Daniel Lemire's blog

, 1 min read

What the heck is interesting research?

I have long advocated that the world would be better off if more people did research. But what the heck is research? In simple terms, I know of two types of interesting research…

  • The engineering path: Pick an unsolved problem that has some importance. Propose a viable solution. For example, no computer has been able to beat a Go champion. Write such a program.
  • The scientific path: Try to explain something significant nobody has managed to explain.

For example, it seems that biological neurons can solve problems (such as vision) using very little energy compared to digital computers. Nobody really knows how this works.

There are a few qualifiers that are important in my (admittedly simplistic) description. You should pick a problem that has “some importance” and “explain something significant”.

These qualifiers cannot be safely omitted. Indeed, there are infinitely many unsolved problems, there are infinitely many unexplained facts. The design space in research is infinite. It is trivial to solve new problems and explain new facts… just make up easy problems and solve them.

People routinely object that I cannot discriminate against “unimportant” problems or “insignificant” questions as I cannot tell what is going to be important or significant in the future.

Except that research is not an abstract endeavour meant to “produce knowledge”. Research is a social process. If the output of your work does not change how a few people think… it was probably in vain.

Almost by definition, research is a wasteful activity bound to fail most of the time. Unsolved problems and unexplained facts remain for a reason. However if your work lacks ambition, you have lost the battle from day one.

Further reading: You and Your Research by Richard Hamming; That’s interesting by Murray Davis.