An inconsistent mapping makes every contact a test

What it is

Stating a convention once is not enough. On a tablet, pen and finger move through several apps and several canvases in one sitting. When the mapping is inconsistent, a rule learned on the last screen cannot be carried to the next, and every contact becomes a test: will this pad-drag pan, lasso, or lay down ink that should not be there. The cost is not half a second. It is dirty data, wrong selections, and a refusal to use the non-dominant hand again. This entry is not about what the convention ought to be. It is about what happens after the convention drifts across apps and modes. Whether the task can finish without a pen is a different question.

Why it happens

A classification rule is a mode. People store “pen is ink, finger is view” as an expectation about this piece of glass, and the next contact is issued under that expectation. App A pans with a finger; app B also draws with a finger. Inside one app, a nib on empty space inks in “write” and selects in “markup.” A missed expectation is a mode error: the action went out, the system ran another semantics. Tablets drift easily because the pen is hardware shared across apps and the finger is a system-level input; people assume the whole device has one mapping. Each probe also trains a stability that is not there. After a few failures the non-dominant hand retreats to the bezel, the bimanual gain is gone, and the tablet is used as a large phone that only taps.

Studying it

Use a mode-error setup. Let people learn “fingers only pan” under a consistent mapping, then—without announcement—change one canvas so fingers also ink, or bounce between two side-by-side apps. Record the intended action of the first contact against what actually happened.

Independent variables: whether the mapping matches across apps, whether a mode switch inside one app changes nib semantics, whether the switch has a visible mode mark. Dependent variables: first-contact error rate, number of probes, stray strokes, whether the non-dominant hand drops out of the collaboration.

Do not tell participants that the two apps use different rules—that removes discovery cost from the measurement. The session has to include at least one app switch; short tasks never form an expectation, so drift cannot be seen.

Where it stops holding

When two apps are genuinely different tasks (a canvas versus an ebook), different mappings can be accepted if pen semantics are visible on entering the canvas (existing ink, a nib-shaped cursor). System back, screenshot, and the dock must not be rewritten by apps, or probing spreads into system gestures. If the user changed the mapping in settings, the inconsistency is chosen—but the current canvas still needs a visible mode indicator. On devices that classify a capacitive stylus as a finger, the “inconsistency” is a classification failure; copy will not fix it.

Applying it

• Use one pen–finger mapping on every canvas in the product. If a mode must be an exception, make the exception visible in canvas appearance and tool state so nobody has to probe.
• Do not swap roles across adjacent tools: fingers pan in markup and draw a stroke in crop, with only a tool icon changing in between, is the expensive kind of inconsistency.
• Do not silently change pen semantics on return from another app. If the system or another app stole the pen, restore the product’s convention on foreground and keep the current tool visible.
• Verify by using notes, browser markup, and photo markup in a row, with no instructions in between. If every “finger drag” starts with a hesitation or an undo, the mapping is still drifting.

Related

• Within the group: K3.02.1 Pen and finger need an explicit division of labor · K3.02.3 Every function still has to work without the pen
• Adjacent: C5.07 Division of Labor Between Pen and Finger · K2.07 Touch Coexisting with Keyboard and Mouse
• Search terms: mode error · inconsistent mapping · pen touch conflict