Is this true? As someone who hasn't seriously studied the history of technology, it doesn't feel true to me.
Two reasons this feels untrue:
1. Before there were semiconductors, exponential improvements in technologies were common. It would be extraordinary if this is now false except when semiconductors were involved. The overwhelmingly likely mechanism for that would be if semiconductors caused such fast improvements in everything that non-semiconductor improvements are a rounding error, but
that seems extremely unlikely?
2. It seems easy to name an arbitrarily large number of technologies where exponential improvements aren't only or mostly due to semiconductors. To pick an example that most tech-adjacent people wouldn't think of, the (average) YoY improvements in car tires are pretty incredible. While advances in simulation have contributed to advances in tires, advances don't seem to mostly come from simulation?
One question I have is: why would someone think this is true?
@danluu I wonder how much of this is people using "exponential" to just mean "very fast"?
Like, I don't know anything about car tires, but I doubt that they are ten or a hundred times better than they were in 2000. Whereas phones and computers obviously are, and AI image generation is like infinity percent better than it was in 2000.
@profjaydaigle Yes, it's possible the author actually means "very fast" and not "exponential".
It's from someone who I wouldn't expect to do that from the other content of theirs I've seen, but I don't know them and don't read all of their stuff, so it's plausible that's wrong. But even then, I'm not sure the statement is true, e.g., are modern battery density improvements attributable to semiconductor advancements? That's been exponential for at least 100 years and it seems fairly fast?
Dan Luu
Jay Daigle
Jacob Christian Munch-Andersen