michael_nielsen

There's a very useful idea for thinking about the history of research and discovery that I call "the velocity of discovery".

It begins with an analogy - actually, more than an analogy - to the motion of waves (e.g. water waves).

A very interesting thing about wave motion is that waves can move faster than the individual particles making up the waves.

Indeed, if you set things up just right the constituent particles may just move up and down, or even backward (compared to the wave), while the wave itself moves quickly forward.

Something very similar happens in research and discovery. You think of individual researchers as individual particles. It's possible for the most exciting research frontiers to move very rapidly round the research ecosystem - far faster than any researcher can move.

An example: suppose in 2004 you happened, by coincidence, to take university classes in linear algebra, AI (including neural nets), and GPUs. You didn't do this with any aforethought - at the time, these weren't necessarily even thought to be so closely connected.

No offense at all meant to my brilliant deep learning friends! Just pointing out that there is a lot of luck in these things.

Another example: someone may become famous doing some type of research. They look like a visionary, way ahead of the curve. But later they continue pursuing what others consider dead end ideas. But perhaps it just shows that they move slower than the general velocity of discovery