Research Practices

Personal notes and principles for doing research well, accumulated from experience and reading.

The Central Claim

> "You don't rise to the level of your ambition, but fall back to the level of your system."

Systems outlast motivation. The goal is building a research system that works even when drive is low, not optimizing for peak performance.

1. Pre-rigorous — you learn rules without intuition (following instructions)

2. Rigorous — you understand *why* the rules work (playing the game properly)

3. Post-rigorous — you know when to strategically break the rules and discover something new

Most research mistakes happen because someone skips phase 2 and tries to operate in phase 3.

1. Make it work

2. Make it right

3. Make it fast

In that order, always. Optimizing before correctness wastes time on the wrong thing.

Connected Papers — citation graph visualization
Consensus — search engine over research literature
Zotero — reference management
Sioyek — PDF viewer tuned for technical reading
Overleaf — LaTeX in browser
Paperpilot — AI paper explainer
Emergent Mind — AI research assistant with trending papers and knowledge graph

The loop that prevents drift:

> Check todo → Locate a feature or bug → Plan and test locally → Push and verify it doesn't break anything → Repeat

Good source control habits are non-negotiable for research projects with multiple experiments. CLI fluency makes this loop fast.

Obsidian Source: Notes / Good Research Practices - wiki/sources/obsidian/notes-good-research-practices.md