemicals, & radio.

Peter Thiel's emphasis is that this stagnation is coming from overregulation. My own emphasis is that we figured out what seem to be the main laws of physics by the 1930s, developed the Standard Model in the 60s and 70s, and every fundamental physics discovery since then has been an evolutionary tweak, not a disruptive re-architecting. (I'm overgeneralizing, of course, but I think the idea stands.)

[jobigoud]

Eh, Moore Law is about the number of transistors per chip, if you can't go closer you can make bigger chips