Conflict simulation, peacebuilding, and development

Downes-Martin: Swarm gaming

Today at the US Army War College War Room, PAXsims associate editor Stephen Downes-Martin argues the need for “swarm gaming”—running many smaller wargames rather than one big one, to explore the decision and issue space in a more agile way.

One approach to deal with the constraint is open adjudication where the players participate with the adjudicators in determining the outcomes of interactions. The wargame becomes a structure within which the participants explore the novel scenario as they decide about novel warfighting concepts. The structure forces decision-making within a competitive environment followed by a cooperative exploration of possible outcomes, and this sequence is repeated as the game progresses.

The requirement can be satisfied by many small games run in parallel where each game is repeated multiple times with game design between each iteration modified by insights generated by the previous iteration. The iterations spawn multiple trajectories and create breadth across the decision and outcome space. Both Matrix Game and map/board based Hobby Game techniques can satisfy these requirements.

Each small wargame has one player per side and one adjudicator who also acts as a data recorder. Each subgame is played many times with players rotating between sides and the adjudication position. Rotating roles is critical for games that explore novel situations, as it forces players and adjudicators to see the situation from different perspectives and be innovative about adjudication. Repetition forces the players to think harder about how to win as they face players who have seen their previous attempts. Whether players stay in the same groups for all the subgames or are shuffled between subgames is an open question. I call this “Swarm Gaming” (not wargaming swarms, that is a different topic).

You can read his full piece at the link above.

10 responses to “Downes-Martin: Swarm gaming

  1. brtrain 11/10/2020 at 1:39 am


    I think you might find Neal Durando’s last four or five posts on his blog here very interesting about game development and the relation of developer to testing. Unfortunately Neal, a very smart guy, has not posted here for some time but what he has written is very good.

    For more on playtesting, I would name off the usual suspects in their usual books – Phil Sabin in Simulating War, James Dunnigan in the Complete Wargames Handbook, and Rules of Play – Game Design Fundamentals (Salen and Zimmerman) has some interesting parts too. They all bang on the same points – multiple iterations by multiple groups in parallel, the need for blind testing (perhaps not so much for professional wargaming, because you are not looking to develop watertight rules, just avoid game-breaking holes, tricks and omissions), and the pursuit of polishedness, so anyone can play the game with no help and stay interested.

  2. Stephen Downes-Martin 10/10/2020 at 2:50 pm

    DR — I deal with the issue of small group discussions and wisdom of crowds during wargames in my Connections North 2019 paper “Group Dynamics in Wargames and How to Exploit Them” (

  3. Stephen Downes-Martin 10/10/2020 at 2:40 pm

    Phil — I agree on the difficulty of doing facilitation/adjudication/data recording simultaneously. One can obviously add a data recorder who will also be able to note emotional ambience of the game, level of animosity or otherwise between the participants etc. which would be of value. It’s a matter of deciding how much one wants to pay for good data — and I’m all for good data!

  4. Stephen Downes-Martin 10/10/2020 at 2:36 pm

    Brian — Can you point me to a write up of commercial game play-testing best practice? I’d like to dig deeper into what we can learn from that (despite the different motives and desired outcomes).

    The 2014 UK Army Wargaming Symposium “looked” like it was doing swarm gaming (see photos at but I don’t know if data was being collected or role rotation was occurring.

  5. D R 10/10/2020 at 2:23 pm

    Stephen the reason for my hesitancy on using Matrix games adjudication for your plan is pretty much voiced by Phil Pournelle,
    (sorry for the delay in reply I had really important Rugby semi final watching to be done :)

    In essence if the results of ‘stuff ‘ : combat or contest results, success, failure, plausibility of an action etc, are decided by 3 people only, then the whole theoretical point of ‘wisdom of crowds’ adjudication (Matrix), eg diversity of opinion and balancing of arguments will, I suggest, rarely occur within such a small group approach.
    I personally have a lot of problems with Matrix as an adjudication method in large groups in any case, for a list of reasons not limited to: probable lack of subject matter knowledge of the players, rank or character bias, the start point for success being over 50% in some methods, and just the avoidance or ignorance of Inherent Military Probability (IMP) a la Burne.
    As you have pointed out in previous pieces about manipulation of results and bias, (conscious or unconscious), I fear Matrix as a method actually encourages that phenomenon – and if it is masked in a small group environment then that effect could be further amplified over a broad group of games.
    However rather than just bleat about it, I suggest that if the games are small and simple, and repeatable – then something like Brian Trains observation about using repeat playtesting methods would clear up some basic rules of thumb, mechanics and probabilities quickly, certainly within a few iterations, and so your results would be a bit more realistic over time than some of the usual Cyber Magic Missiles and Twitter barrages from space. David

  6. brtrain 10/10/2020 at 12:05 am

    Massively parallel wargaming, was my first thought… and why not?
    In a sense, this is in large part what we do with playtesting a commercial game.
    Many people play a game in pairs or small groups (also solo, but the point is–) independently of each other, testing the rules and situation as if it were a complex problem, looking for badly tuned parts of the machine that will make it fly apart, or for Stupid Gamer Tricks… anyway, things that are analogous to unexpected or overlooked events and contingencies.
    Then the groups report back, the designer and developer make changes to address the problems, and you do another iteration… and so on, for as long as you can keep people interested in the project (because they are not paid to be there, and have outside lives).
    The goal with a commercial game is a balanced, interesting enjoyable contest which of course is not the goal of a professional wargame, which is to generate a range of outcomes and responses to problems within the problem (and there are different ways to break apart the problem into manageable smaller bits you can play out).
    But there is no reason why the same processes cannot be used.

  7. Phillip Pournelle 09/10/2020 at 4:30 pm

    I like the concept but I think there will be some challenges.
    To start, the reason why a matrix style game is conducted is to gather the various experts together who have insights into elements of the problem. As an analyst, I’m keen on observing and collecting those argument for future analysis. The experts providing various perspectives on the problem are likely to advance our understanding of the problem.
    Swarm games may offer two opportunities after a series of matrix games. In the first case when enough matrix games (and post game analysis) has been conducted, your participants may have a reasonably level (equal?) knowledge base to then intelligently advance the ball independently. (and remove the group think Stephen is constantly fighting against). Alternatively the area of interest is so unknowable that the utility of swarm games is to generate a very large body of outcomes from semi-random outcomes for further analysis.
    The Game designer of Swarm Games must play close attention to the span of control capacity of the players. Quite often the reason why we have certain team structures (and why military echelons and structures exist) is to divide and effectively synchronize and employ forces. Pay close attention to how many elements a player must control, the range of actions each element can take, and the state each element can take on.
    Related to this is the challenge of having a rotating player take on both roles of adjudicator and rapporteur. I fear a lot of useful data will not be collected, I would advise keeping the rapporteur a separate person. This is especially true when the goal is to generate and collect a large set of argument for future analysis.

  8. Stephen Downes-Martin 09/10/2020 at 1:33 pm

    D R — I included Matrix gaming since it can be run with only three participants (red, blue, facilitator) and can be run quickly. Can you expand on your thinking about why that technique does not help? Thanks for your interest!

  9. D R 09/10/2020 at 12:46 pm

    Stephen, I agree with all of your article apart from the Matrix game comment – I do not think that technique, in general , helps achieve your objective…other than that – violently agree.

  10. Stephen Downes-Martin 09/10/2020 at 12:02 pm

    Here’s a mind map of the thinking that went into the article.

    I usually diagram out my thoughts, write the paper, then iterate between the two.

    (Many of the nodes are linked to external websites, so you might want to right-click on them and select “open in new tab” or whatever similar option is available in your browser.)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: