The Benjamin Franklin Effect
Doing someone a favour can make you like them more, reversing the usual direction.
The claim is that performing a favour for a person increases liking for them, because the actor reconciles the effort with a belief that the recipient was worth helping. It runs opposite to the intuitive order, where liking comes first and the favour follows. The mechanism proposed is self-perception or dissonance reduction.
How it shows up in software
Contribution builds attachment. Users who answer a question in a forum, write documentation, file a good bug report, or mentor a newcomer become more committed to the product than users who only consume. The same logic sits under open source maintainership and under community moderation, where the labour creates the loyalty rather than the reverse.
Using it well
- Ask early users for a specific, small contribution they are qualified to make, such as correcting one documentation page.
- Make the contribution visible and attributed, so the effort becomes part of a public identity.
- Route newcomer questions to slightly-more-experienced users rather than to staff, which converts helpers into regulars.
- Accept the contribution properly, since a merged edit builds attachment and an ignored one destroys it.
Where it turns manipulative
- Extracting unpaid labour by framing it as community, while the output is a commercial asset and the contributor gets nothing, is exploitation dressed as belonging.
- Asking users for favours mainly to create sunk cost, so cancelling feels like abandoning their own work, uses attachment as a lock rather than as a reward.
- Leaning on volunteer moderators to absorb abuse and trauma the platform should be paying to handle transfers risk onto the people most invested.
Where you have seen it
Stack Overflow
Gives answerers reputation, badges, and editing privileges, so contribution earns standing that keeps contributors returning.
Wikipedia
Lets anyone edit and attributes every change to an account, turning a single correction into the start of an editing identity.
Home Assistant
Accepts community integrations and documentation edits through public review, where merged contributions become part of a user's own setup.
What the research says
- Jecker and Landy, 1969Contested
Participants asked by the experimenter to return money they had won rated him more favourably than those asked by a secretary or not asked at all.
A single study with a small sample and an indirect measure. It is the only direct experimental support usually cited, and it has no strong modern replication. Treat the effect as plausible rather than established.
- Bem, 1967Mixed evidence
Self-perception theory proposed that people infer their attitudes from observing their own behaviour, which is the mechanism usually invoked for this effect.
The theory has broad support in attitude research. That support does not transfer automatically to this specific favour-then-liking prediction.
- Franklin, 1791Contested
Franklin's autobiography describes asking a political rival to lend him a book, after which the rival became friendly.
An anecdote from a memoir written decades after the events. It named the idea; it does not support it.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.