Diplomacy · Spring and Fall 1901

Explore Diplomacy openings.

Choose recorded orders and compare their historical outcomes. Add countries or turns to narrow the matches.

Choose orders one turn at a time, from Spring 1901 through Winter adjustments. Each selected turn narrows the results; unselected turns may vary. Failed attempts are included.

What does game quality mean?

These tiers measure game length and signs of players dropping out, not player skill. Tiers 1–3 have a resolved turn in 1904 or later.

Tier 1: No detected terminal dropout.
Tier 2: At least one detected dropout; every dropout began with at most 2 supply centers.
Tier 3: At least one dropout began with 3 or more supply centers.
Tier 4 (excluded): The last resolved turn was before 1904.

We flag a terminal dropout when all of a country’s units hold for at least 3 consecutive movement turns through elimination or the end of the record. Intentional holds can trigger this flag; short absences or players who resume moving will not. Tier 1 therefore does not guarantee uninterrupted play.

Choose an opening below to narrow the results.

Choose an opening

Browse a turn, then select its orders. Unselected turns may vary.

Search recorded orders

Sort openings

Show estimated scores (experimental)

Choices are ordered by frequency.

How scores and comparisons work

Estimated score blends an opening’s mean with that country’s overall mean, giving the overall mean more weight when few games match. Both come from the same fitting subset, which excludes the benchmark’s test games. “Fitting matches” counts games with these exact orders in that subset; historical results use all included games. “Baseline only” means the model uses the country average without an opening-specific adjustment.

This experimental estimate applies to one country’s complete 1901 orders with All press settings. Select the other four turns, including “No recorded orders” where applicable; each card then estimates the complete year, not that turn alone. It is not a forecast made before Spring.

The benchmark found modest improvements over the country average in held-out games from this dataset. Estimates are not evidence that choosing these orders causes a score gain, and may not generalize to new players or future games. Sorting still uses historical means.

Historical mean score is this country’s average final score, from 0 to 1. A solo gives the winner 1 and everyone else 0, including countries that survived. In a draw, each country scores its final supply-center count squared, divided by the sum of those squares across all seven countries.

The baseline is this country’s average across all eligible games, with each game weighted equally. It includes the displayed choice and its alternatives. Other country and turn selections still apply; only the choice you are browsing is unrestricted. Search, sorting and pagination do not change it.

For example, a mean of 0.18 against a baseline of 0.15 gives +0.03 (+20.0%). The percentage is relative to the baseline score, not a change in win probability. It is unavailable when the baseline is zero.

These comparisons describe recorded outcomes. They do not show that choosing an opening caused a better result. Sparse samples contain few matching games; passing the sample threshold does not guarantee reliability.

Historical outcomes

—

matching games

Loading results…

Outcomes by country among matching games
Country	Solo	In draw	Survived	Eliminated	Mean score

Survived excludes Solo and In draw. The four outcomes total 100% before rounding. A solid neutral cell marks each column’s best value; crimson marks its worst. Higher is better except for Eliminated. Ties share highlights; equal-value columns stay neutral.

Outcome definitions and limits

Solo: Finished with at least 18 supply centers.
In draw: Held supply centers in a game classified as a draw.
Survived: Held supply centers when another country soloed.
Eliminated: Finished with no supply centers.

Each table row describes that country in the matching games. Each card describes games with those orders. With one game, Yes/No shows its recorded outcome.

Draws are inferred from empty orders at the end of the record, not confirmed site results. Games with unknown endings are excluded, so these results describe only the included sample.

A match requires all selected orders in the same game. Other countries and turns may differ. Later turns reflect earlier events; identical selected orders can still occur on different boards.