The cost of a physical control is that it cannot be reconfigured

What it is

Where a key sits, how far it travels, who its neighbors are: frozen on the day the tool is cut. Software can move a hit region; it cannot move this key’s coordinates in the housing, or its feel. Non-reconfigurability is the cost paid for eyes-free use and certainty: the function set, the layout, and the force curve barely change over the product’s life. It is not merely “physical keys cost more.” Once interaction is written into material, later versions can only patch the frozen map.

Why it happens

Tools, circuits, certification, and supply turn a key’s existence into a multi-year constraint. Software that wants a new function can occupy an old key (introducing a mode), push the function onto the screen, or ask the user to buy the next hardware. If an old key is redefined, the motor program and the printed word expire together—and neither can be unscrewed. Force curves, spacing, and return springs freeze the same way; discovering in year one that a key is too light or too tight is not a firmware fix. A physical panel is therefore an early, hard-to-revise layout decision. What can be rebound is the software mapping; what cannot is the geometry. Those two are often conflated, so “we’ll define this key later” becomes mode hell after year one.

Studying it

Do layout archaeology across generations: which keys survived from first to third, what they meant in between, how many functions could only land on screen.

Independent variables: whether a key’s meaning was fixed at tooling, whether meaning changed mid-life, whether a soft-key zone could absorb new functions. Dependent measures: errors after a rebinding, fraction of users still acting on the old meaning, number of mode layers added for new functions, hardware respin count caused by an unchangeable layout.

Asking “what do you think this key does now” exposes freeze better than a manual. New-user lab tests miss leftover meanings; use people in the field or at least people who used the previous generation.

Where it stops holding

Modular panels, swappable caps, and magnetic keys rescue some geometry from tool-freeze, though certified safety keys still usually cannot be rearranged by the user. A software-defined knob (the same unmarked encoder, the screen saying what it is today) is still geometrically non-reconfigurable; only the binding moves—what is frozen is pose and feel, not the legend. Very short-lived disposable devices make freeze cheap. When certification writes a key’s existence into a clause, non-reconfigurability can even be a feature: a silent update cannot remove it.

Applying it

• Freeze into geometry only functions expected to stay for the product’s life; leave the rest to the screen or an explicit soft-key zone.
• Forbid “leave this key blank and define it later”—empty keys get stuffed with modes in year one.
• If a shipped key’s meaning must change, treat it as breaking: new cap, new lamp, new firmware default, and a plan for old-meaning false triggers.
• Verify: list functions likely to appear in the next two years and ask which existing key they land on. If the answer is “squeeze onto a key that already exists” more than twice, the physical map is already too small—cut hard keys or add a soft zone rather than stacking modes. Count users who still press the previous generation’s meaning.

Related

• Same group: C10.07.1 Frequent, eyes-free, and safety-critical functions favor physical controls · C10.07.2 Variable, infrequent, label-heavy functions favor on-screen controls
• Adjacent: C10.13 Mode Problems in Multifunction Physical Controls · C10.16 Mechanical Reliability and Lifetime
• Search: non-reconfigurable · hardware lock-in · panel freeze