Path dependence of layouts outweighs efficiency optima

What it is

Once a letter layout has been practised into automatic action at population scale, swapping it for a map that shortens finger travel on paper rarely raises overall throughput. That lock-in is path dependence: the prevailing keymap need not be optimal under present engineering constraints; it persists because already-acquired motor programs keep running. QWERTY is the usual exhibit, but the claim is about any layout that has become a shared skill, not about one particular diagram.

Why it happens

Skilled typing is not a sequence of independent aims. Frequent words, affixes, and punctuation clusters are stored as whole keystroke sequences; fingers fire timed patterns around the home row. Efficiency tables, by contrast, treat every keystroke as a separate Fitts aiming movement. An efficiency optimum assumes an empty skill account: shorter distances and fewer same-finger stretches should help a novice. Path dependence notes that the account is not empty. Changing the layout discards procedural chunks and forces the same lexicon to be recoded onto new coordinates. The larger that skill stock, the harder it is for a geometric saving to beat continued use of the old mapping. Layouts judged “better” on travel alone are treating already-deposited motor programs as zero.

Studying it

Layout comparisons typically use dictation, copy-typing, or controlled phrase sets. Independent variables include keymap, practice duration, whether the keyboard is visible, and letter or bigram frequencies in the text; dependent measures include time per character, error types, same-finger digraph rates, and practice needed to reach a speed criterion. The first tens of minutes with untrained participants inflate a new layout’s geometric advantage; a single session with already-automatized experts writes the old layout’s skill stock down as the new layout being “worse.” Plot the full practice curve, and report cruise speed separately from the tuition paid before that speed. Theoretical indices (total travel, same-finger stretches) are predictions, not substitutes for performance after chunks have formed.

Where it stops holding

The claim concerns letter layouts that already carry stable motor programs. Infrequent typists, writing systems without a population standard, and software keyboards that rearrange keys so users never store positions can still be dominated by efficiency differences. Experimental touch keyboards that morph with context fall outside this claim. “A popular layout is not travel-optimal” also does not mean every incumbent should be replaced: when there is no transferable skill stock, travel tables remain useful.

Applying it

• Default to the system layout users already fluent in; do not promote a new alphabet map for paper savings in key travel.
• If an alternative layout is offered, make it a reversible personal option that coexists with the system layout rather than a one-shot global replacement.
• Judge a new default by practised users’ learning curves, not by novice throughput in the first ten minutes.

Related

• Same group: C6.01.2 Migration cost of alternative layouts exceeds their efficiency gain · C6.01.3 Layouts vary by language and region, and shortcut positions move with them
• Adjacent: C6.02 Key travel and tactile confirmation · C6.17 Path dependence of keyboard layouts
• Search: path dependence · motor program · QWERTY