Bottlenecks shift but they rarely vanish. When a new tool unblocks one constraint, it exposes the next one. Major discoveries often emerge in parallel because the prerequisites finally fall into place at the same time.
When a new tool arrives, we oftentimes say it removes a bottleneck. That is only half true. More often, it just moves the bottleneck upward.
For example, programmers who once spent weeks on a prototype can now ship one in hours because of AI coding assistants. But they are now bottlenecked on something else: deciding which prototype is worth building. There is no unit test for "Is this idea good?"
This pattern appears throughout the history of science.
Charles Darwin and Alfred Russel Wallace landed on the mechanism of natural selection at almost the same moment after geology proved the Earth was millions of years old.
Quantum computing was bottlenecked since von Neumann's day in the 1950s until personal computers made the concept of "computation" salient.
As fields get harder, bottlenecks also become less individual and more collective. Large Hadron Collider papers can have thousands of authors because no single person can master accelerator physics, detector physics, vacuum systems, inverse problems, and theory simultaneously.
At that point, progress depends not only on ideas but on institutions that allocate capital, credit, time, and freedom well.
That is why the next great era of science may come less from a single brilliant thought and more from better systems for many minds, tying us back to Doug's observation in Augmenting Human Intellect.