Home Store Products Research Design Strategy Support News

23November 2009

 Products The Guns of Gettysburg Design Diary The Times, They are A-Changin’

The Times, They are A-Changin’

Action at Austerlitz

Before talking about Gettysburg, I’m going to talk a little about Austerlitz. Austerlitz was, from a chronological perspective, a simple battle. It was characteritzed by steadily rising action, ascending to a peak, and then declining again until completely broken off. If we look at which formations were engaged, hour-by-hour, we can see this simple pattern (it is not always possible to determine exactly when everything happened at Austerlitz, still less what "engaged" means with any precision, but as rough-and-ready overview, I think this is about right):

8:00AM, 2 Dec.Ki,Le
9:00AM, 2 Dec.Ki,Pr,Lg,Le,Da
10:00AM, 2 Dec.Pr,Lg,Ko,Mi,Le,Da,SH,Va
11:00AM, 2 Dec.Pr,Lg,Ko,Mi,Ba,Li,Le,Da,SH,Va,La,Mu
12:00PM, 2 Dec.Pr,Lg,Ba,Li,Co,Le,Da,Va,La,Mu
1:00PM, 2 Dec.Pr,Lg,Ba,Li,Co,Le,Da,Va,La,Mu,Bs,Br
2:00PM, 2 Dec.Pr,Lg,Ki,Do,Le,Da,SH
3:00PM, 2 Dec.Pr,Lg,Ki,Do,Le,Da,SH,Va
4:00PM, 2 Dec.Ki,Do,Le,Da,SH,Va
5:00PM, 2 Dec.

Key: Co=Constantine, Do=Dokhturov, Ki=Kienmayer, Ko=Kollowrath, Lg=Langeron, Li=Liechtenstein, Mi=Miloradovich, Pr=Prebyshevsky, Be=Bernadotte, Bs=Bessières, Da=Davout, La=Lannes, Le=Legrand, Mu=Murat, SH=St. Hilaire, Va=Vandamme.

I call this a single-action pattern, with the first part of the battle consisting of a rise in the number of units engaged, coming to a peak where almost all the major formations of both armies are engaged, and then a falling-off as the losing side begins to break off the action and retreat. In a pattern like this, the battle as an event in time is quite compressed: Austerlitz was over 8 hours after it started.

Action at Gettysburg

Now let's take a look at Gettysburg using the same chronological perspective:

7:00AM, 1 JulyHe,Bu
8:00AM, 1 JulyHe,Bu
9:00AM, 1 JulyHe,Bu
10:00AM, 1 JulyHe,Bu,Re
11:00AM, 1 JulyHe,Re
12:00PM, 1 July
1:00PM, 1 JulyRo,Bu
2:00PM, 1 JulyRo,Re,Hw
3:00PM, 1 JulyHe,Ro,Ea,Re,Hw
4:00PM, 1 JulyPe,Ro,Ea,Re,Hw
5:00PM, 1 JulyPe,Bu
6:00PM, 1 July
7:00PM, 1 July
8:00PM, 1 July
9:00PM, 1 July
Night, 1-2 July
5:00AM, 2 July
6:00AM, 2 July
7:00AM, 2 July
8:00AM, 2 July
9:00AM, 2 July
10:00AM, 2 July
11:00AM, 2 July
12:00PM, 2 July
1:00PM, 2 July
2:00PM, 2 July
3:00PM, 2 July
4:00PM, 2 JulyMc,Hd,,Si
5:00PM, 2 JulyMc,Hd,Si,Sy
6:00PM, 2 JulyMc,Hd,An,Si,Sy,Ha
7:00PM, 2 JulyMc,Hd,An,Si,Sy,Ha
8:00PM, 2 JulyMc,Hd,Ea,Jo,Ro,Ha,Se,Sl,Hw,Re
9:00PM, 2 JulyEa,Jo,Ro,Ha,Sl,Hw,Re
Night, 2-3 July
5:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
6:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
7:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
8:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
9:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
10:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
11:00AM, 3 July
12:00PM, 3 July
1:00PM, 3 JulyHa
2:00PM, 3 JulyPi,He,Pe,Ha
3:00PM, 3 JulyPi,He,Pe,Ha
4:00PM, 3 JulyHd,Ki
5:00PM, 3 July
6:00PM, 3 July
7:00PM, 3 July
8:00PM, 3 July
9:00PM, 3 July
Night

Key: An=Anderson, Ea=Early, Hd=Hood, He=Heth, Jo=Johnson, Mc=McLaws, Pe=Pender, Pi=Pickett, Ro=RodesBu=Buford, Ha=Hancock, Hw=Howard, Ki=Kilpatrick, Re=Reynolds, Se=Sedgwick, Si=Sickles, Sl=Slocum, Sy=Sykes

If we look at Gettysburg in this way, we don’t see a single action like we did at Austerlitz. What we see instead are five different actions, each with its own disctint combinations of troop commitment, with quiet periods in between.

Gettysburg as a Game

If we think of Gettysburg as a game with hourly turns, it is fairly daunting. At one hour per turn, it would take 48 turns to play the 3 days of the historical battle. That would make quite a long game. To estimate how long, take a look at Napoleon’s Triumph, a 9-turn game on Austerlitz featuring armies about the same size as those engaged at Gettysburg. That game takes about 3 hours to play. At the same pace and scale, we would estimate that a Gettysburg game would take 15 hours.

Now, there is no hard and fast rule about how long a game should take to play, but in general, the longer a game takes to play, the less it can be played, particularly once it crosses the magic threshold of a single session, at which time all sorts of issues crop up regarding scheduling the players to get back together, how to leave the game up in the meantime, where to leave the game up, and so on. Taking all of this into account, a 15-hour game could not be played often, no matter how fun it was, just because of the logistical issues. Since I prefer to design games that can be played and re-played, this just isn’t the sort of game I want to make. So, clearly, if I was to make a game from Gettysburg, playing time would represent a substantial problem that the design would have to address.

Faster, Faster

There is one highly effective way to make a game go faster, and that is to cut the piece count. And so, I decided that my Gettysburg game would be at half the unit scale of my Austerlitz game: that is, each piece would represent twice as many men, and so it would only have about half as many pieces in play. From this change alone I expected to cut the playing time down from an estimated 15 hours to about 8 hours.

An 8-hour game is still a long game, but a lot shorter than 15 hours. However, to get within a single session window, I needed to shorten it still more. I didn’t want to cut the piece count further, both because such a major scale change really wouldn‘t support the game system I had in mind, and furthermore, wargames with too few pieces just don’t feel satisfying to people: more like a snack than a meal.

To make further cuts, I would try a radically different tack: variable-length turns. The idea was that periods of action would be played in multiple turns of an hour each, but that a quiet period would be played not in multiple one-hour turns, but in a single multi-hour turn. Thus, if we take a look at our Gettysburg time line again, we would see something like this:

7:00AM, 1 JulyHe,Bu
8:00AM, 1 JulyHe,Bu
9:00AM, 1 JulyHe,Bu
10:00AM, 1 JulyHe,Bu,Re
11:00AM, 1 JulyHe,Re
12:00PM, 1 July
1:00PM, 1 JulyRo,,Bu
2:00PM, 1 JulyRo,Re,Hw
3:00PM, 1 JulyHe,Ro,Ea,Re,Hw
4:00PM, 1 JulyPe,Ro,Ea,Re,Hw
5:00PM, 1 JulyPe,Bu
6:00PM-9:00PM, 1 July
Night, 1–2 July
5:00AM-3:00PM, 2 July
4:00PM, 2 JulyMc,Hd,Si
5:00PM, 2 JulyMc,Hd,Si,Sy
6:00PM, 2 JulyMc,Hd,An,Si,Sy,Ha
7:00PM, 2 JulyMc,Hd,An,Si,Sy,Ha
8:00PM, 2 JulyMc,Hd,Ea,Jo,Ro,Ha,Se,Sl,Hw,Re
9:00PM, 2 JulyEa,Jo,Ro,Ha,Sl,Hw,Re
Night, 2–3 July
5:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
6:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
7:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
8:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
9:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
10:00AM, 3 JulyJo,Ea,Ro,Re,Sl,Se
11:00AM-12:00PM, 3 July
1:00PM, 3 JulyHa
2:00PM, 3 JulyPi,He,Pe,Ha
3:00PM, 3 JulyPi,He,Pe,Ha
4:00PM, 3 JulyHd,Ki
5:00PM-9:00PM, 3 July
Night

Key: An=Anderson, Ea=Early, Hd=Hood, He=Heth, Jo=Johnson, Mc=McLaws, Pe=Pender, Pi=Pickett, Ro=RodesBu=Buford, Ha=Hancock, Hw=Howard, Ki=Kilpatrick, Re=Reynolds, Se=Sedgwick, Si=Sickles, Sl=Slocum, Sy=Sykes

If it worked, this would take the game down from 48 turns to about 33 turns, and therefore take the playing time down from about 8 hours to about 5 hours. Now 5 hours is still longer than the 3 hours of Austerlitz or the 2 hours of Bonaparte at Marengo, but I also thought that I might be able to do a little better in that once the armies locked up, they wouldn’t be doing as much moving and that would speed things up; I had hopes that I could get playing time to the neighborhood of 4 hours. All of this, however presupposed that I could put together a working system for variable-length turns, and the fact was that I went into the project without any real ideas how to go about it.

Pinning the Bell on the Cat

It is all very well to say that you’re going to have variable-length turns, but the first problem you face when you make this decisions is that you need a mechanism for setting the turn length. How would it be decided whether a given turn was one hour, or six, or anything in between? My first thought was to use a sort of an auction system: at the start of each turn, players would more or less bid the turn length, with the low bid winning. In this concept, if one player bid a turn length of one hour, and the other bid a turn length of three hours, the turn length would be one hour.

I can’t say I was ever enthusiastic about this idea. It sounded slow to have an auction before each and every turn, and I dumped it before I even finished the first rules draft. Eventually, I came up with the idea that one player or the other would have the initiative, and that player would get to choose the turn duration, within some sort of minimum-maximum constraints.

The first problem with this approach is that it left open the question of how to determine which player had the initiative. The essential solution here was to make each player put his army under what eventually came to be known as “general orders” — either “Attack”, “Hold” or “Withdraw”. Originally, the player with the initiative for a turn was by default the same as the previous turn, unless one player choose more aggressive general orders (Attacking being more aggressive than Hold, and Hold being more aggressive than Withdraw) than his opponent, in which case he had the initiative. This, however, proved to cause initiative changes to occur too frequently, which caused problems with the play mechanics. And so, I decided to only have initiative change when one player chose Withdraw: so, in the game, initiative belonged to the player who least recently put his army under Withdrawal general orders. (At start, the Confederate player holds the initiative as befits both the strategic situation and the temperaments of the opposing commanders.)

Action and Reaction

One sort of obvious question that I kind of skipped over was why I didn’t just make the turns longer (say 2 hour turns instead of 1 hour)? The answer to this has to do with what turns represent. The main thing you bump into more and more often as you lengthen turns is reaction time. In an actual battle, one army carries out an action, the other army sees it and reacts. When you lengthen game turns, you effectively slow down reaction times. One effect of slowed reaction times, for example, is that one army can surround a unit of the enemy army where historically the enemy unit would have withdrawn or had its flanks protected before it could be surrounded.

Modeling reaction time in wargames can be very complex, in part because in general lower-level commanders can generally react (in a local sense) faster than higher-level commanders. This differential in reaction times is one reason why armies are hierarchically organized — the lower-level commanders can react faster to what they see in front of them than if they had to send reports up the command chain and wait for orders to come back down.

Wargame designs often deal with command-level reaction time differentials by including short-circuits of their turn-length in some situations; examples of these would be defensive-fire and retreat-before-combat mechanisms, which allow the non-active player to take action against the active player during the active player’s turn. However, the longer your turns, the higher the command-level that you need to be able to short-circuit; if your turn length is more than the reaction time of a brigade, you need brigade-level short-circuits; if it more than the reaction time of a division, you need division-level short-circuits, and so on. Because short-circuits add complexity to a game, they are inherently undesirable, but worse than that, they also slow play. Because of this, it is perfectly possible to lengthen turn length in order to speed play, only to end up losing your time savings in the form of long-playing turns because they are full of short-circuits.

What You Don’t Know Can Hurt You

While reaction time in part has to do with issues relating to the command chain and the greater difficulty of co-ordinating larger forces than smaller ones, there is another subtle reaction time issue which is tied to limited intelligence. A commander can‘t react to something he is unaware of. If we look at Gettysburg in the second day, for example, Lee sent Longstreet on a wide circuit around the Union left, a process that took hours but which left the Union surprised and unprepared when the Confederates suddenly appeared to the south of the southern end of the Union line. Now, certainly the Union command‘s reaction time was more than adequate to respond to this if they had known it was happening, but they did not know, and so couldn’t react.

The more you look at military history, the more surprises like this you find, but they are hard to treat in boardgames because the physical pieces are not easily hidden and if both sides have hidden pieces, then detecting collisions between them as they move can be mechanically quite difficult. (Naval games were the first to really grapple with this; the old games Midway and Jutland coming to mind as being among the first if not the first.) One of the big advantages of computer games is that they can handle this sort of thing quite easily and naturally, but it is not something that boardgames do easily and almost never do well. Before talking about this with regard to The Guns of Gettysburg, however, I’m going to go on a bit of a philosophical digression.

Emergent Properties

In philosophy, an emergent property is a property that is not a property of any component of a system, but which is a property of the system as a whole. It can arise either because of the way the components are structured or because of the way the properties of the components interact. Chemical isomers for example, are molecules that consist of the same number and the same type of atoms but which have different chemical properties. It is the structure that gives rise to the properties, not the components themselves.

In The Guns of Gettysburg, an interesting result of the decision to vary the turn length in order to bring down playing time was that it had the effect of varying the reaction times. This variability, far from being a problem in the game’s ability to model reaction times, turned out be an asset in that it could allow the game to quite neatly handle moves such as the Confederate move around the Union left on the second day. Rather than modeling the inability of the Union to react through mechanisms using limited intelligence, the game could do it through varying the turn length. A 5-hour Confederate turn with no Union reaction might be hard to model with 1-hour turns, but it fell quite naturally out of a system that permitted 5-hour turns.

There was, however an unexpected design requirement for this to work. We normally think of the side with the initiative as being the side to move first in a game. (In Chess, white holds the initiative at the start of the game because white moves first.) However, in order to model a lack of reaction time, the side with the initiative in the game needed to be the side that moved second. To take our Gettysburg example again, if the Confederates move first in the 5-hour turn, then they Union gets a 5-hour turn of their own to react. However, if the Union moves first, they can’t use the 5-hour turn to react; instead they have to wait until the next turn, which the Confederates can force to be a 1-hour turn, so that the Union only gets 1 hour to react to a Confederate move of 5 hours.

As an incidental point, the fact that the side with the initiative goes second has given me fits insofar as terminology. Originally, I referred to the side with the initiative as the “active” side, and the opposing side as the “reactive” side, but this meant that the reactive side went first in the sequence of play and the active side went second. There was something of a playtester rebellion at this, as the playtesters let me know (rather strongly) that it made no sense for reaction to precede action. I did see their point, and sought alternative terminology without finding any terms that were liked much better. In the end, I went with “first player” and “second player”, which while wholly lacking any association with the concept of initiative, at least didn’t have any bad associations (unlike, say, “offensive player” and “defensive player” which tangled with the parallel terms used in resolving attacks).

Constraints and Limits

The ability to use variable turn lengths to model variable reaction times was immensely appealing, but it was not hard to think of ways in which the system could be abused. This required that the the variability in reaction times really needed to be constrained so as to maximize its strengths and minimize its weaknesses.

The central problem was to avoid having long turns used to make moves that could not have made historically because the enemy would have reacted early on in the move and prevented it from being completed. (Readers will recognize this as a problem of incorrect modeling of reaction time.) Really, the point of long turns was to model quiet periods and so all long moves needed to be “safe” moves, which would not have been prevented by low-level enemy command reactions. I had some idea that this might involve long-distance line of sight testing, but such a solution was highly problematic for multiple reasons. First, my map design did not really support such calculations. Second, it also was inherently complicated because first it involved testing each moving unit along its entire movement path against a potentially large number of enemy units. Such a system might be realistic, but it seemed inevitably slow, complicated, and completely impracticable. I worried for some time about this, but decided instead to just use the fields of fire limitations instead, which my map and rules already supported, and which were much easier and faster to see; and so I decided that a safe move could not cross an enemy field of fire. The next issue was to inhibit far-flung movements where units were sprayed out like a shotgun blast. And so I decided to add the additional "safe move" constraint that required units to end their move adjacent to or in the same position as another friendly block (including one that had moved earlier that same turn). This allowed a player to successively move units up to extend it along either flank, one unit at a time, which did give the ability to “curl” it around an enemy flank, which is more or less what the Confederates did on 2 July.

Another constraint was to tie the maximum turn length to the number of units that the smaller side has in play. The reason for this is a little subtle, but again goes to reaction time. Again, smaller units have faster reaction times (in general) than larger ones, so the maximum turn length for the early turns, when very few units were in play, needed to be different than the maximum for the later turns, when the entirely of both armies had arrived and were deployed for battle.

A final constraint arose from the need for the player who didn’t have the initiative to be able to attack without having his opponent frustrate him by choosing long turn durations. The solution here had to require that if either player placed his army under “Attack” general orders, the turn was forced to be just one hour long. In addition to being a necessary constraint, it also was in line with the rationale for the entire variable-turn-length system: to use short turns for periods when there was fighting, and long turns for when the armies were waiting or maneuvering.

Time and Tokens

The above presentation is lacking in one critical way, and that is that it doesn’t cover the way the artillery tokens and the turn duration systems interact with each other. Attacking requires the expenditure of tokens (reflecting not only ammunition expenditure, but command and control effects as well) which puts a brake on offensive activity; an army can only attack for so long before it needs to stop and recover, which in the game means refreshing its supply of artillery tokens. These tokens, however, are only drawn at a rate of 1 per hour, so periods of attack must alternate with periods of recovery, which does much to give the game the stop-and-go character of the historical battle. The result is that successive short turns tend to be turns where the players draw down their supply of tokens (when attacking particularly), which they then build back up again during long turns.

The Game in Action

In previous entries, I’ve shown various aspects of how the game plays, and to close this one I’m going to present a time-line of a playtest game, in the same form I used to present the historical battle; one that ended in a Confederate victory early in the morning of 3 July:

7:00AM–10:00AM, 1 July
10:00AM, 1 JulyJo,Bu
11:00AM, 1 JulyJo,Bu
11:00AM, 1 JulyJo,Bu
11:00AM, 1 JulyJo,Bu
12:00PM, 1 JulyJo,Bu
1:00PM, 1 JulyJo,Bu
2:00PM, 1 JulyJo,Bu
3:00PM, 1 JulyJo,Bu,Re
4:00PM–5:00PM, 1 July
6:00PM, 1 JulyJo,Pe,Bu,Si,Re
7:00PM, 1 JulyJo,Pe,An,He,Pe,Bu,Si,Re
8:00PM–9:00PM, 1 July
Night, 1–2 July
5:00AM–8:00AM, 2 July
9:00AM, 2 JulyEa,Ha
10:00AM, 2 JulyHe,Re,Se
11:00AM, 2 JulyHe,Re,Se
12:00PM, 2 JulyHe,Ea,Ho,An,Ro,Re,Ha,Se,Si
1:00PM, 2 JulyEa,Ho,An,Ro,Ha,Se,Si,Ha
2:00PM, 2 JulyEa,Ho,Mc,Ha,Si,Se
3:00PM–7:00PM, 2 July
8:00PM, 2 JulyMc,An,Pi,Pe,Si,Sl,Se,Ho
Night, 2–3 July
5:00AM, 3 JulyMc,Ro,An,Pi,Sl,Se,Sl,Si
6:00AM, 3 JulyConfederate Victory

Key: An=Anderson, Ea=Early, Hd=Hood, He=Heth, Jo=Johnson, Mc=McLaws, Pe=Pender, Pi=Pickett, Ro=RodesBu=Buford, Ha=Hancock, Hw=Howard, Ki=Kilpatrick, Re=Reynolds, Se=Sedgwick, Si=Sickles, Sl=Slocum, Sy=Sykes

The playtest game, like the historical battle, opened with a cavalry screen, but for a more prolonged period as reinforcement entry was slower. The second day saw an extended action in the middle of the day, with both sides attacking and defending, with another flare-up of fighting before nightfall. Finally, a morning attack by the Confederates in the morning on the third day into the Union center was successful, ending the game. The playtest game was in general much less violent than the historical battle, and both sides’ casualties were substantially lighter.

Another Cookie

In closing it occurs to me that once again I’ve produced a design diary entry with no illustrations, which I try not to do. So, to reward you for your patience, I’m going to give you the first several pages of the rule book as a downloadable PDF. (Yes, you’ll get the rest eventually; but you’ll have to be patient, I’m afraid.)

Gettysburg Rules