Aug 26, 2012

Tests of Skill and Tests of Chance

I propose that challenges where resolution is a matter of mere luck should feel different than challenges where resolution is a matter of characters' skills interacting.

By "feel" I mean that the actual steps players and referees go through to resolve the situation should differ. 

Specifically, for challenges where luck is the decisive factor in what the consequences are, player-character agency and control should be decreased, and either referee control or the random factor should increase. Whenever the PCs are not taking action or making choices that affect an outcome, resolution of that outcome must be quick, but I would propose that either a slight delay, or an initial concealment of the results followed by the revelation to the players helps build suspense and excitement. For example, when rolling saving throws in a D&D-like game for PCs or small groups of important NPCs, you ought to try to resolving them one at a time, starting from the person most likely to pass the saving throw and progressing in order of descending probability. I started doing this a year or so ago, and have stuck with it ever since for its suspense-building effect at the table.

Another option is for the DM to roll all the saving throws at once secretly and then announce the various characters' fates in order of severity of consequences, which has a similar effect of mounting tension. One good thing about this method is that players will quickly grasp this patterns, which exacerbates the tension even further - knowing that the further along it takes before their fate is announced, the worse it will be. Even better, once this pattern is known, you can occasionally invert it to preserve the element of surprise.

Tests of luck are about the panic of suddenly having your control of a situation reduced or taken away. Used too much, they discourage planning and engagement with the fictional reality, since doing so doesn't affect one's chances of failure or removal from play. This is one reason I encourage you not to use saving throws for resolving too many effects in D&D like games. It's also why I think that any means of resolving skill tests ought to allow PCs to seek out bonuses and avoid penalties.

Tests of skill need to emphasise at least character, and preferably player-character, competence and agency. They should involve meaningful choices not only prior to resolution (viz picking the relevant skill or action to attempt), but during as much of the process as possible. These choices could be things like position during an attack (for a miniatures-based combat system like d20 or 4e), types of attacks or effects applied to opponents (as in Mongoose Runequest 2 and Runequest 6's combat maneuvers), resources to be expended in accomplishing a goal (WFRP 2e's Fortune Points), applicable bonuses or penalties (Openquest and the 40K adventure games), etc. They may be formal and explicit within the rules, or they may be informal "bonus grubbing".

If they are informal, I recommend the referee spend some time prior to the test explaining what kinds of things they see as causing bonuses or penalties to be assigned to the roll. Openquest does a good job of this, by laying out three basic conditions that create bonuses or penalties (Planning and preparation or the lack thereof; strong roleplaying and engagement with concrete details; the intrinsic ease or difficulty of a task), while in my experience WFRP and the 40K games are awful for this, providing much clearer guidelines for penalties than bonuses, and thereby encouraging penalties to be assessed more frequently than bonuses except in the handful of specific cases where there are clear guidelines (mainly combat).

Tests of skill can be extended tests, where success accumulates bit by bit over time, often involving teamwork. Despite this, I often see means of resolving these kinds of tests that basically allow only one person to make a meaningful choice (i.e. WFRP, where all other PCs do is add a straight 10% bonus to the test if they have the relevant skill), or that frontload all meaningful decisions at the very start and then simply become tests measuring the duration of the work (almost all crafting rules I've ever seen do this).

As a contrast, I would point to the resolution system of a game like Diaspora, where other PCs tag tests with aspects to provide bonuses or penalties, but there are a set categories of types of aspects allowed (zone, personal, etc.) with a limited number of "slots" in each category (one, in most cases). To tag a test, they must also expend a valuable personal resource (fate points). On top of this, they also have the option of making a test themselves (one relevant in the imaginary world) and transferring their bonus success levels to the other test. An "extended" test is one where one is trying to build up a certain number of success levels, which can be spent, IIRC, to reduce the time taken, to improve the final result, etc. This is a structure rich with relevant, meaningful choice throughout it that emphasises player-character agency and skill, and many other games would benefit from adopting a set of principles about how assistance works that are like it, even if they didn't want to formalise it in the same way.

The results of tests of skill should also be relatively enduring, while tests of luck should be fleeting. While I don't uncritically accept the principle of "Let it Ride", where a skill is tested once per scene or situation and that result applied to all possible uses a character makes of that skill, I do think it's preferable to the "Make a Perception test. Now make another one. Now another. Again." school of thought. My main objection to the Let it Ride concept is its stasis across a scene. I think that the results of tests of skill should endure until circumstances change meaningfully. My criteria for meaningful change are not meant to be exhaustive or definitive, but include "a different bonus or penalty would apply than when the previous roll was made", "another character's actions have enabled or interfered with your ability to execute the skill", or "You need to expend a limited resource each time to make this roll"

I'm willing to allow that in combat and certain other rapidly developing situations with lots of PCs and NPCs making decisions that interact with one another in complex ways may make the timeframe for meaningful change into a few seconds long, while in other situations, it may be hours or days.

The important thing is to avoid multiple tests for closely related goals where the circumstances are stable. If you are looking for Bill and Joe in the crowd, then one test is sufficient, rather than two (if this is an opposed test, because Bill and Joe are both hiding, then they should each test against the same roll, rather than having two separate roll-offs between the PC and Bill or Joe).

As part of that endurance, tests of skill should also not be fragmented across multiple PCs unless absolutely necessary. If the entire party is looking for Bill and Joe in the crowd, then four to six Perception rolls (plus whatever Bill and Joe are rolling in opposition) is tedious, boring, and ensures that success is simply a matter of rolling and comparing dice results for ten minutes rather than making meaningful choices. It boggles my mind sometimes that RPGs take the activity of an isolated individual as the paradigm of action resolution, when the actual situation is almost always a group of PCs assisting one another. A simple kludge to cover this situation is to allow PCs to provide bonuses to one another for teamwork, though this only slightly ameliorates the fundamental problem.

Skill challenges in 4e were problematic for many reasons, but there is an interesting idea that I would preserve from them. The idea is that during a skill challenge (where each of the PCs goes in order rolling a skill trying to accumulate successes), the same skill cannot be used by two PCs who are adjacent to one another in turn order. While in a skill challenge this is an artificial constraint that serves to highlight the gameyness of the whole process, I think combining it with the simple bonus kludge would push PCs to take more agency.

The basic idea for something like trying to find Bill and Joe would be that the PCs are all going to work together. At any time, one of them can make a Perception check to look around, but only one PC can do this (perhaps the check happens each round and each time the identity of the PC making it can change). The other PCs will be assumed to be assisting when that happens, and will therefore add a small bonus. If they want to contribute a larger bonus, they must use different skills, powers or tactics to create bonus-worthy conditions for that single check. Similarly, if the Perception check does fail, it's up to the other PCs to use their skills to create a meaningful change in circumstances to allow another check to be made.

I don't claim to have the complete solution here, nor to have exhausted all possible options, but I encourage you to test these out and to develop new solutions based on the underlying principles laid out here.


  1. This is a really thoughtful post. I don't have anything to add really, but this has got me thinking and I wanted to thank you for that. :)