I’m trying a new layout approach in this article. It sacrifices some screen real estate for indented subsections. Do readers like it? Let me know what you think of it!
A lot of the advice here at Campaign Mastery sometimes gives the impression that there’s a shortcut to solving every problem, because offering alternative perspectives on different problems often identifies a solution without the tedium of doing things “the hard way”. This article will address the elephant in the room that has been ignored for far too long – that sometimes shortcuts simply don’t work, usually because there are a wealth of possibilities to choose from and you don’t yet know which one – if any – will be the “right” answer.
That’s when it becomes necessary to roll up your sleeves and actually work hard – hopefully for just a short period of time.
I’m fond of the jigsaw analogy to describe the process of compiling and assembling the right building blocks for a campaign, adventure, or encounter, because it is singularly apt. In any given situation to which you need to devise a solution, there will be some things that are known and some that are unknown. As anyone who’s ever solved a jigsaw puzzle knows, sometimes the edges of a couple of known pieces leave only one piece that can possibly “fit”; most of the creative shortcuts that have been offered come down to either “sorting” the available pieces or knowing which “edges” to look at in trying to identify a match.
As with an algebraic expression, as soon as a second variable enters the picture, or there is a second adjacent empty piece so that there is more than one possible edge to “match” with, everything becomes a lot more complicated. It’s the difference between a disk and a sphere, or between a topographic map and a rather more abstract road map – an additional dimension has been added, vastly increasing the complexity of the problem.
I’ve been in this game (pun intentional) for long enough to know that sometimes, the only answer is to apply a brute force approach to the problem, and to have devised ways of taking as much hard work out of that process as possible.
One of the reasons for the long delay in addressing this aspect of the design process has been the difficulty of explaining those techniques; I’ve started this article half-a-dozen times or more and scrapped it every time. Finally, though, I think I’ve devised a method of doing so. The key is in using a simpler example to demonstrate the process, and educating the reader in a couple of stages, rather than trying to give – and explain – the full technique in one hit.
Basic Tools: The one-variable problem
So let’s assume that you are working on an encounter for your upcoming game session, because an encounter is the smallest possible element of an RPG.
You know some things about the encounter, but not everything. There is one “piece” of the proposed “jigsaw” that is not immediately obvious, because your box contains the pieces from four or five or twenty different “puzzles”.
You may know the climate, the terrain, the creature, the relevance to the overall adventure, the relationship of the adventure to the campaign as a whole, the tone that you are striving for, some of the backstory of the encounter, and even the difficulty of the encounter and how it is to relate to other encounters within the adventure. The missing piece is the actual location.
Example: Hot, dry; grasslands and scrub along a river; a gnoll who has had his abilities supernaturally augmented by a ring; the ring is one of a set, and the arch-villain of the adventure has another; the tone is to be epically spooky; the Gnoll has used the powers granted by the ring to forge/force an alliance of various fallen races, and become dominant in the local region, but something about the location has prevented him from further expanding his sphere of authority; and the difficulty is to be quite high, relative to the PCs, some of which can come from the Gnoll’s augmented capabilities, and some of it from those flunkies that he has gathered around him, but some of it has to come as tactical advantages from the location; and, finally, the PCs need to capture and analyze the ring in order to learn how to defeat that arch-villain, or at least cut him down in size to the point where they have a chance at doing so.
Because you know almost everything about the encounter, the “edges” of the missing “piece” are almost trivial to identify, and I’ve no doubt that any experienced or half-talented GM could immediately think of something to fill the empty hole in the picture, and run this encounter quite successfully. In fact, you can probably infer most of the adventure that surrounds it from the information provided, it’s that complete. That’s not important; what matters is that the simplicity of the situation makes it easy to illustrate the basic tools that will then be applied to more difficult problems.
The basic process is:
- Step One: Identify the parameters
- Step Two: List the possible solutions
- Step Three: Cull the solutions to something manageable
- Step Four: Choose the solution
Step One: Identify the parameters
In this case, we need something that would provide a power base for the Enhanced Gnoll, and that would appeal to such a creature; it has to be something that confers a tactical advantage to him and his forces, but that also explains why his conquests stopped after this victory; and it has to convey that tone of “epically spooky” since nothing else on our list of “known ingredients” is really doing so. These requirements are the parameters that our solution to the problem must satisfy.
Step Two: List the possible solutions
There are a number of possible solutions. A grass hut; a village; a fortified village; a town; a keep; a castle; a ruin; a city; a subterranean enclave within a city; a system of natural caverns; a lost temple; a tower; or some combination of the above. All of these – with the exception of the first – are big enough to contain a graveyard, creating the foundation for the “spooky” tone. They all confer a tactical advantage, if used properly.
Step Three: Cull the solutions to something manageable
How many of them can be called “epic”, though? How spooky are they, aside from the vicinity of the assumed graveyard? And do any of them explain the lack of progress after the conspicuous success? The larger the location in terms of pre-Gnoll population and fortifications, the more it fits the “epic” label, but the harder it becomes to then resolve the second question.
At first glance, the “epic” criterion lets us cross off the grass hut, the caverns, a village (fortified or not), and even a more substantial town lacks the prerequisite aura of significance. Towers are more associated with mages than with “spooky” and its countering. These can all be dismissed without a second thought.
There is also a temptation to add “a keep” to that list of exclusions, but a ruined city with a Keep that has been captured largely intact remains an option. The lost temple, on its own, similarly falls short of the mark, but a ruined city which contains an intact temple is a viable choice.
Another option listed that might be dismissed at first glance as insufficiently spooky is the subterranean enclave within a city, but if there are supernatural manifestations of some sort that have driven the regular occupants out (or killed them), it can remain a contender – that probably means losing the “subterranean” part of the description, though.
None of them, as described, are enough to explain the lack of progress since this victory. We would need to add something more to the location and its backstory to do that, probably in the form of an opposing force which can more or less sustain the situation in equilibrium until the PCs arrive to tip the balance. Mulling that point over leaves only two viable options.
Step Four: Choose the solution
Solution number 1 is a still-viable and occupied city. The Gnoll forced his way through the city walls in a violent and bloody confrontation with the city’s defenders and occupied a temple that they defiled and corrupted, from which a supernatural menace exudes that has driven the residents out. Almost a quarter of the city has been consumed by this malevolent influence, a region in which it is always night and from which foul weather regularly erupts; the citizens have done their best to fight back, but have succeeded in doing nothing more than contain the evil. To defeat the evil, the PCs will have to penetrate and cleanse the Temple and then confront the source of the evil, the Gnoll.
Solution number two is a city that was devastated by the conquest of the Enhanced Gnoll, with a keep that has largely survived intact and now forms the base of operations of the Gnoll. In a desperate attempt to repel the invaders, a temple within the city resorted to forbidden rituals which have raised the dead populace; most simply attack any who venture within the walls of the city, or who attempt to leave, but some few priests and temple guardsmen, transformed into supernatural enemies of all who live, have directed their enmity against the Gnoll and his besieged forces; two evils, each holding the other in check. To succeed in the overall objective (in plot terms), the PCs will have to penetrate the city, avoiding or fighting off its spectral defenders, and cleanse the temple, ending the siege of the Keep and luring the Gnoll out for the ultimate confrontation of the encounter.
The first one sounds epic, the second one more so; both explain the sudden cessation of progress by the Gnoll, though the second one does so perhaps more credibly and certainly with an ironic element that the first lacks; and the first one has supernatural spookiness tacked on, while the second has it as a fundamental part of the concept. And its those last two factors that make the difference between the two, in my book: I would choose the second.
More to the point, this illustrates the core three stages of the process: parameters, solutions, cull solutions using parameters.
Something more difficult: The two-variable problem
With the basic process established and demonstrated by this relatively straightforward example, even though the brute force approach was overkill given the tight definition of the parameters, we can now expand the principles to a more complex problem, and then use that as the process template for expanding the scope of the technique to as many unknowns as have to be filled in.
The two-variable process is considerably longer than the one, as you would expect. At the same time, though, the individual steps will look very familiar after the example provided above.
- Step One: Identify the parameters of variable X
- Step Two: List the possible solutions to X
- Step Three: Cull the ‘X’ solutions to something manageable
- Step Four: Identify the parameters of variable Y
- Step Five: List the possible solutions to Y
- Step Six: Cull the ‘Y’ solutions to something manageable
- Step Seven: Create a list of paired XY solutions
- Step Eight: Cull the list of paired XY solutions
- Step Nine: Choose the XY solution
Steps One to Three
These are quite obviously the same as steps one to three of the single-variable process. An example might be the preceding example without the keyword “spooky” – this was one of the key values used in culling solutions, and was instrumental in making the final choice. Ideally, you want to get your list down to two or three possibilities; four or five would be a more realistic ambition. Six, Seven, or Eight are not out of the question, because there are going to be many more contenders on the original list – “spooky” was used as a guide to restrict the number of choices the first time around. This time, you would have “clearing”, “underwater”, “unnatural mountain”, “young volcano”, “cloud caste”, and a whole heap of others – and that’s still with the “grasslands” terrain as the dominant local feature.
If all you know is that you want the tone to be “epic”, there are LOTS of alternatives that remain available. Narrowing it down requires applying the same process to the second missing (or, in this case, incomplete,) variable – tone.
Steps Four to Six
So that’s exactly what you do in steps four through six. “Spooky” won’t be the only choice that comes out of the process, but it – or some synonym – should be one of them. Some possible tones would be easily ruled out as inappropriate to an Enhanced Gnoll as antagonist – “heist” and “political” and “romantic”, for example. But others, like “Quest” or “Infiltration” or “War” would remain quite viable.
Step Seven: Create a list of paired XY solution combinations
For the sake of argument, let’s say that steps 1-3 give four possibilities, A, B, C, and D; and steps 4-6 give five possibilities, 1, 2, 3, 4, and 5. That means that there are four times five combinations, or twenty in total. This should make it clear why it’s so desirable to cull, HARD, in those first steps.
In fact, I generally go further: some combinations will be obviously incongruous or incompatible. “Epic Quest to the Small Village”? I don’t think so. I won’t bother wasting my time listing or assessing those, in the interests of keeping the results more practical.
Step Eight: Cull the list of paired XY solutions
This is done in exactly the same way as step three of the one-variable process. The difference is that you are now working on a paired solution and seeing how well that pair ‘meshes’ with the things that you know about the encounter already.
Step Nine: Choose the XY solution
Finally, having culled your way down to a handful or less of viable combinations, pick the one that works best or holds the most appeal, and that is most unlike the other adventures that you’ve run recently.
With an understanding of how to expand the process to deal with multiple variables, it remains only to look at the nuances of doing so.
Applying the principles to a larger problem
What if you had three unknowns? Or four? Or five?
The principles are the same. But there are some practical tricks that are important to note.
Culling of each variable is critical. You’re dealing with exponential growth, here. Let’s say you have three variables: if you can cull each down to three choices, that’s 27 possibilities, less any incompatible ones. If your culling leaves you with four choices, that’s 64 possibilities, less the incompatibles. If five possibilities, you have up to 125 to consider.
That’s a huge difference.
It only gets worse when you have four variables. Three Choices = 81 possibilities, four choices = 256 combinations, five choices = 625 potential solutions.
Additional Culling Stages
The more culling you can do in one hit, the less work overall. That means that it’s better to produce two sets of variable pairs, cull each, and then look at the four-way combinations of the culled lists. Call them the XY and AB pairs.
Proof of this claim is simple: let’s start with 4 variables and 4 choices for each. XY combinations: 16. AB combinations: 16. XYAB combinations = 256.
XY and A = 64. Every possibility from this pool that you eliminate also eliminates 4 combinations from the XYAB pool. If you can get rid of half of them, that’s 32×4 = 128 possibilities eliminated – but still leaves 128 to do.
Culling half of the XY possibilities and half of the AB possibilities before you consider the XYAB pool means that you are dealing with 8×8=64 possibilities. Each possibility from the XY pool that you cull removes 16 from the final XYAB pool. Ditto each AB combination that you can rule out.
Additional Culling Criteria
Longtime readers might remember my 2010 article, Scenario Sequencing: Structuring Campaign Flow. A key point of it was the way in which I identify additional criteria for use in adventure generation – Level Of Action, Level of Fantasy, Level of “Cosmic” content, intensity of Tone – rate each adventure according to these criteria and then sequence the adventure ideas within the campaign for variety. For a fantasy campaign, I might use different criteria: “social” content, “arcane” content/emphasis, “spiritual” content/emphasis, “big picture” relevance – but the basic approach would be the same.
In more recent times, I’ve taken to defining and using campaign Themes. If you want to explore that more thoroughly, the place to start is my 2014 article Touchstones Of Unification Pt 1 – Themes, and (to a lesser extent) the entire Touchstones Of Unification series. I also discuss themes at length in the New Beginnings series, especially New Beginnings: Phase 4: Development.
All of these represent additional criteria that can and should be brought to bear in culling the options, especially when you have trouble culling using the techniques described earlier. You see, each unknown reduces the available criteria for culling every other variable.
Think about that for a moment. If you have twenty possibilities and five known criteria, three or four is an entirely reasonable expectation, post-culling (divide possibilities by one more than the number of known criteria). If two of those five become unknowns, five is about the best you can hope for, and six is entirely possible. If three of them become unknown, six-or-seven is as good as it will get, and if only one of the five is known, the best you can hope for is ten possibilities.
Technically, the mathematical relationship is total possibilities, all variables = (P / [k+1]) to the power of (T-k), where T is the total number of criteria, k is the number of them that are known, and P is the size of the initial pool for each variable. The bigger k gets, the more the final number is cut down, and its that final number that ultimately determines how much work is involved.
If you know only that you want the PCs to have an encounter of some type, the field is wide open, and you have innumerable combinations of possibilities. The more you know beyond that absolute minimum, the more you shrink the field, and the more manageable the process becomes. Having additional culling criteria that can be applied to reduce P without increasing T can make a HUGE difference.
Altering the order in which you assess the variables can also make a huge difference. There are three basic approaches:
- Logical dependence,
- Variable Compatibility, and
- Variable Incompatibility.
Where one variable can be shown to be logically dependent on another in large degree, bundling them together as a pair early in the process greatly reduces the number of viable combinations that will survive the culling process.
That doesn’t happen very often, and frequently requires rather special circumstances to apply. Most of the time, you will be left with doors number 2 and 3.
Door number two, in this case, describes situations in which the two variables being combined and culled have something to do with each other, some logical fit in terms of what they are describing – location and terrain, for example. Neither defines the other, but both restrict the viable choices of the other.
Door number three is when the two variables being combined have no apparent relationship. Climate and anything plot-related, for example, are usually incompatible as a combination of variables.
While most variables will fall into one of the two categories, there will always be some that aren’t clear-cut. Many people might suppose that the relevance of an encounter to an adventure, and the relevance of the adventure to the campaign, are either strongly compatible or even dependent. In fact, they are completely independent – there’s nothing to prevent a combat encounter in a romantic plotline, or a humorous encounter in a cosmic quest. And yet, at the same time, it would be a very brave soul who defined them as being completely irrelevant to each other, so they can’t be incompatible, either.
Choosing A Variable Sequence
The objective of preferring one choice over another always has to be in culling more options. Whichever approach does that is infinitely superior to any alternative.
Always pair logically-dependent items together because they will produce a superior cull, by virtue of the dependence.
That’s fairly clear-cut – but from that point on, it becomes rather murkier. There are two lines of argument: The first is that because two compatible items relate to each other, they make a superior choice by culling incompatible options. The second is that because two incompatible items collectively describe more of the encounter, they make a superior choice by culling undesirable/incompatible combinations. Both arguments are valid.
In any given case, one will be more influential than the other. As a general rule of thumb, the more unknowns there are, the less there is to get a grip on in terms of combination viability, and the more dominant compatibility will be over incompatibility – there are simply too many unknowns to make the latter approach viable. But this is very much a “most of the time” thing, and not a hard-and-fast rule.
For that reason, when there is no clear cut choice – four variables, two of which are logically-dependent, means that you put those two together and the partner up the other two by default – I will often cheat by doing things out of order. Specifically, I will generate the pre-cull list of choices for each variable, cull for the known factors, and then do a quick comparison in my head – X with Y? X with A? X with B? Which combination of variables is the most effective?
I am guided in this assessment by the principle of the small variable.
The Principle Of The Small Variable
This is something that I noticed when I was studying basic trigonometry in high school, and found applicable to all sorts of unlikely purposes since – everything from bubble-sort efficiencies through to chained skill checks (where one skill check must be made to give the character a shot at making the meaningful skill check). The principle is simple: the closer to each other two numbers are, the larger the product of those two numbers:
1 x 5 = 5
2 x 4 = 8
3 x 3 = 9
4 x 2 = 8
5 x 1 = 5
The larger the “high number” is, the more obvious this pattern becomes.
1 x 10 = 10
2 x 9 = 18
3 x 8 = 24
4 x 7 = 28
5 x 6 = 30
6 x 5 = 30
7 x 4 = 28
8 x 3 = 24
9 x 2 = 18
10 x 1 = 10
The other fact to notice about this pattern is that it’s fairly diffuse. There’s not a lot of difference between the 4×7 result and the 5×6 result, for example. In this application, it means that a choice of two variables that dramatically reduces the population of one pool is better than a choice that reduces both pools by a smaller amount.
That’s what I look for when choosing variables to partner with each other: the one best combination for culling results in this particular case. And if you have a large pool of choices and a small choice, it’s better to partner them than to partner each with a medium-sized pool.
To illustrate that, let’s look at four-variable combinations: 3, 4, 5, and 7 options, respectively. All things being equal, the best approach might initially seem to be partnering the 3 and 7 pools (XB), leaving the other pair to partner (YA), but appearances can be deceptive: 3×7=21, 4×5=20, cull both to maybe 11 and 10 respectively, 110 combinations to consider in total – instead of the ‘raw combination count’ of 3x4x5x7=420. Compare that with an alternative configuration: 3×4=12 and 5×7=35, cull to 6 and 18, 6×18=108 combinations to consider. The third configuration is sort of in-between these: 3×5=15 and 4×7=28, cull to 8 and 14, 8×14=112 combinations.
The principle of the small variable is to always partner small variable pools, all else being equal.
But most of the time, it won’t be. What if the 5×7 option was such that 75% of the results could be excluded instead of half? Instead of 35, culling to 18, we get 35, culling to 9 – and 6×9 is only 54 combinations to consider.
Or things might go in the other direction, and the 5×7 combination only culls 25% of the options: 5×7=35, which culls by 9 to 26, 26×6 = 156 combinations – almost triple the workload!
I have gone into this in a fair amount of detail because I want to emphasize that these are NOT trivial choices, and NOT always obvious. The fastest approach is to list all the possible values for each variable, cull, and THEN decide which pairs to partner up, unless the choice is clear through the power of logical dependence.
The same approach can be used for the creation of Adventures, and even campaigns. It’s all a matter of listing the alternatives and using parameters to cull the choices down to a manageable selection.
I described at some length the process of writing the current adventure in the Zenith-3 campaign in Paving Over Plot Holes: A Masterclass in Adventure Creation. When I started working on that adventure, aside from the campaign-connectivity subplots, all I knew about it was that it was a mystery analogous to a locked-room mystery (but one in which a different parameter was the “impossibility”. So I listed my options, making sure to avoid all the pitfalls of bad mystery writing, culled the different combinations, and selected one. Then obfuscated the heck out of it.
I’ve used the same general approach when I had nothing but the name of a (minor) villain to go on – in fact, the encounter that leads off that Zenith-3 adventure was created in just this manner, simply because I had too many options to pick from, none of them compelling.
You won’t, and shouldn’t, need this technique every time. Encounters should, generally, flow logically from the adventure that they are part of, and adventures should flow naturally from the campaign that they are part of, and campaigns should be the product of inspiration and experience. But it is an indispensable tool in my repertoire for those occasions when something has to happen but you don’t know what.