Patience is higher for the first action than for later repeats

What it is

The first time the album opens, two seconds of spinning can still be read as “it is getting ready”. After one photo, another two seconds to swipe to the next feels like a trick. Patience on the first action is higher than on later repeats: the same delay, spent as a cover charge, can be explained as preparation; spent on an already-known act, it can only be explained as regression. Patience is not a personality trait. It is a judgement of whether this is still learning / still opening.

The previous leaf said the first hit is objectively slower. This one says people budget more patience for it — so putting the cover charge on the first call fits the patience curve better than putting it on the fifth.

Why it happens

The first time carries structural uncertainty: how heavy is this feature, will it ask for permission, what does a result look like. Waiting can be coded into “getting to know this object”. On repeat the structure has closed; each call should pay only steady-state cost. Delay at first-hit level switches the explanation from “it is preparing” to “why is it still like this” or “is it broken”. Once the explanation flips, the same two seconds go from acceptable to offensive.

Repeat also activates comparison: the second trial takes the first as an anchor. First at 2 s, second at 2 s, the anchor is never refreshed to “it can be faster”; people do not thank the stability, they wonder why fluency did not speed it up. First at 2 s, second at 0.3 s, the anchor refreshes; a fifth trial back at 2 s offends more than never having been fast.

Studying it

On one task, compare “1st in session” and “5th” at the same delay: acceptability ratings, retries, and spoken attribution (preparing / broken / regressing).

Independent variables: trial index, whether delay matches the first hit, whether a clearly faster response appeared in between. Dependent variables: acceptability, attribution class, premature cancel rate on the 5th.

Telling people “the first one will be slower” raises patience and makes the lab unlike the field. Nobody in the field is warned about the cover charge.

Where it stops holding

Experts opening a daily tool have almost no “first-time patience” — every open is a repeat. Rare heavy jobs (taxes, annual summary) feel first every time, and the patience budget is issued again. After one error, the next trial’s patience is drained by the error and is no longer issued as first-time. When progress is visible, patience on repeated waits drops more slowly, because comparison is of stage, not of duration.

Applying it

• Leave unavoidable cover charges on a true first (opening the feature, first fetch after permission). Do not levy them again on every list page, every send.
• Accept repeated paths against steady-state delay. “Users tolerated it the first time” is not the budget for the repeat path.
• If a repeat path occasionally returns to first-hit slowness, treat it as a fault, not as “sometimes it goes cold”.
• How to check: five identical operations, delay held at the first-hit level. From the third onward, impatience, retries or “why is it still this slow” should appear. Cut the last four clearly shorter than the first; the impatience should go — that is the patience curve working.

Related

• Same group: I1.07.1 The first response carries extra init cost and is usually slower than steady state · I1.07.3 Warmup or preload can move first-hit cost to a moment the user does not feel · I1.07.4 A large first–steady gap makes users misjudge overall performance
• Nearby: I1.02 Continuity-of-thought threshold · I1.05 Latency jitter
• Search terms: first-time patience · repeated action latency · expectation after warmup