Software, engines, and artwork

Third-party disclosures

DiploBench relies on open-source software to run the tournament and adapted artwork to render its replays. These are the exact versions, sources, authors, and licenses used by this release.

Game engine1.1.2

python-diplomacy · AGPL-3.0-or-later

Gunboat baselineMIT

RuleBasedPlayer / DumbBot ruleset

Replay mapCC BY-SA

Diplomacy.svg · version 3.0

Order notationMIT

vizdip · version 0.3.0

01

python-diplomacy

Game engine and adjudicator

DiploBench pins diplomacy==1.1.2 as its game engine. The harness uses it to create standard games, enumerate legal orders, accept validated orders, advance phases, and adjudicate movement, retreats, and adjustments.

Version 1.1.2 was published by Philip Paquette and is licensed under the GNU Affero General Public License v3 or later.

02

RuleBasedPlayer / DumbBot

Deterministic gunboat baseline

DiploBench’s src/diplobench/dumbbot.py is adapted from the DumbBot ruleset in diplomacy/research, commit 0ce19e236cf9bb602510daef1fdaa2067a02f204.

The adapted implementation is Copyright © 2026 Andrew Lavery. The upstream work remains Copyright © 2019 Philip Paquette and is licensed under the MIT License.

gunboat-normal parameters12 effective values · all defaults

This entrant supplies no tuning overrides. These are the complete effective DumbBotTuning defaults used in the tournament.

spring_attack_weight = 700.0
fall_attack_weight = 600.0
spring_defense_weight = 300.0
fall_defense_weight = 400.0
spring_strength_weight = 1000.0
fall_strength_weight = 1000.0
spring_competition_weight = 1000.0
fall_competition_weight = 1000.0
build_defense_weight = 1000.0
remove_defense_weight = 1000.0
alternative_diff_modifier = 5.0
play_alternative = 0.5
gunboat-aggressive parameters12 effective values · 8 overrides

This entrant overrides the eight seasonal attack, defense, strength, and competition weights. The remaining four values inherit the normal defaults.

spring_attack_weight = 1100.0
fall_attack_weight = 1000.0
spring_defense_weight = 150.0
fall_defense_weight = 200.0
spring_strength_weight = 1200.0
fall_strength_weight = 1200.0
spring_competition_weight = 700.0
fall_competition_weight = 700.0
build_defense_weight = 1000.0
remove_defense_weight = 1000.0
alternative_diff_modifier = 5.0
play_alternative = 0.5
View full adapted DumbBot source850 lines · all gunboat parameters and order logic

This is the complete src/diplobench/dumbbot.py used by the tournament harness, including its default weights, configurable tuning fields, validation, factor maps, and movement, retreat, build, and disband selection code.

Open raw source ↗
03

Standard Diplomacy map

Adapted SVG replay board

The replay board uses an adapted version of Diplomacy.svg by Martin Asal, licensed under Creative Commons Attribution-ShareAlike 3.0 Unported.

DiploBench hides the source artwork’s starting-position units, normalizes SVG transforms for browser rendering, and applies live ownership colors and replay overlays. The adapted SVG remains available under CC BY-SA 3.0.

04

vizdip order notation

Replay overlay visual grammar

The replay overlay retains an order-rendering grammar adapted from vizdip 0.3.0 by KarmaMi. Movement arrows, hold rings, and differentiated support, convoy, retreat, build, and disband marks are implemented directly against the DiploBench replay schema.

DiploBench does not include the original React 15 package, its js-diplomacy runtime, or its map asset. The retained notation grammar is licensed under the MIT License.