This technical article is, first and foremost, targeting board gamers with a deep knowledge of the THS, the Conan game system. Even for this audience, this article would likely interest only those who wish to understand the creation process of a game system. For every other readership this article will only be of limited interest.
From adventure mode to versus mode
For a long while, I was thinking that the Versus mode would only require two Skelos books put up against each other. It would only be a formality. When we started the first tests I realised the way to go for this mode would be to offer a game of interest at least equivalent to the Adventure mode. The exercise was all the more arduous because the two modes had to share the same Systemic Consistency. It was not about adding a system to another but rather rethinking dialectically the two systems in order to feed off each other. Each system must be as alike as an outgrowth of the other, in such way it would be impossible to distinguish the root from the bud.
The need to desynchronize the game rhythm: the emergence of Activation Potential
One of the interesting aspects of the Adventure mode as operated in Conan is based on the asymmetric activation rhythm between the two sides. One side (the Overlord) turns out to be particularly regular in their number of activations (cf. Boxed text) while the other side (the Heroes) seems to be more arrhythmic. Actually, the number of activated Heroes during their turn could vary greatly (from 0 to 4 activations, with all the possible spectrum empirically covered). Consequently, this ensures some desynchronization of the action between the two sides.
Despite the intrinsic symmetric system of the Versus Mode, it was a major concern to keep this arrhythmia between the opponents. By introducing a new cycle called Activation Potential, the asymmetry of the number of activations could be ensured. Strongly embedded (as its Systemic Consistency is high since its analogy with THS Energy Cycle) this new cycle has, however, a specificity that distinguishes it from the former and allows it to fulfil its purpose: the marginal growth of its utilisation cost joined to a Recovery Constant (with the second never being ideally equivalent to the first for the balance between spending and recovery could be only realised over a term longer than a turn).
Hence, whether to activate a second tile (or even a third or a fourth) results from a real arbitration. Indeed, the savings generated from non-activated tiles is substantial enough to impact future game turns, and thus the rhythm of the game. And it is precisely the notion of game rhythm which was in balance here.
The introduction of the Activation Potential broke the synchronization between two Overlords. They are now free to over-consume gems when needed (e.g. for a charge, to make a special effort). They also could anticipate a future burst of activity on the board by hoarding gems.
In the Versus Mode, an average game lasts around 45-60 minutes for approximately 12 turns for each player. By calibrating the Recovery Constant to 2 gems in association with an Energy Pool of 3 immediately available gems and 7 gems in expectation, we empirically tend to 1.5 activations per turn (with 1 gem cost for the first activation, 2 gems for the second, 3 for the third and 4 for the fourth, being respectively 1, 3, 6 and 10 gems spent for 1, 2, 3 or 4 activations during the same turn). Consequently, the number of activations for each game is relatively equivalent between the two players (about 18 activations for each player in one game). But the distribution of the activity peak is strongly desynchronised.
On the following example: the two players have activated the same amount of tiles during the whole game (Σ of activations = 18). We can easily observe a great difference between both players regarding their respective game dynamic.
(*) The Activation Potential could be described as an additional circulation line on the book of Skelos. 10 gems of a new color move along a system of communicating vessels similar to the Energy circulation. Each turn, a player recovers 2 of these gems. For an umpteenth tile activation, the player has to spend on this cycle, an equivalent number of gems. Then like on the Adventure Mode, they have to spend gems on the energy cycle, according to the tile’s position on the river. (cf. diagram at the end of the article).
The struggle against excessive savings
Previously, we mentioned the problem of the excessive savings into the Overlord gem management. Saving is not an issue when only one of the two players is concerned. However, it becomes more problematic in a mode that only confronts Skelos books.
Three tools have been implemented in the game to encourage players to look further down the river for the units they need:
1/ Activation Potential anew
In addition to the arrhythmia it encourages, this new cycle tends to increase the average expense of a tile’s activation. The reason for this is very simple: because the second activation is no longer systematic, it becomes useless to consider activating the first tiles only, in order to save gems during one turn.
In other words, going from an average of 2 tiles activated each turn to 1,5 tiles activated, without decreasing the Energy Recovery Constant (5) nor the Energy Pool circulating onto the cycle, mechanically generates an inflation of the average gem expenditure for each activation.
2/ The implementation of exertion limit link to Punctual Improvements
The inflation’s effect described below can only be fully operational if the remaining gems are not immediately used for an alternative allocation.
In other words: everything that competes with a given tile for gem allocation, leads to lower his activation probability, and all the more if its current cost is high. Whether this competition comes from other tiles (and the Activation Potential tends to reduce such competition) or it exists in the form of Punctual Improvements. So, to limit the allocation’s dilution, the capacity to spend gems on Punctual Improvement was drastically restrained. From the implementation of the exertion limit results an interesting corollary: the necessity of the arbitration. Since reaching the exertion limit for one character comes down to deprive another ally from a Punctual Improvement.
3/ The tiles’ over-specialization
This is the last, and probably the most powerful, lever used to incite gems’ prodigality. In the Conan game, the tiles are only distinct from each others in degree, or for their skills, but never in their nature. So, each time a player has to decide which tile to activate, they undertake an intuitive calculation to estimate the profitability of each gem spent. They may make a decision between the activation of only one powerful tile but with a high cost on the river, and the activation of another tile, less powerful, but less costly. The second option could save gems that can be spent later in Punctual Improvement to increase the power of the characters used. In other words, as long as the tiles differ only in degree, the river performs like a one-dimensional continuum on which each choice is too often limited to efficiency maximisation (Efficiency/Cost ratio). By strongly specialising the tiles (attributing some characteristics to some tiles while depriving other tiles from them) and if these characteristics are essential to a scenario, the one-dimension of the continuum is broken (in fact, each time a characteristic is not shared with other tiles, an additional dimension is superimposed on the initial continuum). Basically, in a situation where a solution requires a given characteristic, it is most probable that the activation of a tile with this characteristic will be expensive as this characteristic is rare and must be used frequently. In the Batman game, the implementation of such tile specialisation has gone through the episodic introduction of Thought and Manipulation.
The High Variance Dice appearance
During the Versus Mode’s first tests, we quickly hit a snag: the major difficulty for the Attacks to overcome the fixed Passive Defense. We already thought about this problem during the development of the Conan Adventure Mode. To avoid it, the Heroes Armour is expressed as dice (So with a result that may vary) whereas for the Overlord units it is expressed as a Constant. The fixed Defense is not a problem for the Heroes as they can vary the intensity of their attacks in regard to the Defense threshold they have to overcome. But units cannot increase their attack, so the fixed Defense becomes crippling. The first solution considered was to express all the Defense with dice, like Hero’s Defence. These tests immediately reminded me why this option was rejected some years earlier: the whole game was weighed down because of continuous dice rolls, which impeded its fluidity substantially. It would also be possible to simply increase the damage done by each tile without increasing their Defense at the same time. In doing so, the fixed Defense would be overcome more frequently... except that the Adventure Mode would have been strongly impacted, resulting in the Heroes being slaughtered. In reaction, to re-equilibrate the game, the number of Overlord units should be lowered when opposed to the Heroes, what I refuse to do (it seems to me that one of the major interest of the Adventure Mode lies in the asymmetry, a handful of Heroes facing a horde of villains). It took some time before the solution appeared by itself in front of a cup of coffee: if it was impossible to change the expectations under pain of denaturing the game, on the other hand it was quite possible to work on the dispersion around these expectations...
(*) The impossibility for the Overlord to increase their attack power with Punctual Improvements, contributes to limit the arbitration between tiles as stated earlier in the text. If we authorise such an improvement, we mechanically reduce the interest to go further in the river to choose a tile at high cost rather than activating the first one and improving its power.
(**) The huge dispersion around the expectation (the polarisation of the ventilation of the power on the dice) simulates perfectly the characters and weapons with low precision but high power. This is the case of clumsy brutes on a hand to hand fight, they barely touch their opponents, but when they do, they are especially deadly. It could also represent the recoil of heavy firing arms (Colt 500 or shotgun for example) incapacitating for the precision.
It is astonishing to note that despite the astronomical number of games using the dice as a random resolution tool, the Variance (the expectation of the squares of the deviations of a variable from its mean) remains a dimension a priori strongly absent from the problems of Game design. However it is a powerful simulation tool (cf. above). For a reason that escapes me, most game authors seem to focus primarily on expected value (weighted arithmetic mean), thereby neglecting a lot of the information the dice are carrying. But the Variance is just as interesting to exploit as the expected value. Furthermore, in the Versus Mode, the Variance is overriding, as since increasing it without changing the expected value could lead to open the spectrum of results (increasing the dispersion measured below by the coefficient of variation). This allows one to overcome the fixed Passive Defense. Thus, we introduced 2 new dice, Black and White, corresponding respectively to a Red and Yellow dice with polarised distributions.
The five dice are presented in the table below:
Now let see how these dice combinations perform against fixed Passives Defenses.
The graph above and on the left are respectively typical attack profile for Goons and Leaders.
A simple glance at the graphs above shows that, regardless the attacker’s profile (Goons or Leader), the high dispersion configuration remains more advantageous when a threshold is overcome. In case of the Goons’ profile, whenever there is a Defense (1 or more), the attacker substantially benefits from an attack with high variance (a low variance attack is only an advantage while facing creatures without any defense, such as dogs). For Leaders’ profiles, as they are depicted here, after 2 Passive Defense points have been reached, the curves never meet again. It is worth noting the importance of such a gain: for a Goon configuration, the probability to overcome a 1 point Passive Defense is 33% higher for a high dispersion configuration than for a classic configuration. This probability becomes four times higher, when the target has 3 points of Passive Defense. In the same vein, a Leader configuration with a high dispersion turns out to be 55% more efficient to overcome a 3 point Passive Defense than a classic configuration (increasing the efficiency from 36% to 56%). For a 4 point Passive Defense, the 8-percentagepoints- gap between classic and high dispersion configuration represent a 57% advantage. Facing 5 point Passive Defense, the erratic configuration is four time more efficient. Beyond that point, overcoming the armour is nearly impossible for the classic configuration, while it is still probable (>10%) for the high dispersion configuration.
Those who are still reading these lines will understand that the work on the THS leaves little room for chance and improvisation. Adding a new game mode is never trivial, but rather requires a systematic global rebuilding of the system outside of any particular game mode. Adding a new game mode without rethinking the whole system in its entirety would put aside important notions like Systemic Consistency or Integration. The Systemic dimension is far from being anecdotal. It is, in fact, consubstantial to what a game is. The Systemic dimension is opposed to a bricolage consisting of stacking games mechanics, for a functional game but without unity (I.e. endowed to an endogenous logic with the principle of all the actions covered by the system). Without this founding principle of the Systemic Consistency, the Versus Mode would have only been a second game standing alongside the Adventure Mode. But it had to be more. The two modes have to be both sides of a same coin, like twins sharing the same DNA and evolving in different environments.
Article from Frédéric Henry (designer of Conan and Batman: Gotham City Chronicles)