A four-table trial conducted across 1,842 sessions between January and March 2026 found that enabling auto-muck as the default action reduced misclicks in the final sixty seconds of a session by 9.1 percent, from 4.7 to 4.3 per hundred hands. The effect was measured only on hands where the client had already folded or where showdown exposure carried no informational value, and it disappeared entirely at tables with fewer than three active players. The finding matters less for the headline number than for what it reveals about where late-session errors actually originate: not in the decision to fold, but in the mechanical act of confirming it.
Methodology and the Problem of the Last Minute
The trial was run on a mid-stakes no-limit hold'em client with a persistent session timer, drawing on a pool of 611 players who opted into telemetry. Hands were tagged by elapsed session time, and a misclick was defined narrowly: any fold or check action taken within 1.5 seconds of the preceding action, subsequently reversed or contested in chat, or logged by the client as an input outside the intended button's hitbox. That definition excludes the ordinary fast-fold, which is why the baseline rate sits at 4.7 per hundred rather than something higher.
Session-60 misclicks — the shorthand for errors in the sixtieth minute and beyond — have a distinct profile. They cluster on the fold button, they correlate with tables where the player is not in the current hand, and they spike when a player is multi-tabling across four or more windows. The trial deliberately capped participation at four tables to keep the sample interpretable; results at six and eight tables are being collected separately and are not reported here.
The 9.1 percent reduction is smaller than the trial's designers expected. An earlier pilot at two tables showed a 14 percent reduction, but that pilot had a smaller denominator and a higher baseline misclick rate (6.2 per hundred), which inflates percentage reductions. Normalizing to the four-table baseline, the two-table figure would land near 10.5 percent. The number that should travel is the absolute one: roughly four fewer misclicks per thousand hands played in the sixtieth minute and beyond.
Why Auto-Muck and Not Auto-Fold
Auto-muck removes the confirmation step at showdown, not the decision to fold. A player who folds still clicks fold. What changes is the second interaction — the muck confirmation after a hand ends without a showdown, or the exposure prompt when a losing hand would otherwise be shown. Those secondary prompts arrive at exactly the moment a player's attention has already moved to the next table or the next hand, and they are the source of a disproportionate share of contested actions. In the trial, 61 percent of flagged misclicks involved a showdown-exposure prompt rather than a fold decision.
This distinction matters for how the result should be read. Auto-muck is not a decision aid. It is an interface simplification, and its measured benefit is a reduction in confirmation errors, not an improvement in hand selection. Any operator citing this figure as evidence of better player decision-making is overreading it.
Where the Effect Concentrates
The reduction was not evenly distributed. Broken out by table count, the four-table cohort accounted for nearly all of the measured effect; the one- and two-table cohorts showed reductions within the confidence interval of zero. That is consistent with an attention-allocation story: at one table, a player has enough slack to absorb a confirmation prompt without error, so removing the prompt changes little. At four tables, the prompt competes with three other windows for the same half-second.
| Tables active | Baseline misclicks per 100 hands (min 60+) | Post-change | Delta |
|---|---|---|---|
| 1–2 | 3.1 | 3.0 | −0.1 (n.s.) |
| 3 | 4.2 | 3.9 | −0.3 |
| 4 | 5.4 | 4.9 | −0.5 |
| 4, session 90+ | 6.8 | 6.0 | −0.8 |
The session-90+ row is the most interesting and the least powered. Players who stay past ninety minutes show a larger absolute benefit, but the subsample is 214 sessions and the confidence interval is wide. Treat the 0.8 figure as directional.
The Player-Selection Confound
One confound deserves flagging. Players who enabled auto-muck by default were not randomly assigned in the first two weeks of the trial; the default was rolled out to a subset of accounts and opt-out was available. Players who opted out may differ systematically in risk tolerance or in how much they care about hand-history completeness, and hand-history completeness is precisely what auto-muck trades away. The trial's later weeks used a forced default with an opt-out logged, and the effect held at 8.7 percent in that cleaner window — close enough to the headline that the confound does not appear to drive the result, but not identical.
There is also a real cost. Auto-muck means the player's own mucked hands are not displayed in the hand history by default in some client versions, which matters for players who review sessions to check their own fold frequencies. The trial did not measure how many players turned the feature off specifically to recover that data, and that number would tell us something about how much players value post-session review versus in-session smoothness.
What the Number Does and Does Not Support
A 9.1 percent reduction in a narrow, well-defined error class is a real result, but the error class is narrow. Misclicks in the sixtieth minute are not the same as losses in the sixtieth minute, and the trial made no attempt to convert misclicks into expected-value terms. A contested fold at a $1/$2 table costs less than the same fold at $5/$10, and the sample skews toward the lower stakes. Anyone extrapolating the percentage to a dollar figure is doing work the data does not do.
The more defensible claim is comparative: auto-muck default reduces confirmation-prompt errors at four tables by roughly half a misclick per hundred hands, and the effect scales with table count and session length. That is a modest, real, and mechanistically plausible improvement in an interface that most clients have left unchanged for years.
It also raises a question the trial cannot answer. If secondary confirmation prompts generate this many errors at four tables, what is the error cost of the prompts players cannot turn off — the ones attached to bet sizing, to time-bank activation, to the all-in confirmation that appears only above a threshold? Auto-muck was an easy default to flip because the cost of mucking is mostly informational. The next prompt on the list is not, and the same 9 percent logic will not automatically apply.