A few months ago, on ConSimWorld’s For The People discussion board, Taylor Golding posed the following question to me. My answer may be of mild interest to some. But in any case, I’m reposting it here so I have a record of it.

Does For the People completely reflect the strategic friction experienced by both sides in the American Civil War, especially friction as evidenced by organizational challenges, key assignments and appointments, and external events?

A famous nursery rhyme, based on the defeat of King Richard III of England at the battle of Bosworth Field in 1485, postulates the importance of small details and chance:

For want of a nail, a shoe was lost,
For want of a shoe, a horse was lost,
For want of a horse, a battle was lost,
For want of a battle, a kingdom was lost,
And all for want of a horseshoe nail.

So wrote George Herbert in 1640. Two hundred years later Clausewitz captures this incident and uncountable others in his famous exposition on the nature of friction in the conduct of operations. I have always interpreted this friction to be a factor on the “friendly” side of the issue, even while realizing that the enemy is subject to his own friction. Over the intervening century and a half a number of really smart people continue to consider Clausewitzian friction during the study of military (and any organizational) dynamics. Colonel John Boyd folds the concept into his famous “Patterns of Conflict” briefing and thus influences contemporary thought with his OODA loop paradigm. By 1988 even modern mathematics, as popularized by James Gleick in his book Chaos, seems to focus on how seemingly small details can influence great events – the so called “Butterfly Effect”.

But in the very same book, Gleick also provides the counterpoint to total chaos in natural and man-made systems, which is observably not present in all natural nor man-made systems. I’m speaking of Lorenz’ “Strange Attractors” and Mandelbrot sets. Even to mathematical dilettantes like myself it is clear that Herbert’s deterministic view that every cause, no matter how small, has an effect doesn’t really capture reality. Small perturbations can seemingly influence events, but ultimately these perturbations… in most, but not ALL systems… get swallowed up by grosser influences – the strange attractor.

Which is a good thing for us operations analysts or we might as well wrap fish entrails around our fingers and howl at the moon to make our predictions. Instead we rely (often, but no longer always) on simulations. And while the giant simulations we used during the Cold War to analyze systems, strategies, and concepts used random numbers to create a certain variance in the results between runs, the fact of the matter is that the basic simulations were pretty deterministic. Battle outcome was predicated on the numbers and types of platforms on each side, rates of fire, expected lethality of weapons and expected reductions in vulnerability of targets. Behind this approach is the basic assumption that if the analyst could just capture all the factors and get their influences modeled properly (including probabilistic (i.e. stochastic) representations at this level), then the result would be a valid prediction of the hypothetical future given that series of inputs.

And so it seemed to be. The models were inaccurate in replicating the results of the Six-Day and Yom Kippur Arab Israeli wars, but not so inaccurate that the differences couldn’t be excused by normal mathematical variance or differences between the Israeli’s & NATO and the Arabs & the Warsaw Pact in terms of doctrine, training, planning and organization. It wasn’t really until the 1991 Gulf War that the Cold War models were cruelly exposed in their inadequacies. None foresaw the total dominance of the battlefield by the Allied powers that resulted in such a short war. Of even more concern was that the casualty predictions were off (thankfully!) by two orders of magnitude.

Now for those who believe that every effect must have a cause, these observations mean that new simulations must be created that use the factors that actually influence the outcome of a battle, campaign or war. But to me, it means that the outcome of a war rests in a realm of human behavior that is far less amenable to accurate mathematical expression… that the “big factors” such as popular support, morale, training efficacy, political ramifications, public perception must be included as well as logistics, command and control issues. In fact, I think the least important factor is the traditional Killer-Victim scoreboard on the battlefield.

So after a page and half of philosophical pontification, most of it abbreviated and no more than a wordy train of consciousness… or more accurately, what I use in place of a train of consciousness…, I’m ready to consider For the People.

For the People is not a simulation of the American Civil War. It is a game that simulates the ACW. Now designer Mark Herman may, and probably should, disagree with me on this, but I’m going to draw the distinction as I see it. If FTP were a simulation, I do not believe that the South could win, because… in historical retrospect… there was no way that the South could accomplish its goals of independence. Now that bald statement has been and will be long discussed because there are still imponderables. The most telling of which is to ask, under what conditions would the population of the North have acceded to the division of the Union? And I’m willing to kick that question around as a discussion, but for the purpose of this paper, let’s leave it as the most extreme example of an unmodelable gross factor that interferes with simulating the conflict.

If you’ll accept the fact that the big factors, not only northern political will, but population, infrastructure, cultural influences, diplomatic exchanges and so forth… and not only by their direct effects, but by the interactions of all these factors from the second to the Nth order effects…, must be included in a simulation, then you’ll have accept that the simulation would be unimaginably large and complex. But worse, it’s be boring as hell.

If one is designing, publishing and selling games, then the last thing one wants is boring. But at the strategic level, to get players to accept the game as a depiction of an historical event, then in some fashion the game must make at least passing reference to all the big factors just mentioned, while still focusing on the stuff that the market audience finds interesting.

And what does the market find intriguing about the ACW? What do they believe at the emotional level? What aspects get the juices flowing? Different things for different people obviously, but I think that some common views are: the honorable “Lost Cause”, the perception of CSA leadership superiority, the overwhelming strength of the Northern in population and industrial power, the importance of Emancipation, “King Cotton”, and at least in Texas the article of faith that any Johnny Reb could outfight three Billy Yanks at the tactical level any day of the week, despite all evidence to the contrary.

So to satisfy the market, the game must address these items in some fashion. It is not strictly necessary that the designer be totally congruent with a purchaser’s view… indeed it’d be impossible to be congruent with more than one, much less many. But if the factors aren’t there somehow, then the game can only “make it” as a pure game. (The approach taken by A House Divided.) But the engine in a Card Driven Game, the cards themselves, allow the designer to alter the board condition/position arbitrarily and then attribute the effect to some cause that the players’ recognize as an important factor in the war.

But note that what is actually being effected… the part of the conflict that captures our interest and imagination -- the dashing Stuart, the steady Longstreet, the dour, willful and bold Jackson, and of course the brilliant, gracious gentleman astride Traveler. Instead of the military campaign influencing the diplomatic, economic and cultural foundations of each side; what we see is the counter-reflection of the big picture as it is imperfectly projected, as thru poor glass, back onto the military picture. Now, admittedly, in FTP this is not as extreme as in some games because we take all N dimensions and project them into a one dimensional plane called “Strategic Will”. The mathematician in me rebels at the gross simplification. The wargamer in me exalts. The wargamer doesn’t want to take time to feed factors into a computer, putz with probability distributions, or need to keep a database of all the troops and logistics and precinct polls. The wargamer wants to mull over the benefits of using a campaign card for a special Naval move by Grant instead of two less bold moves. And when he decides to “run the guns”, he vicariously experiences a shadow of the thrill and apprehension that must have been present when Grant, thru a cloud of cigar smoke and over a half full bottle of whiskey, forced the decision down McPherson’s and Sherman’s throats. (This moment is recounted in both Grant’s and Sherman’s memoirs, and it’s strikingly similar. Sherman was convinced that this was the height of folly. I recommend both books to even the casual reader.)

Based on my conversations with Mark in play testing FTP, EOTS and a little bit of Washington’s War (let’s not go there!), I believe that he designs intuitively. He starts with the big factors in mind, but loves the chrome as much as the next guy. It’s this instinctive balancing of what will be fun to have in a game (FTP cavalry counters come to mind) and what is necessary to make the game behave somewhat like the actual combatants were constrained to behave due to the big issues.

So having established (thru long winded rhetoric) that FTP is a game that simulates, and not a simulation in the sense of an analytical tool, does the game capture the chaos and the friction of the ACW?

Brian Sielski argues semantically on ConSimWorld that there is no way that that a handful of cards can capture every possible piece of friction. In one sense he’s correct. But that perspective has its roots in the front end of the deterministic simulation approach with which I opened this discussion. The inherent assumption that if we could (as an intellectual experiment) identify every single factor, that if we could we relate them perfectly to each other, that then all chance would be removed from our model and we could predict perfectly the outcome. Not much of a game, but a great simulation.

But I argue that even if we could determine every factor and every influence of each factor on all others, that we would still not be able to predict the outcome because the nature of the universe is that these factors themselves behave probabilistically.

This discussion has been going on since mankind climbed out of the trees. In religious circles it manifests itself as free-will or predestination. In mathematics and physics as a deterministic or stochastic universe. I don’t think we’re going to decide it here; the two points of view will remain. I don’t think I’m going to convince anyone of my point of view. I am just stating how I think because I was asked. And I assume Taylor Golding already knew most of what I would say and asked just because the CSW board was in a quiet mode and needed to be stirred up.

I think that For the People adequately captures strategic and operational friction. Ignore the labels on the cards… random events occur that influence the military situation and the resulting Strategic Will. In some cases the events can be turned to our benefit and in some we must roll with the punch from an apparently malicious fate. In each battle and other key cusps (interception, retreats, blockades, leader losses) there is an unpredictable outcome due to the roll of dice. When a battle roll comes up “six” then Chamberlain refuses the left and counterattacks at just the right moment. When it comes up “1”, a brigade is advanced too far into the peach orchard and unhinges a strong position. All of that is Clausewitzian friction, the die roll just sums up all of the miscommunications, lack of situational awareness, the unanticipated fence just at the limit of effective enemy fire and all the flapping of millions of butterfly wings all over the world.

How more accurate can the functioning of random nature be? And yet, though the events are random, we still have enough control of the system to influence it toward one or another strange attractor… one defined as Confederate victory, the other as Union victory.

Views: 97

Replies to This Discussion

Great post, Don. Would like to see something like this as an article submission for F&M if you have time for it.

BTW, have you read The Black Swan (author escapes me just now)? Peter Perla recommended it to me and it's quite an interesting read. Gleick's book is excellent also.

As a fellow mathematical modeler, I've also spent a lot of time pondering these issues. Although I've not played FTP, I think your commentary here is more universal. I've observed several things that I think are important to consider.

1. Many people I work with and deal with regularly are convinced that the purpose of modeling is to predict the future. As you stated about the gulf war, modeling is an abysmal failure at prediction, except that it's better than just about anything else. Steven Rieber (Intelligence Analysis and Judgmental Calibration, from International Journal of Intelligence and CounterIntelligence, 2004) did research into the value of predictions via expert analysis and found that while modeling does poorly, gray beards and pundits do infinitely worse. The take away is the recognition that models provide information about relationships between variables in the environment. Understanding these relationships can help mitigate the impact of unforeseen events, and that's about all. Using tools like modeling for prediction suffers from what economists call the Lucas critique, or more metaphorically, the uncertainty principle of Heisenberg. Changing your course changes the conditions, thus invalidates the prediction. This is a crucial recognition.

2. I think there is a conflation, generally, not just in your article, of games, simulations, and models. As you state. FTP is a game that simulates the Civil War. To some extent I disagree with this, although perhaps this is semantic in nature. I would be more inclined to state that the game (any wargame really) simulates the constraints of an event, and thus affords the players the experience of those historical constraints within which to make decisions. I think this is an important distinction. In modeling, what we are doing is exploring the relationship between variables and studying the impact of stochastic events on the system. We can then use this method to inform the constraints we create in a game, but a model cannot be a game. A simulation is a model run based upon some set of initial conditions. A game is a decision-making environment, which can be informed by models and simulations, but is not, to my mind, a simulation in anyway other than its simulation of environmental constraints. And since prediction is part of the discussion, a game is also not a great predictor, but it is a great tool for understanding the constraints of an environment and a great tool to consider the mind of one's opponent, and that's really the crux. While the mathematician in me also wants to quantify, the critical thing to understand about a game is that it's purpose is to examine decisions under constraint, and thus, to glimpse the mind of any potential opponent. And that is completely unquantifiable.

3. Models and other quantitative methods can never completely incorporate decision-making (creative thinking, panic, heroism, cowardice) in the context of environmental constraint. To my mind, this is the great gulf that exists between wargamers and modelers. My experience in the modeling world has shown me a group that rejects anything that is not quantifiable. Where data doesn't exist, the issue is often simply ignored or given a bell-curve. The issue here can be summed up by something a fellow modeler once said to me, "the great thing about knowing math is that you don't have to read as much." And this is why the large ocean of theoretical constructs is usually ignored, despite its ability to usefully fill in the holes where hard data is absent. But the problem for quants is worse than that. Because of the lack of theoretical immersion, much of the quantitative work is done sans a theoretical constru
Looks as though much of my post was cut off once I could no longer edit it. I reported it to JK. I don't have a copy but will recompose from memory...

...sans theoretical construct. Thus, empirical relationships are discovered that have no theoretical justification and utilized despite this shortcoming. Barefoot empiricism is all too common in mathematical models.

But Modelers are not wholly at fault. I've seen posts on CSW as recently as a few months ago that dismissed quantitative methods entirely. I can't provide a reason for this thinking as frankly I can't fathom it. Fear of math seems the most likely culprit to my mind, as description of why quantitative analysis is faulty is usually couched in perceptions about the method that are completely wrong. The quant vs. qual debate is also alive and well in academia, and frankly I can't imagine a stupider debate. The two are completely symbiotic, yet practitioners of one form or another insist that their method is the correct one.

NTL, the truth is that wargaming is a qualitative method that is (should be) informed by quantitative research (models, etc.).

4. Your metaphoric use of strange attractors is brilliant!
Thanks for calling me brilliant. I appreciate it, but you are putting your analytical and observational reputation seriously at risk!

I'm not sure where I picked up the idea of applying strange attractors to wargaming, but I am sure that it's not an original perspective.

As for your observations comparing experts to simulations, I quite totally agree. My opinions on this are strongly influenced by Kahneman, Tversky and Slovic as well as Robyn Dawes. They're a bunch of cognitive psychologists that have done work in decision making and decision theory. (Actually two different fields, the first is observational of human behavior, the second is the more normative mathematical tools with which members of this discussion group are familar.) In many respects these two schools can be loosely group in the fashion you used, qualitative vs. quantitative.

To overly condense the work of these four scholars and probably hundreds of supporting grad school students, one gets down to this key observation. Humans are really good at determining what factors are important in making a decision, and frankly are reliably terrible at integrating those factors when making a decision. Mathematical models are great at integrating the factors, but totally atrocious at determining which factors are important to making the decision. The later are usually manifested by data collection followed by regression models, factor analyses and so forth. The model (and mathematicians who refuse to read) have no way to sort the spurious correlations from the actual ones. They lose Power (in the statistical sense) by leaning over backwards to avoid the Type I error. But that's just the start of it.

The solution, of course, is to use a hybrid approach where humans do what they do best, and math models (and their computers -- both silicon and carbon based) do what they do best. That happens in some environments, but because there are stove pipes in academia that prevent the scholars from synergizing, it'll be awhile before this becomes a commonly accepted practice.
To some extent I think wargames are, in fact, a hybrid approach, as you describe.

Any chance you'd be willing to synthesize your original post into an F&M article?
I honored that you think it is worthy.

Unfortunately, I'm quite busy at the moment with some other activities. However, that's not ruling out an article in the future, but it would have to be the indefinite future.

Are you planning to be at WBC this year, Jon? One of the items on my plate is putting together an hour long presentation on Wargaming as a Profession. For my preferred format, I hope to rope a number of pro's into constituting a discussion group. That would definitely give you a "quid pro quo" to hold over my head!

For that matter, any other practicing (or retired) OA's out there?
Alas I believe my wife would have something quite negative to say about it at this stage.


Help Center

Latest Activity

Robert Johnson commented on neal ekengren's group Florida Wargamers
"Created a new group for the new community…"
4 hours ago
Profile IconAlex Gerazounis and Chris Obonsawin joined ConsimWorld
5 hours ago
John Kranz posted a blog post
9 hours ago
Profile IconSteen Kastoft Hansen and Gerard Vandervelden joined ConsimWorld
David left a comment for Jamey Cribbs
"I will look forward to meeting to play some wargames once we get past the pandemic. We met some…"
Jan 19
David left a comment for Shawn King
"I noticed you played Enemy Coast Ahead. Which of the games have you played - Doolitte Raid or…"
Jan 19
John Kranz posted a blog post

Early Access: CSW Community

We are now open for early access for our all-new CSW Community.We invite you to join today for…See More
Jan 19
John Kranz commented on John Kranz's blog post Moving to New Platform
"We're now providing early access to our new CSW Community platform. Anyone can sign-up for…"
Jan 19



CSW Related Links

Please be sure to check-out these CSW services.

© 2021   Created by John Kranz.   Powered by

Badges  |  Report an Issue  |  Terms of Service