Writing Up Your Findings

An experiment you do not write up is an experiment you will forget. Here is how to write a small findings post people will actually read. That means exact prompts, model versions, dates, and the raw CSV.

40 min · Reviewed 2026

The Write-Up Is the Discovery

Researchers say: if you did not write it up, you did not do the experiment. Writing is where you discover what you actually learned — the parts that survive a second look and the parts that dissolve on contact.

A template that works every time

Title: the result in one sentence
TL;DR: the finding in 1-2 sentences
Motivation: why you ran this experiment
Method: what you did, reproducibly
Results: one figure or table plus the story it tells
Limitations: what this does not show
Takeaway: the generalizable lesson

Writing principles that travel

State the result first, then explain how you got there (inverted pyramid)
Every number has a confidence interval or a sample size
One chart is worth more than paragraphs
Never bury a negative finding
Leave a 'things I still do not understand' section at the end

Amateur write-up	Professional write-up
Sells the result	Describes the result with caveats
Hides the ugly data	Shows all the data
No reproduction instructions	Includes exact prompts, models, seeds
One chart, no caption	Every chart self-explanatory with caption
No limitations section	Limitations front and center

Where to publish

Personal blog or Substack — fullest control
LessWrong, Alignment Forum — AI-curious audience
arXiv (if you have standing) — for citations
Tendril's Experiment Log (planned for a future update) — for students
A class Discord or club channel — gets fastest feedback

If a paper is the story, the write-up is the map. Without the map, the territory was never really yours.
— Adapted from research methodology texts

The big idea: the write-up is not the part after the work. It is the part where the work becomes knowledge. Ship the post. Build the habit. The rest compounds.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-creators-writing-up-findings

What is the core idea behind "Writing Up Your Findings"?
1. An experiment you do not write up is an experiment you will forget. Here is how to write a small findings post people will actually read. That means exact prompts, model versions, dates, and the raw CSV.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
Which term best describes a foundational idea in "Writing Up Your Findings"?
1. TL;DR
2. writeup
3. reproducibility
4. publishing
A learner studying Writing Up Your Findings would need to understand which concept?
1. writeup
2. reproducibility
3. TL;DR
4. publishing
Which of these is directly relevant to Writing Up Your Findings?
1. writeup
2. TL;DR
3. publishing
4. reproducibility
Which of the following is a key point about Writing Up Your Findings?
1. Title: the result in one sentence
2. TL;DR: the finding in 1-2 sentences
3. Motivation: why you ran this experiment
4. Method: what you did, reproducibly
Which of these does NOT belong in a discussion of Writing Up Your Findings?
1. TL;DR: the finding in 1-2 sentences
2. Surface uncertainty in the UI so users can weigh it
3. Motivation: why you ran this experiment
4. Title: the result in one sentence
Which statement is accurate regarding Writing Up Your Findings?
1. Every number has a confidence interval or a sample size
2. One chart is worth more than paragraphs
3. State the result first, then explain how you got there (inverted pyramid)
4. Never bury a negative finding
Which of these does NOT belong in a discussion of Writing Up Your Findings?
1. State the result first, then explain how you got there (inverted pyramid)
2. One chart is worth more than paragraphs
3. Every number has a confidence interval or a sample size
4. Surface uncertainty in the UI so users can weigh it
What is the key insight about "Publishing beats perfecting" in the context of Writing Up Your Findings?
1. A 70 percent write-up published today beats a 95 percent write-up never published. You can always write a follow-up.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
What is the key insight about "Be kind to your future self" in the context of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Write the post as if your future self (reading it in a year) needs to reproduce your result from scratch.
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
What is the key insight about "Review date" in the context of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. Reviewed in 2026. Treat fast-changing product names, prices, availability, and policy details as examples to verify befo…
4. They make different models directly comparable
Which statement accurately describes an aspect of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. They make different models directly comparable
4. Researchers say: if you did not write it up, you did not do the experiment.
What does working with Writing Up Your Findings typically involve?
1. The big idea: the write-up is not the part after the work. It is the part where the work becomes knowledge. Ship the post. Build the habit.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
Which best describes the scope of "Writing Up Your Findings"?
1. It is unrelated to foundations workflows
2. It focuses on An experiment you do not write up is an experiment you will forget. Here is how to write a small fin
3. It applies only to the opposite beginner tier
4. It was deprecated in 2024 and no longer relevant
Which section heading best belongs in a lesson about Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. A template that works every time
4. They make different models directly comparable

← Back to interactive lesson

Tendril · Creators · AI Foundations

Writing Up Your Findings

40 min · Reviewed 2026

The Write-Up Is the Discovery

A template that works every time

Title: the result in one sentence
TL;DR: the finding in 1-2 sentences
Motivation: why you ran this experiment
Method: what you did, reproducibly
Results: one figure or table plus the story it tells
Limitations: what this does not show
Takeaway: the generalizable lesson

Writing principles that travel

State the result first, then explain how you got there (inverted pyramid)
Every number has a confidence interval or a sample size
One chart is worth more than paragraphs
Never bury a negative finding
Leave a 'things I still do not understand' section at the end

Amateur write-up	Professional write-up
Sells the result	Describes the result with caveats
Hides the ugly data	Shows all the data
No reproduction instructions	Includes exact prompts, models, seeds
One chart, no caption	Every chart self-explanatory with caption
No limitations section	Limitations front and center

Where to publish

Personal blog or Substack — fullest control
LessWrong, Alignment Forum — AI-curious audience
arXiv (if you have standing) — for citations
Tendril's Experiment Log (planned for a future update) — for students
A class Discord or club channel — gets fastest feedback

If a paper is the story, the write-up is the map. Without the map, the territory was never really yours.
— Adapted from research methodology texts

The big idea: the write-up is not the part after the work. It is the part where the work becomes knowledge. Ship the post. Build the habit. The rest compounds.

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-creators-writing-up-findings

What is the core idea behind "Writing Up Your Findings"?
1. An experiment you do not write up is an experiment you will forget. Here is how to write a small findings post people will actually read. That means exact prompts, model versions, dates, and the raw CSV.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
Which term best describes a foundational idea in "Writing Up Your Findings"?
1. TL;DR
2. writeup
3. reproducibility
4. publishing
A learner studying Writing Up Your Findings would need to understand which concept?
1. writeup
2. reproducibility
3. TL;DR
4. publishing
Which of these is directly relevant to Writing Up Your Findings?
1. writeup
2. TL;DR
3. publishing
4. reproducibility
Which of the following is a key point about Writing Up Your Findings?
1. Title: the result in one sentence
2. TL;DR: the finding in 1-2 sentences
3. Motivation: why you ran this experiment
4. Method: what you did, reproducibly
Which of these does NOT belong in a discussion of Writing Up Your Findings?
1. TL;DR: the finding in 1-2 sentences
2. Surface uncertainty in the UI so users can weigh it
3. Motivation: why you ran this experiment
4. Title: the result in one sentence
Which statement is accurate regarding Writing Up Your Findings?
1. Every number has a confidence interval or a sample size
2. One chart is worth more than paragraphs
3. State the result first, then explain how you got there (inverted pyramid)
4. Never bury a negative finding
Which of these does NOT belong in a discussion of Writing Up Your Findings?
1. State the result first, then explain how you got there (inverted pyramid)
2. One chart is worth more than paragraphs
3. Every number has a confidence interval or a sample size
4. Surface uncertainty in the UI so users can weigh it
What is the key insight about "Publishing beats perfecting" in the context of Writing Up Your Findings?
1. A 70 percent write-up published today beats a 95 percent write-up never published. You can always write a follow-up.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
What is the key insight about "Be kind to your future self" in the context of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Write the post as if your future self (reading it in a year) needs to reproduce your result from scratch.
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
What is the key insight about "Review date" in the context of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. Reviewed in 2026. Treat fast-changing product names, prices, availability, and policy details as examples to verify befo…
4. They make different models directly comparable
Which statement accurately describes an aspect of Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. They make different models directly comparable
4. Researchers say: if you did not write it up, you did not do the experiment.
What does working with Writing Up Your Findings typically involve?
1. The big idea: the write-up is not the part after the work. It is the part where the work becomes knowledge. Ship the post. Build the habit.
2. Surface uncertainty in the UI so users can weigh it
3. Abstain from answering when uncertainty is too high
4. They make different models directly comparable
Which best describes the scope of "Writing Up Your Findings"?
1. It is unrelated to foundations workflows
2. It focuses on An experiment you do not write up is an experiment you will forget. Here is how to write a small fin
3. It applies only to the opposite beginner tier
4. It was deprecated in 2024 and no longer relevant
Which section heading best belongs in a lesson about Writing Up Your Findings?
1. Surface uncertainty in the UI so users can weigh it
2. Abstain from answering when uncertainty is too high
3. A template that works every time
4. They make different models directly comparable

← Back to interactive lesson