Once again, I’ve made a mistake. When looking at the battle generation capabilities of the Unity version of Field of Glory (FoG(U)), I assumed that the user-made scenarios created for the previous version of Field of Glory would only run in that older version. While strictly true, it didn’t take that much poking around to find that FoG(U) has a built-in tool that will upgrade a scenario from the old version to the new version with a rather obscure key combination. (Alt-Right + F5 if you don’t want to look it up, although you need to do so anyway to figure out the file structure).
Having made this discovery I’ve decided to test out my theory about improved AI in FoG(U) by playing the same scenario in both the old and new versions. For this exercise, I’ve chosen The Battle of Bannockburn.
For the Braveheart generation, we remember Bannockburn as the final scene of the movie. It is also the close of the book The Scottish Chiefs (although I despair ever getting through to that point in the book, as painfully slow and utterly ahistorical it is proving itself to be). Naturally, Braveheart had it all wrong. It depicts the battle as a spontaneous affair, taking place as Robert the Bruce finally gives up and agrees to swear allegiance to Edward II in exchange for support for his claim to the crown – a depiction that has no basis in reality. In fact, I have read that the Battle of Stirling, as shown in the movie (sans bridge) was more of an accurate representation of Bannockburn. While superficially the case, I think it is more an indication of the contortions through which a fan of Braveheart must go to connect the movie back to historical reality.
The initial setup of the Bannockburn battle in the original Field of Glory. The Scots have prepared an ambush from the woods while the English are moving by in column.
This is another battle that simply can’t be reproduced within the Field of Glory engine. The actual battle was fought over two days but the way the game is set out will necessarily fight it as a single day. As deployed, the battlefield mostly resembles the second day and the introductory text gives the date corresponding to the second day of battle. Of course, having just fought the Scots on the previous day, the English would not find themselves ambushed as depicted in this scenario.
Like the campaigns of Wallace, the specifics of the Battle of Bannockburn are preserved through a mix of written histories (some of which have only partially survived) and the heroic tales of Scotland’s independence. There is room for different interpretations of the orders of batttle, the tactics, and other details. I found a fairly extensive description of the battle that I am using as authoritative, mostly because it is sufficiently detailed. It also includes some excellent maps, made by one of the website’s authors.
In undertaking this exercise, I was hoping to obtain a clear victory in the old Field of Glory, one which I could attribute to the AI mistakes. That would then give me a way to compare the experience to the new Fog(U) AI. Unfortunately, that is not what happened.
I chose to play the Scottish side because, well, “Freedom!” and because I assumed, having won the actual battle, they might have an advantage in the scenario. Part of the problem is that (and you can see this in the above screenshot if you look at the minimap), the English are spread out backwards through rough terrain. This creates for the English AI a problem in that, to develop a decent attack, they must first consolidate and coordinate their forces in front of the enemy.
Against an aggressive English player, the Scots could play a defensive battle. I would think this would be a significant advantage, allowing the Scots to use both the terrain and their initial good order. Against the AI, however, the Scots need to go on the attack themselves. The first day of the real battle, the Scots aggressively attacked mounted knights with their schiltron formations, so as a representation of the whole battle, Scottish maneuvering makes sense. Day two, however, probably did follow more the Scottish defensive strategy that an English player versus Scottish AI might produce.
As the Scottish player, I was required to wheel my infantry out of the woods and form up a new battle line running parallel to the Bannock Burn. In doing so, the right side of my line got pretty badly mauled by the English knights. Towards the end of the game, when my left wing finally engaged with the rear of the English, I began to catch up in points. In the end, however, a battle based entirely on a Scottish attack is going to have a tough time resolving itself within the time allotted. My battle was a draw.
Sixteen turns was not enough to complete the battle, although it was getting close.
My next attempt to was to try playing the battle from the other side. As I discussed above, I anticipate the some of the difficulties in playing the English would be easier to figure out as a human player than for the AI. I also anticipate that if the battle was a draw for the AI English, it would be an absolute blowout for a human English player.
It took four turns to get my English into place for what looked a lot more like the historical battle than when I played as the Scots.
Sure enough, both of these were true. In my first four turns, I deployed the English knights facing the Scots in the woods, resulting in something that looks a lot more like the historical Battle of Bannockburn (more on this below) than the “ambush from the woods” that the scenario describes. I also realized that the English foot (see the green units at the lower left of the mini-map) take a long time to bring into play, because they are initially trapped by rough terrain. It was also a lopsided English victory.
My next attempt was to try to put together (quickly) something that I could play as the Scots, but was much closer to the historical battle. As I said above, this was a battle fought over two days. On the first day, the English began converging on the Scottish position and, assuming they had a decisive advantage, moved into battle piecemeal in an attempt to relieve a Scottish siege of Stirling Castle. Referring to the screenshot above, day one saw the Scots deployed in the woods facing to the left of the screenshot. They had prepared their position, including a series of hidden pits on the right side (in the screenshot, that is) of the Bannock Burn (the blue creek on the left side). As the armies engaged, the English were surprised by the effectiveness of the Scottish spear formations, which broke up the relief force.
At the point where the main English force was moving forward to, they believed, scatter the Scottish position, Sir Henry de Bohun road ahead of the army and challenged King Robert (the Bruce) to single combat. In something that reads more like fiction, the two fought and Robert smote Sir Henry with his axe. Cheered by their king’s victory, the Scottish spears surged forward and trapped the English, before they could form up, among the pits, rough terrain, and Bannock Burn ford. Unlikely as it seems, the Scottish foot scattered the English knights and caused extensive casualties. Both sides retired from the field for the day.
Again, a better written and almost certainly more accurate account of the battle can enjoyed here.
Realizing that the ford where they had originally crossed was well defended, the English used the night to begin crossing the Bannock Burn further downstream, away from the Scots’ defenses. While demoralized, the English still had both numbers and a professional army, and assumed that, once push came to shove, they would inevitably defeat the Scots.
Editing the scenario, I’ve deployed the English closer to where they may have begun the day, facing the Scottish line. I also shuffled the Scots around to match their Day Two deployment.
As the battle began on the second morning, the English knights were arrayed against the Scots, themselves formed up in the woods. To reproduce this, I’ve edited the scenario to move the English army forward and array them in a prepared battle line. There is an issue that is apparent in the above screenshot. One of the factors behind the Scottish victory was that the English had hemmed themselves in between the creeks to either side and the bad terrain to their rear. We can see these physical limits on the Field of Glory map, but by no means are the English “crowded” into that area. Reworking this aspect of the scenario is far beyond what I’d like to do here. My edits consist only of moving around the initial placement of the existing armies – no modifying terrain or order of battle.
While I was obvious wrong about the ability to convert scenarios to the new version, FoG(U), there was another feature I had wondered about. The editor in the new program does not appear selectable, and I never knew why. I finally looked it up. The editor function in FoG(U) was not completed. As a result, creating a user-made scenario for FoG(U) requires first creating it in the original Field of Glory, and then converting it to the new system using the process I’ve now just learned. I’m generally not a big scenario maker, but it is good to know how it all works.
The result with playing the newly-edited scenario was an improvement. As with the first time through, I played as the Scots against an English AI. Overall the battle felt much better than the original version. The English engaged immediately, and the nature of the battle was similar to the historical progress. As a result, the fight did not run out of turns as it did my first time through. Instead, I lost.
A second try through backed up my experiences with the original version. In this try, I took no initiative with the Scots, only engaging when I had the advantage. Working this way, I established an early lead and was moving steadily towards victory. Once again, however, I ran out of turns before a victory could be established. What seems to be the key to this result is that the AI English are not particularly aggressive but the slightly shorter scenario (15 turns) requires that engagement begins as soon as possible. On the basis of this observation, I continued on my journey to look for difference in the AI for the two versions.
By turn 2, the English knights had charged the Scottish lines. This took four or five turns in the old Field of Glory.
By the time I got this far, I was pretty sure what I was going to see and I was neither surprised nor disappointed. In the new AI, the English charged the Scottish lines immediately. When playing the Scots defensively, the results were pretty similar to those we saw in the old version of the program, they just happened a lot faster. The rear lines of foot did move forward more than in any of the previous games, but that did not seem to make much of a difference in the outcome.
Final results are within the range of historical possibility.
Also of interest, the finally tally of casualties not only had a margin of several turns allowing a finish in the allotted time, but the figures are within a range that matches the historical outcome. While these casualties are perhaps on the low end for estimated losses in that battle, remember several factors. First, this is the second day of a two day battle, so (particularly the English) start out already in the red. Second, the Field of Glory battles end when one side “breaks.” This means there is little in the way of pursuit of fleeing forces, when much of the casualties of a medieval battle took place. Historically, the English were killed and captured as they attempted to retreat back across the Bannock Burn and were trapped by the Scottish infantry.
Finally, in this version, unlike any of the other attempts at this scenario, the Small Folk actually played a roll. One part of the tale of this battle is how Bruce’s camp followers, non-combatants traveling with his army, rushed forward as the English became trapped in the bad terrain while trying to recross the creek. Waving sheets and brandishing knives, the fell upon the panicked English army. At the end of this scenario, I brought the Small Folk (represented in the scenario) forward and they managed to route one battered English unit that had been reduced through combat with Scotland’s front line.
Apparently, there is some dispute about the exact location of the battle. There is a Visitor Centre, not too far from the location that we’ve mapped out in this scenario. A main feature of that attraction is a “3D Game” which allows visitors to fight the battle, and then receive a debriefing on how the battle actually played out. It appears to be quite popular, requiring tickets in advance and having already sold out for the academic year for school tours.