"Citations don't matter", says the one with gazillions of citations. "We believe in slow science", say the groups who publish one paper every week. Like it or not, the system evaluates researchers with citations. Here are practical steps for junior folks 🧵

Three key ideas: 1⃣ Make your work accessible 2⃣ Make your work memorable 3⃣ Make your work usable Let's break them down.

*How do you make your work more accessible?* 👉 Disseminate your work through dynamic, engaging media rather than relying solely on a static, boring PDF. But HOW? Make a website, write a social media thread, write a blog post, create a video explainer.

*How to make your work memorable?* There are several actionable steps. The simplest one? Come up with a good name!

Needless to say, the paper title is incredibly important! Spend time designing a distinct, memorable title. Source: Good Citizen of Computer Vision ()

People memorize your work better when you can tell a clear story that *puts your work in the context of existing research*. Distill your core idea into one concise figure. Examples:

*How to make your work usable?* Your work is cited because it empowers others to do their work. Don't simply dump your messy codebase online. Put effort into polishing and organizing it so people can easily build upon your work.

To further increase your research impact, check out this post that I found particularly insightful: "Build what you need and use what you build." - @Michael_J_Black

Share the many facets of your work. To finish up your work, you typically need to - write up evaluation script - implement custom kernel - reproduce baselines - process messy data, clean up annotations - visualize results All these have values. Share generously to help others.

Of course, aiming to maximize citation counts is the WRONG goal. Focus instead on producing research that has real impact. Over time, your work will naturally be appreciated and recognized. Make them accessible, memorable, and useful!