Good Research

Good research aims for three things: it should matter, be correct, and be easy for others to understand.

1. Pick problems with taste

• Work on questions that are important, not merely publishable.
• Optimize for impact, not minimum viable novelty.
• Look for problems where you have unusual insight or comparative advantage.
• Read widely so you know the field, then think independently so you are not trapped by its bad habits.
• Talk to people, find strong collaborators, and share ideas freely. Execution matters more than secrecy.

2. Use taste to find better ideas

• A strong paper usually starts from a problem that genuinely matters.
• Prefer projects that only you, or very few people, are well positioned to do.
• Notice areas where current work feels obviously confused, weak, or incomplete.
• Accept that luck matters. Good ideas often come from timing, exposure, and chance connections.

3. Execute ruthlessly

• Start with the part most likely to fail. De-risk early.
• Kill projects that do not work.
• Also kill projects that technically work but are unlikely to matter much.
• Re-prioritize when a much more important idea appears.
• Put in unreasonable effort on the parts that matter: experiments, controls, evidence, and skeptical checks.
• Keep one clear central idea. Every experiment should support it.

4. Make the paper feel complete

• A paper should answer the obvious questions a careful reader will have.
• Do the most natural extensions if they are essential.
• Leave only non-essential follow-up work for others.
• The reader should not finish thinking, “the obvious next test was missing.”

5. Write for humans

• A paper should communicate one main idea.
• Know who the reader is and write for that person.
• The abstract should state the topic, problem, method or result, and why it matters.
• The introduction should tell a story: where the reader starts, what world the paper lives in, and why the contribution matters.
• Figures should stand on their own with clear takeaways.
• The conclusion should answer “so what?” rather than repeat the abstract.

6. Write clearly, not theatrically

• Clarity matters more than style.
• Read your writing aloud or use text-to-speech to catch confusion.
• Be specific, avoid unnecessary jargon, and cut what does not help the message.
• Good writing is mostly readable writing.

7. After submission, luck still matters

• A good paper can be too early.
• A good topic can cool off before publication.
• Someone else may publish first.
• Awards depend on reviewers, committees, timing, and taste.

Because of that, do not aim primarily at awards. Aim to consistently produce work with a high-impact distribution. Awards are a noisy sample from that distribution.

Practical summary

• Choose important problems.
• Build strong collaborations.
• Read a lot, then think independently.
• Test risky assumptions first.
• Drop weak projects quickly.
• Focus the paper around one idea.
• Write so the reader can immediately see why the work matters.

Best paper awards are mostly downstream of doing important, rigorous, well-communicated work, plus timing and luck.