The line
Where this turns into manipulation
Knowing how the brain reacts to the fear of losing, to scarcity and to an excess of options serves to take effort off whoever uses. The same knowledge serves to charge a toll, because the same knowledge about behaviour is a lever in both directions, and the screen in the two cases can be identical. The difference is whether or not there is something true behind it. That is the line I try to draw, bias by bias. And there is one more risk, which no manipulation matrix captures because it coerces no one: designing well for a body you presumed without noticing you presumed it.
The illusion of choice
An excess of options empowers no one. It tires, and consent comes out of tiredness.
Start with choice, because it is the easiest example to recognise. The Hick–Hyman lawCut what does not change the decision. Splitting into steps does not reduce the cost of choosing, it spreads it.Open the law → says that each extra option costs deciding time, and anyone who has tried to cancel a subscription knows what happens when someone uses that against you: the third menu wins by fatigue, not by argument. Privacy consent uses the same raw material by another route, which is not William Edmund HickMeasured decision time as a function of the number of alternatives, and found the logarithmic relation that bears his name.One reference in this work:1952On the rate of gain of informationSee on Wikipedia ↗See in the bibliography →’s logarithm but the accumulated fatigue of twenty little boxes and a long path to “no”, with “yes” one click away. And when the behavioural data collected to personalise each screen is added to that labyrinth, design stops serving a decision and starts predicting and steering it. No one was coerced at any moment. It was autonomy that got left along the way, one option at a time.
The ethics of the methods
Measuring non-conscious reaction requires a consent the person understands.
The measurement methods I describe on the Methods page widen what can be known about a person: the pupil that dilates, the skin that starts to conduct more, the wave that rises when the screen asks too much. None of it passed through the awareness of whoever used it, and responsibility grows with the reach. Informed consent, in this case, means the person understood what kind of data they are handing over, whether it is the emotional reaction or the body’s signal, and what will be done with it; clicking on a notice is not that. In Brazil this has an article in the law. The LGPD treats biometric data as sensitive personal data, and for it consent has to be specific and set apart, not the same little box as everything else. Then comes the reading, which asks for the same caution I demand of the studies cited in the laws, because a biometric signal records that something happened, not what the person wanted. And then comes safekeeping, because this data is more intimate than any browsing history, and treating it with the same carelessness is the quietest way of abusing it.
Who doesn’t show up in the count
The risk that coerces no one and only leaves people out.
Using the knowledge against the person is the risk everyone sees, and it is the one that choice and methods deal with. There is a third, and it coerces no one: a law measured in one kind of body, applied as if it held for everyone, simply leaves people out, and in a way that generates no complaint, because those left out are usually not in the room. That is why “which body are we talking about?” is an ethical question, and not only one of precision. It decides who counts as a user.
The matrix, bias by bias
Each row gives the principle, the use that crosses the line, and what to do instead.
| Principle | Where it turns into manipulation | What to do instead |
|---|---|---|
| Hick–Hyman law | Piling up options in privacy and cancellation flows until consent comes out of exhaustion. | Cutting what doesn’t change the decision, and giving “no” the same number of clicks as “yes”. |
| Parkinson’s law | Setting a tight deadline for someone who has no command of their own time, and calling it productivity. | A deadline negotiated with whoever will meet it, and spelled out: what happens when it expires, and what doesn’t. |
| Salience bias | Hiding the cancellation and highlighting the promotional path, the most banal and most effective dark pattern. | Giving both exits the same visual weight when both are legitimate choices. |
| Clustering illusion | Presenting noise as a pattern, and resting the decision on a chart that confirms what was already wanted. | Actively looking for what refutes the finding before taking it to the meeting. |
| Zeigarnik effect | Leaving the task artificially unfinished: a fictional “step 1” to force step 2. | Signalling the real end of each stage, and never inventing a stage to hold on to someone who has already finished. |
| Negativity bias | Focusing on negative messages to provoke action through fear or aversion. | Saying what is at risk without blaming the reader: what happens if nobody does anything, and what can be done now. |
| Scarcity bias | Creating an artificial sense of urgency to force impulsive decisions. | Using scarcity only when legitimate, indicating real deadlines or limits clearly. |
| Choice-supportive bias | Validating options already chosen, reinforcing decisions with no room for reassessment. | Leaving the decision reversible on the same screen where it was taken, with no new step to go back. |
| Anchoring bias | Influencing the perception of value through manipulated comparison, such as inflated prices struck through. | Showing the reference price with a date and where it was charged; if there is none, striking nothing through. |
| Social proof | Using generic or fabricated collective validation to influence decisions. | Saying where the number comes from and when it is from, and showing nothing when there is no data. |
| Mere-exposure effect | Repeating stimuli to create misleading familiarity and influence unconscious preference. | Stating explicitly the reason for the repetition: saying it is a recommendation, and why. |
| Illusion of control | Giving a sense of control over an automated system the person does not actually control. | Real control alongside the perceived kind, and saying clearly where the limit of each action lies. |
What is left of this
The role of whoever designs, today, includes protecting the person from the weaknesses the interface itself can exploit in them, and that begins before the screen, in the biases of whoever draws it, which enter the product uninvited when nobody looks for them. I don’t think a list of prohibitions solves this, because the list ages and the possibility of using the knowledge against whoever uses it does not. What solves it is treating ethics as a quality criterion, of the same kind as contrast and response time, and not as an appendix read at the end. In a field where the same knowledge turns into profit as easily as it turns into care, designing realities is reaching into deep layers of someone else’s cognition, and the question that separates craft from manipulation comes before any of the 38: if the person knew exactly how this works, would they still click?