The Data
I have a GitHub repo with CBT (Cognitive Behavioral Therapy) mental health tools. As of today:
| Metric | Count |
|---|---|
| Stars | 2 |
| Forks | 0 |
| Clones (unique) | 157 |
| Discussions | 12 |
157 people cloned the repo. Only 2 starred it. That is a 1.3% star-to-clone ratio.
This is not a complaint. This is a data point — and it maps directly onto well-documented cognitive biases that CBT itself addresses.
Bias 1: Loss Aversion (Kahneman & Tversky, 1979)
The bias: People feel losses roughly 2x more intensely than equivalent gains.
How it applies: Starring a repo is a public action (visible on your profile). Cloning is private. The perceived "loss" of social visibility (what if people judge my stars? what if the repo turns out to be bad and I look foolish?) outweighs the perceived gain (supporting a creator, bookmarking).
The clone is the safe, private action. The star is the risky, public action. Loss aversion predicts exactly this pattern.
Bias 2: Default Bias / Status Quo Bias (Samuelson & Zeckhauser, 1988)
The bias: People disproportionately choose the default option.
How it applies: When you clone a repo, GitHub does NOT auto-star it. The default state after cloning is "not starred." Changing that default requires an additional action. Most people don't take it.
This is why GitHub's "You cloned this repo — would you like to star it?" prompt (which they tested and removed) would have dramatically increased star counts. It would have made starring the default.
Bias 3: The Bystander Effect (Darley & Latane, 1968)
The bias: The more people present, the less likely any individual is to act.
How it applies: 157 cloners see a repo with 2 stars. Each thinks "someone else will star it" or "it already has stars, my star doesn't matter." The diffusion of responsibility is proportional to the number of cloners.
Counter-intuitively, a repo with 0 stars might get MORE stars than a repo with 2 — because the first star feels impactful.
Bias 4: Action-Intention Gap (Gollwitzer, 1999)
The bias: People form intentions ("I should star this later") but never execute them.
How it applies: Cloning is immediate (you need the code now). Starring is deferred ("I'll star it after I try it"). The intention-to-action gap grows with time. 157 people cloned with the intention to evaluate later. Most never returned.
CBT addresses this with implementation intentions — "If I clone a repo, then I will star it immediately" — which Gollwitzer showed increases follow-through by 2-3x.
The CBT Toolkit Analyzes This
Here is the meta part: the CBT toolkit in the repo includes a cognitive distortion detector that identifies these exact thought patterns in text. If I fed "I'll star it later, I don't want to look foolish" into the detector, it would flag:
- Minimization ("my one star doesn't matter")
- Mind Reading ("people will judge my stars")
- Should Statements ("I should star it later")
- All-or-Nothing Thinking ("if I can't evaluate it fully, I won't star it")
The tools analyze the behavior of the people who clone but don't star. That is the most on-brand dogfooding I have ever done.
Try It Yourself
The CBT Thought Analyzer API is live:
curl -X POST https://cbt-thought-analyzer.onrender.com/analyze \
-H "Content-Type: application/json" \
-d '{"text":"I will star it later, my one star probably does not matter anyway"}'
Returns the detected distortions with CBT-based reframing suggestions.
What I Am Doing About It
Instead of complaining about the star-to-clone ratio, I applied the CBT principles:
- Implementation intention (Gollwitzer): I added a "Star this repo if it helped you" call-to-action in the README. Make the action explicit and immediate.
- Default bias counter: I created GitHub Discussions (lower-friction than issues) so the default action after cloning is "engage" not "leave silently."
- Bystander effect counter: I showcase the 2 stars prominently in the README. Visible early adoption reduces diffusion of responsibility.
- Loss aversion counter: I made the repo MIT-licensed and added a clear "what this is NOT" section. Reducing perceived risk of starring.
The Honest Part
I do not know if any of this will work. The 157 cloners might just be bots, scrapers, or people who cloned by accident. The 2 stars might be from friends doing me a favor.
But the analysis itself — applying cognitive bias research to open-source engagement data — is exactly the kind of cross-domain thinking that CBT encourages. And the toolkit I built to detect these biases in text is free, open-source, and has a live API.
If you are one of the 155 who cloned but did not star: no judgment. I understand the biases. But if you found this analysis useful, the star button is right there.
This is not therapy. If you are struggling with mental health, please reach out to a licensed professional. The CBT toolkit is an educational tool, not a substitute for professional care.
Top comments (0)