Text must be extremely compressed without becoming ambiguous

What it is

Copy on a watch must be short enough to finish in one look, but not so short that “2” could be two kilometres, two stops, or two hours. Compression removes modifiers and repetition, not units, names, or negation. Ambiguity here is concrete: the same string can honestly be read as two different next actions in the current setting. This entry is only about whether the words can be misread. It is not about how many foci share a screen, and not about whether those words arrived by paging or scrolling.

Why it happens

A glance takes a few words; the brain fills empty slots from context. Slot-filling is fast and wrong easily: “2:30” during a run is pace, in a calendar it is a clock time. “Alex” is enough when there is one Alex; with two threads of that name it is a coin flip. Units, proper names, and polarity (not, cancelled, expired) are the constraints slot-filling can least spare; deleting them saves the most expensive pixels. Icons do not inherit those constraints automatically: a rightward chevron is a turn in navigation, a read receipt in mail, a skip in a player. The right unit of compression is “the few characters that still uniquely determine the next step,” not “fewer words are better.”

Studying it

Use forced-choice paraphrase: show the copy for a second or two, offer two next actions that both make sense, and see which one people pick. Contrast with/without unit, full name versus given name, polarity word kept versus dropped.

Independent variables: unit present, completeness of the object name, polarity word present, icon present at the same time. Dependent variables: rate of choosing the intended next action, hesitation between the two readings, whether the critical constraint can be recalled afterward.

If participants already know this is navigation versus mail, the experimenter has fed them context and ambiguity cannot be measured. Place the same string in two plausible app settings and see whether it “reads” in both. Do not replace this choice with a readability formula—formulas reward short and do not punish double meaning.

Where it stops holding

User-chosen shortenings (contact nicknames, short names for frequent addresses) are not ambiguous in that person’s context; they are not templates for general UI. Legal, dose, and payment fields cannot compress unit or polarity; two lines are cheaper than eating the word “not.” Compression ratios differ by language: Chinese characters carry more per glyph; English abbreviations collide more (min / mi / m). A numeral that repeats in a complication slot is disambiguated by slot memory, not by unit; that memory does not transfer to a one-shot notification.

Applying it

• Cut adjectives and boilerplate; keep unit, proper name, polarity. Read the result aloud; if two next actions can be heard, put back the constraint that was cut.
• When a person name is not unique in the current list, add a field that splits it (surname, group, time). Do not rely on an avatar alone.
• Icons may repeat an assertion the copy already makes; they must not carry unit or negation by themselves.
• Verify by cutting the copy out and showing it to someone who did not see the previous screen: “what would you do next.” Two equally reasonable answers means compression went too far. Fix until only one action is reasonable, then check the words still fit in a glance.

Related

• Within the group: K4.05.1 One screen carries one focus · K4.05.3 Paging costs more than scrolling
• Adjacent: K4.01 Wrist Raise and Glance · K4.06 Watch Faces and Complications · S3.04 High- and Low-context Information Density
• Search terms: abbreviation ambiguity · glance copy · unambiguous compression