I often see this (gp's) sentient on HN and it is quite surprising to me given how people here specifically work in technology. Are we not intimately familiar with how one typically starts with small trial groups before we scale, so that we understand the effectiveness and safety? Or are we just "move fast and break things" and leave a mess in our wake with no one spending time to clean anything up. I guess that would explain enshitification.

But seriously, S-curves aren't just about how the middle part looks exponential. Both ends are slow. Slow to start and slow to end. When technology B replaces technology A it is (almost) always WORSE than technology A initially. The difference is that its theoretical maximum is higher. Because guess what? In the long run if you stack a bunch of S-curves together, you get an exponential curve. And this is how so much technology has improved, including transistors, solar, batteries, and so on. Sometimes this slow start can be on the order of decades! The Motorola DynaTAC 8000X (1983) wasn't "a failure," it was a step to the IBM Simon (1992) which was a step to the iPhone (2007).

So if you're exclusively chasing things that are better *now*, you won't make any progress. You have to invest.

https://en.wikipedia.org/wiki/Technology_readiness_level