Twenty years and almost no update

The question everyone asks when they return to a game that is "still good after all these years" is: how many years is that, really? Twenty since the release, without a single update? The short answer: yes, twenty, and almost no updates. The longer answer is more interesting, because the year of release cannot be found by looking at the first link.

Dice Wars in six sentences

It is played on a random map made of irregular territories that were formed by joining small hexagons. Each territory holds a stack of dice, at most eight. You can attack only a neighbouring enemy territory, and only from a stack of at least two dice. Both sides roll all the dice in their stack, the higher sum wins, and in a tie the defender wins. After a victory all the dice but one move to the conquered territory; after a defeat the attacker is left with a single die. At the end of your turn you get new dice according to the size of the largest connected area you control, and they are distributed at random.

These rules are everything. The map is different in every game, you play against the computer, and it is one of those games that can be explained in a minute and learned in an evening. According to a description on the bit-tech site from 2010, you can easily lose a couple of hours to it.

A detective story about the year of release

The original game was made by a Japanese developer who publishes games under the label GAMEDESIGN (gamedesign.jp). Several sources give Taro Ito as its author. This comes from the JayIsGames review, from Wikipedia and from the credits of unofficial versions. The official pages do not name him, so the name should be taken with reserve.

The year of release is harder. Five different numbers appear on the internet, and each has a reason:

The year 2006 is therefore the one that has support in the archive and in the contemporary reviews, and it makes twenty years and four months. Anyone who counts the Windows version from July 2004 gets 22 years. The figure of 25 years, which the year 2001 would give, is a trap.

Really no updates?

Almost. The web archive keeps three different builds of the Flash file: from 6 July 2006 (98,171 bytes), from 8 November 2006 (98,268 bytes) and from 31 August 2007 (100,394 bytes). The last one added code for displaying the loading progress and embedded fonts. What changed in the game itself is not known: the file was not taken apart.

Since August 2007, according to the archive, not a single byte has changed. The file the server offers today has the same size. So the game has stayed essentially untouched for about nineteen years, although for the first fourteen months it was still being tuned.

In 2016 the author also released his own rewrite in HTML5, which refers to the original Flash file as "the previous version". It is new code. Whether the rules are identical has not been verified, although the code matches the rules from the review of 2006. No official sequel or port for mobile phones, Steam or consoles was found (the app stores and the web were searched; absence of evidence is not evidence of absence).

And the end of Flash? Adobe switched it off on 31 December 2020. The game survived because the author already had an HTML5 version, and the Flashpoint archive saved it before the end (a record of 4 May 2018).

What happened around it

Dice Wars had no official sequel, but it had descendants:

• KDice (1 December 2006): a multiplayer game by Ryan Dewsbury, modelled on Dice Wars. JayIsGames gave it its award for the best multiplayer game of 2006.
• Dicewars DS (2007): an unofficial game for the Nintendo DS with play over Wi-Fi for up to seven players.
• Open clones: versions in Qt (2015), LibGDX (2016), Rust (2017) and Godot (2022), and a bachelor's thesis from the Faculty of Information Technology of Brno University of Technology (VUT, 2018), from which a client-server game in Python with five levels of AI came out.
• Mobile clones, unofficial, for example Dice Wars Lite or Conquer – Epic of Dice Wars.

The clones suggest that the rules are so clear that they can be written again from a description. That not one of them replaced the original is an impression, not a measurement.

What makes it unique

According to the JayIsGames review of June 2006 (a rating of 4.7 out of 5 from 235 voters) it was an excellent single-player game for a few minutes that lacked a multiplayer mode. Three things that seem important from the rules and from the reviews:

1. The rules are small and the depth is great. One sentence of rules and a few dozen decisions in every game.
2. The map is different in every game. There is no "learned" map that you can memorise.
3. The opponent feels human enough. Its artificial intelligence is, according to the code of the HTML5 version, simple: it picks a random attack among those where the attacker has at least as many dice as the defender, and if one player holds more than two fifths of all the dice, it attacks only them. It looks like strategy and often that is enough. (What the AI did in the Flash version was not checked.)

To this it must be added that Dice Wars is a game about risk. Almost every decision is a bet: attack with a 60% chance, or wait? That bet is the reason the game is remembered. It is also the reason why some players hate it.

What it lacks for today's player

What "today's player" wants cannot be said with a single number. From the article When the game was finished, however, a few facts follow: more than half of playing time belongs to older games, the twenty most played games hold about 58% of the time on PC, and 62% of American adults watch short video at least a few times a week. A game that wants to catch on in such attention needs the kit that is ordinary today and that Flash in 2006 could not have had:

"Not mentioned" only means that it is not in the sources that were read. The original game was not examined in every detail.

This does not mean that the original game is bad. It means that the core is fine and the kit around it is missing. And that can be added, if the core is kept.

What Audamir is

Audamir is a single-player web strategy game with dice, inspired by the original. Plain HTML, CSS and JavaScript, no dependencies and no server; it runs offline and can be installed. The rules of the Principalities^[1] mode match the original (eight dice per territory, the largest connected area, a tie goes to the defender, random irregular territories). Everything else is extra kit:

• three levels of opponent, measured by tournaments,
• 18 minigames and 28 puzzles without chance,
• a campaign with 13 missions and bosses, the Iron Run and the Threat ladder,
• the Avatar^[2], the player's own die that comments on the game, and the Collection,
• 272 steps of achievements, a daily challenge, a record and an analysis of a game,
• five languages (three of them are so far machine drafts marked beta) and accessibility.

Forever is a planned extension of the game: an endless run on generated missions, which does not exist yet (the details will come in a separate article). The plan of extensions has 33 chapters today, and more are added as play and measurement show.

The unexpected thing is the pace. The whole game came into being from 17 October 2026, in about three and a half days, together with an assistant. How that went is told in the articles of the Development diary category, the very first of them about reconstructing the beginnings from copies of a folder.

Update of 2 November 2026

On 2 November 2026 both files of the official HTML5 version were read in full and repeatedly; the description is in the article What is in the source code of the original (forthcoming article no. 42). The original text of this article stays as it was, except for the places that were changed or left out on 4 November 2026 (the name of the mode is explained in the note at the end of this section); four places are made more precise.

• "Whether the rules are identical has not been verified" (section Really no updates?). The rules of combat and reinforcement in the code of the HTML5 version match what the reviews of 2006 described: a tie goes to the defender, after a win all the dice but one move over, new dice are given by the largest connected group, and the reserve is at most 64. The match with the Flash version of 2006 was not checked, though, because only the version of 2016 was read.
• "The rules of the Principalities mode match the original (… random irregular territories)" (section What Audamir is). That holds for combat, the tie and reinforcement, not for everything in the brackets and not for everything else. The original does not deal the territories to the players at random but in the order of their birth (so the human always gets the first territory); its map does not come from gluing but from the growth of territories one at a time from a core and a ring (a simulation from the description: 30.0 territories of 20.5 cells, 5,000 maps); and its opponent has rules of its own. Principalities do not copy the procedure of the original. The copy is the separate game Tribes, which has been in the development version of the game since 2 November 2026 (at the time of writing 0.90.5; Tribes: a copy that can be measured (forthcoming article no. 43)).
• The opponent (section What makes it unique and the table row "Opponent: one level"). "One level" is right. The rule of two fifths is more exactly this, though: when someone holds more than two fifths of the dice, the others attack only them, and that player attacks anyone (not "attacks only them"), and when the numbers of dice are equal the attack happens with a probability of 0.9, not always. The opponent also includes a wrong sorting of the "leader"; the details are in the article What is in the source code of the original (forthcoming article no. 42).
• "Dice Wars in six sentences". The sentence about the territories that were "formed by joining small hexagons" is approximate: a territory arises by growth from a core and a ring.

Sources and notes

• Year of release and version history (web archive, search of 20 Oct 2026): change log of the site of 25 Oct 2005 (launch of the site 12 Apr 2004), the Windows game Dice Strategy of 2004, the list of games of 12 May 2006 (without the Flash version) and of 22 June 2006 (with it), the first record of the page of the game of 23 Jun 2006, the builds of the file 0 (6 Jul 2006, 8 Nov 2006, 31 Aug 2007).
• The official HTML5 version with the line about the year 2001 and the rules and the simple AI in the source code; the main page refers to the Flash version as the previous one.
• Reviews: JayIsGames, 26 Jun 2006 (4.7 out of 5, 235 votes), Gigazine, May 2007, bit-tech, 23 Oct 2010.
• Descendants: KDice on Wikipedia (1 Dec 2006), Dicewars DS (version 0.4.0 of 11 Oct 2007), clones on GitHub: Qt, LibGDX, Python from the VUT bachelor's thesis and the documentation for it. The existence of the clones is verified, their other properties are not.
• The end of Flash: Adobe, the page of the end of support (31 Dec 2020), the record in the Flashpoint database (4 May 2018).
• Audamir is an independent game inspired by the original. The data on its development are from the git history and the documentation of the project.