Gentleman Johnny

On June 13th, 1777, “Gentleman Johnny” Burgoyne and Major General Guy Carleton inspected the forces of Great Britain assembled at Saint-Jean-sur-Richelieu and about to embark upon and invasion of the American colonies from Canada. The force consisted of approximately 7,000 soldiers and 130 artillery pieces and was to travel southward through New York, by water and through the wilderness, to meet up with a second force moving north from New York City. The act of capturing the Hudson River Valley and, in particular, the city of Albany, would divide New England from the mid-Atlantic colonies, facilitating the defeat of the rebel forces in detail.

Poor communication may have doomed the plan from the start. The army which Burgoyne counted on moving up from New York City, under the command of General Howe, was committed to an attack on Philadelphia, to be executed via a southern approach. Thus, when it needed to be available to move north, Howe’s army would be separated from the upstate New York theater not only by distance, but also by George Washington. Burgoyne did not receive this important information and set out on his expedition unaware of this flaw.

Nonetheless, Burgoyne began well enough. As he moved southward, the colonial forces were unaware of his intent and strength and friendly native forces screened his army from observation. He captured Crown Point without opposition and successfully besieged and occupied Fort Ticonderoga. Following these successes, he embarked on an overland route from Skenesboro (at the southern reaches of Lake Champlain) to Fort Edward, where the Hudson River makes its turn south. This decision seemed to have been taken so as to avoid moving back northward, a retrograde movement necessary to use Lake George’s waterway. It may well also indicate Burgoyne’s lack of appreciation for the upstate New York terrain and its potential to allow a smaller colonial force to impede his movements.

Live Free or Die

In order to deal with the enemy blocking his path, Burgoyne sent forth his allied Indian forces to engage and run off the colonials. Having done so, they proceeded to loot and pillage the scattering of colonial settlements in the area. This had the perverse effect of driving otherwise-neutral locals into the rebel camp. As the fighting portion of his army made the trek to Fort Edward rather rapidly and uneventfully, Burgoyne discovered he had two serious issues. First, he finally received communication from Howe informing him that the bulk of the New York army, the forces with whom Burgoyne was planning to rendezvous, were on their way by sea to south of Philadelphia. Second, the movement through the wilderness had delayed his supply train, unsuited as it was to movement through through primal woodland.

Burgoyne’s solution was to again pause and to attempt to “live off the land” – requisitioning supplies and draft animals from the nearby settlers. Burgoyne also identified a supply depot at Bennington (Vermont) and directed a detachment to seize its bounty. What he didn’t know is that the settlers of Vermont had already successfully appealed to the government of New Hampshire for relief. New Hampshire’s General John Stark had, in less than a week’s time, assembled roughly 10% of New Hampshire’s fighting-age population to field a militia force of approximately 1,500.

When Burgoyne’s detachment arrived at Bennington, they found waiting for them a rebel militia more than twice their number. After some weather-induced delay, Stark’s force executed an envelopment of the British position, capturing many and killing his opposing commander. Meanwhile, reinforcements were arriving from both sides. The royal force arrived first and set upon the disarrayed colonial forces who were busy taking prisoners and gather up supplies. As Stark’s forces neared collapse, the Green Mountain Boys, under the command of Seth Warner, arrived and shored up the lines. The bloody engagement continued until nightfall, after which the royalists fell back to their main force, abandoning all their artillery.

Stark’s dramatic victory had several effects. First, it provided a shot in the arm for American morale, once again showing that the American militia forces were capable of standing up to the regular armies of Europe (Germans, in this case). Second, it had an opposing dilatory impact on the Indian tribe’s morale, causing the large, native force that Burgoyne had used for screening purposes to abandon him. Third, it created a folklore that persists in northern New England to this day. Stark became a hero with his various pre- and post- battle utterances preserved for the ages. Not the least of these was from a letter penned well after independence. Stark regretted that his ill health would prevent him from attending a Battle of Bennington veterans’ gathering. He closed his apology with the phrase, “Live free or die: Death is not the worst of evils,” which has been adopted as the official motto of the State of New Hampshire.


The delays put in the path of Burgoyne’s march gave the Colonials time to organize an opposition. New England’s rebellion found itself in a complex political environment, pitting the shock at the loss of Ticonderoga against the unexpected victory at Bennington. The result was a call-to-arms of the colonial militias which were assembled into a force of some 6,000 in the vicinity of Saratoga, New York. General Horatio Gates was dispatched by the Continental Congress to take charge of this force, which he did as of August 19th. His personality clashed with some of the other colonial generals including, perhaps most significantly, Philip Schuyler. Among the politicians in Philadelphia, Schuyler had taken much of the blame for the loss of Ticonderoga. Some even whispered accusations of treason. Schuyler departure deprived Gates with the knowledge of the area he was preparing to defend, hindering his efforts. Burgoyne focused his will to the south and was determined to capture Albany before winter set in. Going all-in, he abandoned the defense of his supply lines leading back northward and advanced his army towards Albany.

On September 7th, Gates moved northwards to establish his defense. He occupied terrain known as Bemis Heights, which commanded the road southward to Albany, and began fortifying the position. By September 18th, skirmishers from Burgoyne’s advancing army began to meet up against those of the colonists.

Having scouted the rebel lines, Burgoyne’s plan was to turn the rebel left. That left wing was under the command of one of Washington’s stars, General Benedict Arnold. Arnold and Gates were ill-suited for each other leaving Arnold to seek allies from among Schuyler’s old command structure, thus provoking even further conflict. Arnold’s keen eye and aggressive personality saw the weakness of the American left and he realized how Burgoyne might exploit it. He proposed to Gates that they advance from their position on the heights and meet Burgoyne in the thickly-wooded terrain, a move that would give an advantage to the militia. Gates, on the other hand, felt his best option was to fight from the entrenchments that he had prepared. After much debate, Gates authorized Arnold to reconnoiter the forward position where he encountered, and ultimately halted, the British advance in an engagement Freeman’s Farm.

Game Time

In my previous article, I talked about some new stuff I’d stumbled across in the realm of AI for chess. The reason I was stumbling around in the first place was a new game in which I’ve taken a keen interest. That game is Freeman’s Farm, from Worthington Games, and I find myself enamored with it. Unfortunately it is currently out of print (although they do appear to be gearing up for a second edition run).

How do I love thee? Let me count the ways. There are three different angles from which I view this game. In writing here today, I want to briefly address all three. To me, the game is interesting as a historical simulation, as a pastime/hobby, and as an exercise in “game theory.” These factors don’t always work in tandem, but I feel like they’ve all come together here – which is why I find myself so obsessed (particularly with the third aspect).

The downside for this game, piling on with some of the online reviews, is with its documentation. Even after two separate rule clarifications being put out by the developer, there remain ambiguities aplenty. The developer has explained that the manual was the product of the publisher and it seems like Worthington values brevity in their rule sets. In this case, Worthington may have taken it a bit too far. To me, though, this isn’t a deal-breaker. The combination of posted clarifications, online discussion, and the possibility of making house rules leave the game quite playable and one hopes that much improvement will be found in the second edition. Still, it is easy to see how a customer would be frustrated with a rule book that leaves so many questions unanswered.

Historical War Gaming

This product is part of the niche category that is historical wargaming. Games of this ilk invite very different measures of evaluation than other (and sometimes similar) board games. I suppose it goes without saying that a top metric for a historical wargame is how well it reflects the history. Does it accurately simulate the battle or war that is being portrayed? Does it effectively reproduce the command decisions that the generals and presidents may have made during the war in question? Alternatively, or maybe even more importantly, does is provide insight to the player about the event that is being modeled?

On this subject, I am not well placed to grade Freeman’s Farm. What I will say is that the designer created this game as an attempt to address these issues of realism and historicity. In his design notes, he explains how the game came about. He was writing a piece on historical games for which he was focusing on the Saratoga Campaign. As research, he studied “all the published games” addressing the topic, and found them to be lacking something(s).

I’ll not bother to restate the designer’s own words (which can be accessed directly via the publishers website). What is worth noting is that he has used a number of non-conventional mechanisms. The variable number of dice and the re-rolling option are not exactly unique, but they do tend to be associated with “wargame lite” designs or other “non-serious” games. Likewise, the heavy reliance on cards is a feature that does not cry out “simulation.” That said, I am not going to be too quick to judge. Probability curves are probability curves and all the different ways to roll the dice have their own pros and their own cons. Freeman’s Farm‘s method allows players to easily understand which factors are important but makes it very difficult to calculate exactly what are the optimal tactics. Compare and contrast, for example, to the gamey moves required to get into the next higher odds column on a traditional combat results table.

Playing for Fun

All the above aside, a game still needs to be playable and fun to succeed. We seem to be living through a renaissance in the board gaming world, at least among the hobbyists that have the requisite depth of appreciation. A vast number of well-designed and sophisticated board games are produced every year covering a huge expanse of themes. More importantly, ideas about what makes a game “fun” and “playable” have evolved such that today’s games are almost universally better than the games of a generation or two ago. Gone are the days (such as when I was growing up) when slapping a hot franchise onto a roll-the-dice-and-move-your-piece game was considered effective and successful game design. You can still buy such a thing, I suppose, but you can also indulge yourself with dozens and dozens of games based on design concepts invented and refined over the last couple of decades.

From this standpoint, Freeman’s Farm also seems to have hit the mark. It is a nice looking game using wooden blocks and a period-evoking, high quality game board. I’ve read some complaints on line about boxes having missing (or the wrong) pieces in them. This would definitely be a problem if you don’t notice it and try to play with, for example, the wrong mix of cards. The publisher seems to be responsive and quick to get people the material they need.

Game Theory

The real reason I’m writing about this now is because the game has a form that seems, at least to my eyes, to be a very interesting one from a theoretical standpoint. Contrast to, say, chess, and you’ll notice there are very few “spaces” on this game’s board. Furthermore, the movement of pieces between spaces is quite restricted. This all suggests to me that the model of the decision making in this game (e.g. a decision tree) would have a simplicity not typical for what we might call “serious” wargames.

Given the focus of that last post, I think this game would get some interesting results from the kind of modeling that AlphaZero used so successfully for chess. Of course, it is also wildly different from the games upon which that project has focused. The two most obvious deviations are that Freeman’s Farm uses extensive random elements (drawn cards, rolled dice) and that the game is not symmetric and non-alternating. My theory, here, is that the complex elements of the game will still adhere to the behavior of that core, and simple, tree. To make an analogy to industrial control, the game might just behave like a process with extensive noise obscuring its much-simpler function. If true, this is well within the strengths of the AI elements of Alpha Zero – namely the neural-net enhanced tree search.

Momentum and Morale

A key element to all three of these aspects revolves around rethinking on how to model command and control. It is novel enough that I wrap up this piece by considering this mechanism in detail. In place of some tried-and-true method for representing command, this game uses blocks as form of currency – blocks that players accumulate and then spent over the course of the game. Freeman’s Farm calls this momentum; a term helps illustrate its use. From a battlefield modeling and simulation standpoint, though, I’m not sure the term quite captures all that it does. Rather, the blocks are a sort of a catch-all for the elements that contribute to successful management of your army during the battle, looking at it from the perspective of a supreme commander. They are battlefield intelligence, they are focus and intent, and they are other phrases you’ve heard used to describe the art of command. Maybe the process of accumulating blocks represents getting inside your enemy’s OODA and the spending blocks is the exploitation of that advantage.

Most other elements in Freeman’s Farm only drain away as time goes by. For example, your units can lose strength and they can lose morale, but they can’t regain it (a special rule here and there aside). You have a finite number of activations, after which a unit is spent – done for the day, at least in an offensive capacity. Momentum, by contrast, builds as the game goes on – either to be saved up for a final push towards victory or dribbled out here and there to improve your game.

Now, I don’t want to go down a rabbit hole of trying to impose meaning where there probably isn’t any. What does it mean to decide that you don’t like where things are going and your going to sacrifice a bit of morale to achieve a better kill ratio? Although one can construct a narrative as to what that might mean (and maybe by doing so, you’ll enjoy the game more), that doesn’t mean it is all simulation. The point is, from a game standpoint, the designer has created a neat mechanism to engage the players in the process of rolling for combat results. It also allows a player to become increasingly invested in their game, even as it is taking away decisions they can make because their units have become engaged, weakened, and demoralized.

I’m going to want to come back and address this idea of modeling the game as a decision tree. How well can we apply the either decision trees and/or neural networks to the evaluation of game play? Is this, indeed, a simple yet practical application of these techniques? Or does the game’s apparent simplicity obscure a far-more complex reality that prevents the application of these computer learning techniques by being applied by someone who doesn’t have Deep Mind/Google’s computing resources? Maybe I’ll be able to answer some of these questions for myself.