Transparency
Telling people how you are influencing them rarely stops the influence working.
Transparency here means disclosing the mechanism, not just the terms: saying that an option is preselected, that a feed is ranked, or that a comparison is drawn from a particular group. The consistent experimental finding is that disclosure does not destroy the effect, which removes the main excuse for not disclosing. Disclosure is necessary and it is not sufficient.
How it shows up in software
Transparency in software is the line saying this setting is on by default, the sponsored label on a result, the explanation of why a post ranked, and a changelog for the terms. Most products achieve legal disclosure and almost no comprehension, because everything true is available and nothing true is prominent.
Using it well
- Disclose the mechanism at the point of decision, in the same visual weight as the thing being disclosed.
- Test comprehension rather than presence. Ask users afterwards what they agreed to, and fix the copy until they can say it.
- Write the sentence you would be comfortable reading aloud to the user. If the honest version of the disclosure changes their choice, you have found the real problem.
- Version and publish changes to defaults, ranking and pricing, so long-term users can see what moved under them.
Where it turns manipulative
- Using disclosure as cover for a practice you would otherwise not run turns transparency into a permission slip. The experiments show the influence survives disclosure, which means disclosure does not launder it.
- Burying the disclosure in a policy, a tooltip or a collapsed section is technically honest and functionally deceptive.
- Disclosing a partial truth, such as naming the data collected but not the inferences drawn from it, is the modern version of fine print.
Where you have seen it
Google search Sponsored labels
Paid results carry an in-line label at the top of the result rather than a legend elsewhere on the page.
Instagram Why you are seeing this post
A per-post explanation of the ranking signals, reachable from the post itself.
Basecamp and similar plain-language terms
A summary in ordinary English alongside the legal text, with a published history of changes.
What the research says
- Loewenstein, Bryce, Hagmann and Rajpal, 2015 (Journal of Marketing Research)Well evidenced
Telling participants in advance that a default had been set to influence them, and why, did not significantly reduce the default's effect on their choices.
- Bruns, Kantorowicz-Reznichenko, Klement, Luistro Jonsson and Rahali, 2018 (Journal of Economic Psychology)Well evidenced
Four transparency variants, including full disclosure of purpose and expected effect, left the nudge: A change to how choices are presented that shifts behaviour without removing any option.'s effect on contributions essentially unchanged.
- Ben-Shahar and Schneider, 2014 (More Than You Wanted to Know)Well evidenced
A broad review of mandated disclosure across consumer finance, health and privacy, concluding that disclosure regimes routinely fail to change decisions because people do not read or cannot use them.
This is the counterweight. Disclosure that satisfies a regulator and nobody reads is compliance, not transparency. Position and comprehension are what matter.
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.