Writing feel is a subjective experience and cannot be fully measured by objective figures such as latency and accuracy
Aliases: writing feel · paper-like · haptic feel
What it is
Latency, reporting accuracy, and pressure bit depth can be measured, and they do affect writing. Together they still are not feel: people also use drag, vibration spectrum, sound, barrel inertia, even how the ink looks, to judge “like writing on paper”. Two devices with almost the same objective figures can be called plastic and paper. Feel has to be collected on its own; a latency pass cannot be declared a feel pass.
Why it happens
In multisensory integration, mechanics and acoustics are often taken as “material” earlier than a millisecond gap. Skating from missing friction, a shrieking high frequency, a hollow light barrel, will swamp already-low latency. The reverse also holds: some people will take slightly higher latency for obvious paper drag. Priors matter: fountain-pen habits and pencil habits name the same µ differently. Lab forced choice (which is more like paper) can rank, but does not yield a “feel score” reusable across studies. Hard-coding subjective items into one metric (a weighted sum of sound, µ, latency) fails on the next cohort, because the weights were this cohort’s. Objective figures still must be measured—they explain some variance and are the knobs engineering can turn—but acceptance needs a separate subjective report after free writing, or the screen will be optimised into something that follows and aims well and that no one will write on for long.
Studying it
Holding latency and accuracy fixed (or reporting them), change surface/nib and collect forced choice, Likert “like paper / willing to write a long time”, and fatigue after fifteen minutes. Regress how much variance in the subjective scores the objective quantities explain.
Independent variables: µ, sound level, latency, accuracy, barrel mass. Dependent measures: paper-likeness, long-session willingness, fatigue, R² from objective to subjective.
A low R² is not a failed method; it is evidence that objective figures are not enough.
Where it stops holding
In latency-critical sketching, the time component of feel grows and friction’s weight drops. Some accessibility users depend more on visual alignment than on shear. A procurement contract that only names a latency cap lets vendors ignore the surface—an indicator-design problem.
Applying it
- Specify latency/accuracy and the surface–nib pair together, plus a subjective gate (willingness to continue after fifteen minutes of writing).
- Do not title the writing experience with “lowest latency in the industry” alone.
- Have target users (not only engineers) blind-choose surfaces, then open the objective data.
- How to check: two configs close in latency and accuracy, different in µ and sound; blind writing should separate them. If it cannot, ask whether only subjective noise remains; if it can, a latency table must not claim the two experiences are the same.