Multiple options expose the system’s instability — a cost and a form of honesty

What it is

Three emails side by side, each a different tone, promise, close. The user says “you haven’t even decided.” One request drew three legal paths; the comparison UI laid the fork on the table. The system looks unstable, and the system no longer pretends there is one answer. Exposing instability is the double consequence of juxtaposition — the cost is a discounted impression of competence; the honesty is that the distribution is no longer hidden behind a single overwrite.

Whether spreading variability is more honest than a singleton is a different argument about display strategy. This entry governs what happens on the comparison floor: discount and candour arriving together once instability is seen.

Why it happens

The default model of a tool is a function: one knob, one setting. Legal forks side by side slap that model; competence is read as “not even sure itself.” Meanwhile a singleton dresses every draw as “the system’s output”; users only find the fork by retrying, and retry is more easily read as a fault. Juxtaposition moves discovery forward and spends one layout on those experiments.

Cost and honesty are therefore not two product phases; they are two faces of one display. Dropping juxtaposition can keep an impression of stability, at the price of hiding the draw; keeping juxtaposition means letting “you haven’t decided” into the room.

Studying it

Same prompt, singleton versus three juxtaposed. Measure: competence ratings, whether people understand “there could be others,” trust, whether they still retry. Independent variables: size of between-option differences (how visible the fork), whether a line says “these are possible drafts.” Dependent variables: impression of stability, awareness of variability, surprise when another answer shows up later.

Split by difference size. A tiny fork under-exposes, and people still feel stability; a huge fork steepens the competence discount. Do not collapse both into one main effect of “juxtaposition.”

Where it stops holding

Several options on a closed answer do not expose instability; they expose that the system cannot do the item, a discount with no matching honesty. When the user is exploring (composition, melody), “not decided” is the feature. High-stakes output that needs a single accountable signature can read exposed instability as dumping the decision. This entry does not treat whether relative evaluation is easier, nor position bias.

Applying it

• Beside the set, say in one line: these are different possibles from the same request, not the system throwing dice. Do not leave people to read a fault out of the fork.
• Accept that competence ratings may fall. If the product cannot bear “you haven’t decided,” do not use juxtaposition to compare; reverting to a singleton is a choice to hide the draw.
• Do not add a “recommended” highlight on top of juxtaposition to pretend a decision was already made. The highlight swallows the honesty and leaves only the cost of “why are the other two still here.”
• Check: after juxtaposition, ask “how sure is the system of the answer” and “would another generation differ.” The first may sit at medium; the second should already be “yes.” If surety is driven to zero and read as a fault, only the cost was paid, and the fork was not explained as a distribution.

Related

• Same group: L3.07.1 Juxtaposition drops the evaluation load from good-or-bad to which-is-better · L3.07.2 When options differ too little, comparison collapses into picking noise · L3.07.3 More options bring choice difficulty and decision regret · L3.07.4 People favour the first and left positions; order must be controlled or randomised
• Nearby: L3.01 Generating Multiple Options · L1.01 Mismatch Between Stochastic Output and Deterministic UI
• Search terms: exposing instability · visible sample variance · honesty of multi-option