PlaySolitaire’s frozen August 2026 dataset reports 37.0% completion for Draw 1 and 38.0% for Draw 3. That difference does not establish that drawing three cards makes Solitaire easier. The players chose their modes; they were not randomly assigned to otherwise identical conditions.
Analysis type: a calculation from two published totals, followed by a made-up example. There is no new player dataset here, and the August figures are not September measurements. The main report contains the full results.
Recalculate the published comparison
| Mode | Eligible starts | Observed wins | Wins / starts | Published 95% confidence range |
|---|---|---|---|---|
| Draw 1 | 138,759 | 51,334 | 37.0% | 36.7–37.2% |
| Draw 3 | 12,190 | 4,634 | 38.0% | 37.2–38.9% |
The arithmetic is 100 × wins / starts, rounded to one decimal. Starts fall from July 15, 2026 at 12:50:26 UTC, inclusive, to August 4 at 07:00 UTC, exclusive. Wins were counted before August 6 at 07:00 UTC. This allowed at least 48 hours for each included start to produce a recorded win. The methodology explains how records are checked and duplicates removed.
The data includes only play for which analytics consent was given. Unlimited undo is included; Daily and Klondike winnable or golden pools are excluded. Included starts with no recorded win still count toward the total starts. One person may have played many times: these totals count attempts, not individual players.
How an overall comparison can reverse
The following counts are invented for explanation. They are not a breakdown of our users, and “newer” and “experienced” were not measured in the published dataset.
| Illustrative group | Draw 1 | Draw 3 |
|---|---|---|
| Newer players | 18 wins / 90 starts = 20% | 1 / 10 = 10% |
| Experienced players | 8 / 10 = 80% | 63 / 90 = 70% |
| All starts combined | 26 / 100 = 26% | 64 / 100 = 64% |
Draw 1 has the higher completion rate in each group. Yet Draw 3 has the higher combined rate because most of its attempts come from the higher-completion group. To find the combined rate, add the wins and divide by all starts. Simply averaging the two group percentages would give the wrong answer because the groups are different sizes.
This arithmetic proves that selection can produce a reversal. It does not prove that this particular mechanism explains the real 37.0% and 38.0% figures. Experience, persistence, devices and other differences are possible explanations, not findings established by these two rows.
What the confidence ranges do and do not show
The table’s confidence ranges use the Wilson method to describe uncertainty from the number of recorded attempts. They cannot correct for different players choosing each mode. They also do not make repeated games by one person independent. Whether the ranges overlap does not, on its own, show what changing the draw mode does to difficulty.
To test whether draw mode causes a difference, a study would need comparable players and deals. It would also need clear rules for counting replays, abandoned games and repeated players. We have not run that study.
Use each mode for its actual challenge
Draw 3 constrains access to the waste; the Draw 3 Solitaire page explains the packets. To compare the experience for yourself, play classic Solitaire and then try Draw 3 while keeping your practice rules consistent. A few personal games can help you choose a mode, but they cannot show which mode is easier for everyone.
The reproduction script recalculates both published percentages and every fraction in the made-up table. No historical v1 counts are combined with these v2 rows.