In interaction design, what do Fitts's law and Hick's law each predict, and how are they commonly misapplied?

## Fitts's law: pointing **Fitts's law**, from experimental psychology in the 1950s, predicts how long it takes to move to a target and acquire it. In the formulation most used in interaction design: > movement time = a + b × log2(distance / width + 1) where **distance** is how far the pointer or finger must travel, **width** is the target's size along the direction of movement, and **a** and **b** are constants measured for a given device and person. The logarithm term is called the **index of difficulty**. Three consequences follow: - **Distance and width trade off.** A target twice as far away and twice as wide has the same difficulty. - **Returns diminish.** Because the relationship is logarithmic, enlarging a small, distant target helps a lot; enlarging an already large, nearby one helps very little. - **It is about movement, not decision.** Fitts's law says nothing about how long people take to decide where to point. The law holds across pointing methods, including fingers on touchscreens, although the constants differ and a finger adds imprecision of its own. ## What it means for layout 1. **Put actions near where attention already is.** In a benefits application, the 'Continue' action belongs right after the last question on a page, not in the far corner of a wide screen. 2. **Keep related controls together.** Actions used in sequence, such as choosing a document and then uploading it, should sit close to each other. 3. **Open menus at the point of interaction.** A contextual menu that appears where the pointer already is has almost no travel distance. 4. **Use screen edges for pointers, not for touch.** A mouse pointer stops at the edge of the screen, so an edge target is effectively infinitely deep in that direction. A finger does not stop there, and touch platforms often reserve edges for system gestures. 5. **Separate destructive actions.** The law works in reverse too: placing 'Withdraw claim' away from 'Continue' makes it harder to hit by accident. ## Hick's law: choosing **Hick's law**, often called the Hick-Hyman law, predicts decision time when a person chooses among alternatives. In a common formulation, reaction time grows with log2(n + 1), where **n** is the number of equally likely choices. The logarithm reflects that people can narrow down familiar options in stages rather than considering each one in turn. The conditions matter: the options should be **familiar**, the person should know what they want, and the options should be roughly **equally likely**. When options are unfamiliar and must be read one by one, the task is a **search**, and time grows closer to linearly with the number of items. ## Where each law is misapplied | Misapplication | Why it is wrong | Better reasoning | |---|---|---| | 'Fewer options is always faster', so split one list into nested steps | Each step adds its own decision, reading and navigation time | Group and label one list well; nest only where groups are meaningful | | Using Hick's law to judge a long, unfamiliar list | Scanning unfamiliar items is closer to linear search | Offer type-ahead search or a predictable order | | 'Bigger is always better' for every target | Returns are logarithmic, and space given to one target is taken from others | Enlarge the small, distant, frequently used targets first | | Treating Fitts's law as mouse-only | It models pointing in general, including touch | Apply it with different constants per input method | | Treating either law as proof a design works | Both model idealised tasks | Use them to form hypotheses, then observe real use | ## A benefits application example An applicant must choose which of fourteen benefits to apply for. Hick's law alone might suggest cutting the list or splitting it into steps. But this is not a choice among familiar, equally likely options: most applicants do not know the benefits' official names. A better design groups the options by life situation, such as 'I have lost my job' or 'I care for someone', uses plain-language labels, and keeps them on one page so no branch hides the right answer. For a 'country of birth' question with around two hundred options, the right tool is type-ahead search, because nobody decides among two hundred alternatives; they look for one they already know. ## Using the laws well Both laws are **models of idealised tasks**. They are good at predicting the direction and rough size of an effect, and at explaining why one layout beats another. They are not a substitute for watching people use the product: cite them to form a hypothesis, then check it.