https://www.youtube.com/watch?v=yJDv-zdhzMY
It's hard to pin down, but I'd define a fundamental advance as one that either (a) multiplies the number of addressable problems, or (b) constitutes a sudden and sizable nonlinear jump in capability over a very short period of time.
It does often result from the combination of existing things, but it's always a discontinuity. A fundamental innovation is when 1 + 1 + 1 = 1000, not 3. It's a big monumental win that unlocks whole sections of the "technology tree" that were not reachable at all before.
An example of (a) would be the final integration of all the components necessary for a CPU onto a single die. An example of (b) would be what the development of the alternating current transformer did for energy transmission over long distances.
The most dramatic examples are both. Examples of these would be the first radio transmission or the first rotary electrical generator. These were utterly unprecedented innovations that changed human life radically. They weren't linear at all.
You're right. I haven't seen many of these in my lifetime. I can think of some low-magnitude borderline cases but nothing Earth-shattering that has touched me directly. But if I'd lived from, say, 1900 to 1970, I'd have seen many in rapid succession.
You seem to be arguing that such things do not exist-- that the fitness-weighted combinatorial state space for technology is completely smooth and incrementally traversable. I think you should study a bit of combinatorics as it applies to areas like evolutionary information theory. Discontinuities in slope, fitness valleys between significantly different local maxima, and isolated peaks unreachable by incremental hill-climbing are common features of real-world fitness landscapes.
Arguably Windows 3.1 to anything in the DOS-based line ending with Windows Me is incremental, and Windows NT 3.1 to anything ending in the line of which Windows 8.1 is the most recent (until Windows 9 is released) iteration is incremental, but the shift from the DOS-based line to the NT based line when XP extended the NT line into the consumer space was not incremental.
So we will probably have self driving cars which incrementally change our transportation system until it's completely unrecognizable but no step will be particularly large.
This reminds me a lot of the micro/macro-evolution debate in biology. There are people who think macroevolution is just a lot of microevolution, and there are people who think macroevolution is sometimes something completely different -- that discontinuous leaps occur. I am of the latter camp there too. The Cambrian explosion was a sudden non-linear leap to give one example.
Going from horse to car is not the same as going from Model T to Honda Civic.