The "Curve of Knowledge"

As a student I was worried that I wouldn't be able to come up with new ideas, or discover the Next Cool Thing!... I think a lot of people worry about that.

Honestly: I still worry that I don't have another "amazing" idea in me...


To help ease that fear as a PhD student, I jotted down a concept I called the "Curve of Knowledge". It works like this:

1. Come up with new ideas or notions for projects as they occur to you naturally
2. Spend a little time vetting the idea, making sure it's not obviously flawed
3. Look through the literature, assuming somebody already had this idea decades ago
4. When you inevitably find that the idea has been done, see what the date on the paper says.

As you practice coming up with ideas and growing in your domain knowledge, you'll notice in Step 4 that this offset shrinks. You'll start naturally coming up with ideas that have only recently been executed, maybe even being only a few weeks old! e.g. I've had new project ideas develop when Kepler data is released, only to have somebody else publish the paper a month later, before I can execute it

You could negatively view that outcome as being "scooped"... but instead I saw (and still see) it as an indicator that I have arrived at the cutting edge!

You also will notice that Step 1 becomes easier as you go. Ideas will more naturally come up as you gain experience. Creativity and idea generation are skills that you can practice and improve upon! Especially in science, you can practice them in a principled manner and get quantitively better at creativity!

Bonus: Old but not New, Forgotten Ideas

One thing I didn't mention in the video: sometimes when you're doing literature searches on projects you'll find that this idea was first outlined or speculated about 60 years ago, but the data/technology didn't exist then to execute it, and nobody has written about it since! If this occurs, proceed with the idea immediately! I have repeatedly come to the conclusion that I have no truly original ideas, but am satisfied with having ideas that have been forgotten by the community or haven't been considered for a couple generations.

You get triple bonus points from me if your original ideas cite 50-, 60-, even 70-year old papers that pseudo-scooped you, but nothing from 0-20 years ago!