To me science is about defining the bounds of the possible. To do that you also need to push the bounds and occasionally update everything you thought you knew. In the case of CS where we find new, lower bounds, I find that especially exciting.

I meant to append this to my original reply but I’ll say here: I enjoyed Knowledge Based AI from Georgia Tech. The lectures tallied about different types of knowledge acquisition as it applies to humans and them how we’ve mapped that to machines. That along with HCI were my favorite OMSCS courses.

In your case, seeing a new, lower bounds helped push you to look in different directions and re-examine your assumptions. In KBAI we weren’t allowed to share code (of course) but were given very wide leniency in what we could share. Posing my performance numbers and reading other’s gave me a sense of what’s possible and had a similar effect as your story.