The auspicious appearance of @jbhuang0604.bsky.social on Bluesky inspires me to look back at my notes on “how to get cool research ideas,” which I started jotting down way back in 2001, at the start of @belongielab.bsky.social at UC San Diego (1/18)
Comments
Log in with your Bluesky account to leave a comment
Not me! I think Serge is right about Győző. It's really fun to see a photo of the lab again, I haven't seen it in many years. My old desk is hidden behind the wall, it was just below the rightmost windows. Why didn't we play more pool during those days?
Also on the bottom left you can just barely see the arm of the (terrible) couch where some people used to nap. That and the pool table were defining features for me.
I definitely played pool a bit -- with Victor (Gyozo)! I heard the pool table was for robots, but that pre-dated me... After you graduated, we moved down to the Vision & Graphics lab, and then to the new building. We wished we had a terrible couch! Had to sleep under our desks before deadlines.
There’s no silver bullet, but these six prompts provide structure to the thought process (in the context of academia).
1. “Taco Bell” research
2. “Why don’t you just X?”
3. Magical thinking
4. “If it’s not worth doing, it’s not worth doing well”
5. Interdisciplinary research
6. Moonshots
To make a splash, you need to grab some ingredients from a different cuisine. A couple years into his PhD studies, @borrrrrrrrrris.bsky.social combined boosting and multiple instance learning in an online framework to solve a feature tracking problem. (4/18)
Back then, CV and ML were still separate fields, and with MILTrack, Boris discovered two great tastes that taste great together: https://bbabenko.github.io/miltrack.html (5/18)
2. “Why don’t you just X?” Imagine Reviewer 2 taking a quick glance at your abstract and smugly asking this question. Your related work section better anticipate a bunch of possible Xs. (6/18)
If the answer to this is that X honestly works pretty darn well, you can do your best at making the case that "this one goes to 11," but don't expect the paper to stand the test of time. (7/18)
Comments
1. “Taco Bell” research
2. “Why don’t you just X?”
3. Magical thinking
4. “If it’s not worth doing, it’s not worth doing well”
5. Interdisciplinary research
6. Moonshots
(2/18)
This isn’t a bad thing — indeed, it's a reasonable description of most of my group's papers — but such work is unlikely to make a splash. (3/18)
https://youtu.be/evUWersr7pc?si=QK2AqmU85TUZFa6h