English | 日本語
LAB TOOLS — RESEARCH LOG & THINKING LOG

Research Log & Thinking Log

Tools for recording — The purpose is shared; the method is yours
The purpose is shared; the method is yours. 目的は共通、手段は各自。

Why keep two kinds?

Research LogThinking Log
What to writeWhat you did (facts)What you thought (misalignments, questions, hypotheses)
FrequencyOne entry a dayEvery Monday (three lines is enough)
RoleReproducibility, progress, responsibility for the researchNot letting the seeds of questions slip away
Who reads itYour future self, the team, your supervisorYou, first of all

The Research Log is a record of "what you did." Experimental conditions, analysis settings, papers read, results obtained, what to do next. It is enough if, looking back later, you can immediately tell what you did. Once interviews or experiments begin, this becomes mandatory as part of your responsibility for the research.

The Thinking Log is a record of "misalignment." It differed from what you expected, you are not convinced by the explanation, something bothers you for some reason. Inquiry begins with a sense of misalignment, so unless you keep a record of it, no question is born. It is not a place to keep perfect information; it is a place to keep your own misalignments.

Shared rules

  1. Write in your own words. You may have AI summarize, but do not paste it as is. Use 50% as a base and edit it yourself
  2. Prioritize continuing. Do not strain. Three lines is enough for the Thinking Log
  3. Keep it in a searchable form. Make it findable by date, title, or keyword
  4. Decide where it lives. Notion / local Markdown / Obsidian / GitHub — any is fine, but pick one place and do not scatter

Members' examples (from the July 2026 Lab Forum)

Reference points for finding a form that suits you. There is no right answer.

A: Notion calendar + AI summary in the evening
Records by day on a Notion calendar, with titles written so the content is clear at a glance. On leaving for home, has Claude summarize the day, then corrects and adds to it and carries it over to the next day. Instructs the AI to include questions for the AI in the summary.

B: Notion weekly view + long titles
Uses a weekly calendar view and records the content by writing longer titles. Uses Notion AI to search the past.

C: Local Markdown, bullet points, three lines
Does not use Notion; manages everything in local Markdown files. Uses bullet points because there are several tasks a day. Values "keeping it going" and writes without straining. The Thinking Log is about three lines. Shares the files with Claude for reference.

D: Obsidian, weekly Thinking Log, in English
Keeps a Weekly Thinking Log in Obsidian. Runs a reading log for papers and textbooks and a rule-based Research Log. Values rewriting in one's own words and is also practicing recording in English. Since all data is local Markdown, migrating to another LLM in the future is not a problem.

E: Division of labor between Notion and GitHub
Records research concepts and trial and error in Notion week by week, and moves to GitHub once the code is settled. Has set rules for Claude on how to write commit logs, so that commit messages following the rules are generated automatically.

F: A Notion space per paper
Creates a Notion space for each paper and manages importance, priority, and status. Attaches the PDF and summarizes it in one's own words. Sometimes summarizes with AI, but tries to write in one's own sentences.

G: Separating daily logs from topic pages
Runs daily logs and topic pages separately. Recently started keeping conversations with Claude directly as a log. Before reading a paper, has Claude produce a chapter-by-chapter weighting guide.

Detailed logs for experiments and analyses (Lab Notebook)

When you run experiments or analyses, keep a detailed log separate from the Research Log.

What to write: date, purpose, procedure (equipment, settings, parameters), results (tables, graphs, photos), interpretation, next steps.

Example entry (Markdown / Jupyter)

# 2024/06/01 Thermal conductivity measurement of geothermal fluid
## Purpose
Measure the change in thermal conductivity with pressure and compare experiment with simulation
## Procedure
1. Compute with TOUGH2 at 10 / 15 / 20 MPa
2. Compare with experimental data
## Results
| Pressure (MPa) | Temperature change (°C) | Thermal conductivity (W/mK) |
|---|---|---|
| 10 | 5 | 0.58 |
| 15 | 5 | 0.61 |
| 20 | 5 | 0.65 |
## Interpretation
- Thermal conductivity rises with higher permeability
- Difference between experiment and simulation (needs re-verification)
## Next steps
- Acquire additional data with varied permeability

Good and bad examples

GoodBad
Date and conditions are stated explicitlyOnly "ran the experiment," conditions unknown
Results are organized in tables and graphsNumbers are scattered
Interpretation and next steps are presentEnds with "acquired data"

A Google Colaboratory format example is in Notion (Data Management).

Working with AI (Tips)

Summary of this page: Research Log is facts, Thinking Log is misalignment / In your own words, keep going, searchable, in one place / The method is yours: choose from examples A–G / Keep a separate detailed log for experiments and analyses / The misalignments in the Thinking Log become Idea-0 on the Paper Card.