While watching the special features from Numb3rs season 3 on DVD, I got to thinking about one of the phenomena of TV shows – that some episodes you really like, and some you don’t, and some episodes are really popular and some are not (and these categories never completely coincide. This is true even of my favorite TV series.

At the same time, I was turning over in my head various ideas for blog posts, and one of them was about the efficiencies of improvement, and how they could be taken into account in planning your game prep to give the greatest bang (ie overall improvement) for your buck (in this case, for your development effort).

I’m not an expert in time-and-motion studies, I should state that up-front. In fact, I firmly believe that people are different enough that such studies (however appliccable in general) are not always 100% on the mark in individual cases. That’s why real elections never perfectly follow the predictions of polls, and why no-one ever agrees completely with the ratings of anything – be they TV shows, movies, or whatever. Statistics are great at overall situations, not so hot in dealing with individual situations. So everything I’m about to write should be taken with not a grain of salt, but a pillar. On the other hand, sometimes an amateur who’s not afraid to experiment can see the forest more clearly than the expert who starts by focussing on the species of tree!

Another ‘fair warning’ while I’m giving caveats: I don’t really know exactly where this blog is going to go, this is very much a case of thinking out loud. I have some ideas as to approach, but what conclusions I will reach – if any – are a complete mystery at this point. That’s the exciting thing about this particular post!

The Objective

The Goal of the process that I’m hoping to develop is to increase the amount of fun that everyone has at the game table. The methodology of achieving that goal is what this blog post aims to develop, by determining some means of specifying which areas of game prep can benefit most from a little extra attention on the part of the GM, and which areas should be shortchanged to get the time required. But at the same time, I want to increase the GM’s fun as well, to make game prep less of a chore and more of a hobby.

That means that two seperate qualitative factors will have to be taken into account: how much the GM enjoys doing it (or, more specifically, how much work it is for the GM), and how big a benefit the campaign will recieve from extra attention. The first is analagous to the rating you as an individual would give to the various elements of game prep, if you were measuring how much you enjoy doing them. The second is analagous to the ratings your players would give at the end of a day’s play – how much fun everyone collectively has at the table.

A Working Taxonomy

Everyone divides up their tasks into subtasks, and everyone does it differently. As Johnn and I have found while working on another project, creating a unified and comprehensive taxonomy for everything that goes into a roleplaying game is incredibly difficult and full of hard choices.

To try and keep this post on-topic, I’m going to simplify the stages of game prep into a small number of substages which are necessarily less detailed and less comprehensive than would be used in a real life situation. Those substages are:

1. Story – what is going to happen in the course of the day’s play, or the adventure overall
2. Metaplot – how things fit into the big picture, any external limitations or complications that the PCs will have to take into account, and the long-term significance of the adventure
3. Background – the origins of the current situation, anything needed to make the plot more challenging by limiting or restricting the PCs scope for choices, and anything else that is needed to justify the in-story events.
4. Locations – maps, blueprints, and location descriptions
5. Encounters – who or what gets encountered (in generic terms), and what that encounter adds to the developing awareness of the plot by the PCs
6. Personalities – the roleplaying elements necessary to take that generic encounter and make it unique and individual
7. Character Mechanics – the game mechanics used to describe the character in-play
8. Script – snatches of prewritten dialogue, bullet points of information that NPCs are to relate to the PCs, and any narrative that the GM is to provide ex-cathedra
9. Support – props, miniatures, any game supplements that should be on-hand, and anything else that can add to the playability of the adventure but that might not be strictly necessary

Okay, the list is longer than I originally thought it was going to be – I kept thinking of ‘one more thing’ to add. It has holes: where do rewards fit into it? Should they be their own category, or under Character Mechanics, or Encounters? But it’s sufficiently functional for today’s purposes, it just means that any examples might need to be bigger than I was expecting.

Developing Your Own Taxonomy

I should emphasise that each GM should develop their own taxonomy to describe what they do in game prep – because not every GM will do all these things, and certainly not every time. Some GMs might not sperate Personalities from Character Mechanics, or Metaplot from Story, or Metaplot and Background. The next time you’re doing game prep, you should generate your own, unique, taxonomy – the discriminating factor being whether or not you think of it as a seperate task, something you can do at another time.

Dependancies

Quite often, a GM will consider one of these tasks to be dependant apon the outcome of another. Some GMs will define the personality of an NPC first, and use them to derive the game mechanics; some will generate the NPC’s stats first and build the personality as a reflection of those characteristics. Still other GMs won’t seperate the two into seperate tasks at all – they might start with a general idea of personality, develop the character stats, refine the personality to accommodate the game mechanics, then tweak the character. Or it might be as simple as getting the personality down on paper while the game mechanics are fresh in the GMs mind or vice-versa – this is another way of linking the two into a single task.

A similar relationship exists between Metagame and Story, and between Metagame and Background, and between Story and Background. How much does the location influance the nature of the encounter that is to occur there – and which comes first?

The rule of thumb to use in developing your own taxonomy is that if a step is completely dependant on the results of another step, those two should be combined into a single step in the GM’s thinking.

How Much Fun?

Having established the categories, the next step is for the GM to rate how much fun he has actually doing them. A one-to-five scale is fine, where zero = ‘I have to be forced to do this at gunpoint’, and five = ‘I would do this every day if I could’. We’ll label this “R” and use it to determine how much the GM tries to avoid doing the task. The illustration is a rough indicator of how I would rate the elements of the taxonomy I derived as an example.

How Easy?

The next step is to determine how easy the task is for the GM. Every one of us has strengths and weaknesses. This value will also be used to rate how much return the GM gets on his efforts. I could have the GMs rate it, but it should be more enlightening to determine it by comparing the time spent on the task with the number of pages of output. This method also takes into account the actual tools that the GM uses, whether it be a fractal mapper or character generation software.

Data Capture

The first step is to gather the data. For two, four, or five prep sessions, track how long you spend on a task, and the number of pages of information that you produce. Don’t worry about filtering out diversions and time lost to things like phone calls – those represent the inevitable overhead that will come from interruptions with any longer-duration effort. All we care about is the start time and end time and the difference in minutes.

Becoming more aware of how you work is a by-product of this information that can yield it’s own benefits. For example, you might find that you had clear ideas of the NPCs that you had to create, but that by the time you had finished doing the maps for an encounter, that information had become vague and hard to remember – meaning that you can make an immediate gain in efficiency by generating the NPCs before you start work on the maps.

The example above shows some fictitious results based very loosely on my own work practices, spread over four different game sessions. The first session totals exactly 4 hours of prep time, the second totals 5 minutes more than that, and the other two total less. All told, exactly 900 minutes of prep time have been divided up – an average of exactly three-and-three-quarters hours per game session. It can instantly be seen that I spend most of my time on the story, on the metaplot, and on scripting narrative and dialogue – over 60% in fact – leaving less time for anything else.

Data Conversion

Once you have a reasonable basis for your calculations – the work done over several game sessions – you can get the totals in each category of your taxonomy and perform a simple division to get the number of pages per minute. Don’t be surprised if the results seem alarmingly low – it can take an hour or more to write a page of text, and that would translate to 0.016 pages per minute. The next step is to adjust these to fit the same 1-to-5 scale that we used for fun, and will also use in later steps.

To make the conversion, subtract the lowest value result from the total of each step. Multiply the result by 4 and divide by the highest of the individual results. Then just add 1. The example to the right shows this being done in two stages: the subtraction and then the rest of the calculation. The result rates the efficiency of the GM’s work on the 1-5 scale that we need.

The first thing to notice is that all the activities that the GM has rated as enjoyable are in fact his least efficient – if the R is 4 or 5, the efficiency is ranked as 1, 1.2, or 1.3. The area in which I show the greatest efficiency is the one on which I spend the least time, and the one that I enjoy the least (and so would tend to avoid), character mechanics. So that’s already really interesting information. Even though the page count for the things I enjoy doing is high, so is the amount of time that I’m investing in them – probably because I enjoy it!

The question we can now begin to examine, having been armed with this information, is whether or not that degree of effort is really justified, or should I improvise more plot and spend more time on the things that I enjoy less? How much could I improve the game – and how much would I detract from it?

Extreme Session Post-Mortems

The next step is to do some session post-mortems. Pick two sessions that stand out in your mind – one for being more successful than most, and one for being as close to an absolute disaster as you can remember. In each, attempt to categorise how effective each prep element was at contributing to the enjoyment, based on comments at the table, your own observations, and any feedback that you remember.

What we are really interested in here are the totals, and the differences, between the two. Presumably, in the one that everyone enjoyed, everything would rate high – but there can be surprises.

I have two particular sessions in mind – one where I lost track of where we were in the plot, which limped a bit to begin with, and in which I got the abilities of a couple of characters confused. Even in this absolute disaster, there were a couple of things that people enjoyed, and a few areas that were complete messess. I can also remember a session in which everyone had a LOT of fun, everyone was laughing at the table and participating – but there were a few moments when things faltered, when things didn’t work as well as they should, and in which I had to scramble to keep things on some sort of sensible track.

The totals rank the strengths of the campaign in the perception of the players – the things that you are doing right. The higher the total out of ten, the better. The differences rank the differences between the two sessions – in other words, the reasons why one was a disaster and the other was not.

Above are the ratings I have assigned to the two sessions I have in mind, together with their totals and differences. And right away, a number of conclusions leap out at me. Metaplot is my strong point, that’s no surprise; but second most important is story, and that was also one of the big differences between the two. Ranking only marginally behind story in importance are background and scripted elements, the first of them being ranked the same between the two sessions, but the second being only just a little bit less critical than story in letting the second scenario down. Only less important again are Personalities and Character Mechanics, and here is where things really start getting interesting: Personalities ranked as the equal biggest difference, while character mechanics actually ranked higher in the disaster than it did in the success. Right away that tells me that less time spent on the game mechanics of the NPCs and more time spent working up their personalities or story might have made the difference between a mediocre session (at worst) and the disaster that it proved in hindsight to be.

Putting The Picture Together

In a perfect world, the last set of results would be all that mattered, because a GM could spend as much time on each element as it needed; it would only be a question of establishing a minimum standard in each for play, and you would have a winner every time.

Unfortunately, we all live in the real world, where time is limited and we absolutely have to take efficiency into consideration. (And now we get into the section of the article where I’m not really sure of my technique. But let’s dive in, anyway).

To start with, the activities that I enjoy doing are obviously a part of my GMing style, and so these will obviously bias both the rating of importance, and the priority that I should give to additional efforts. We need to take that bias out of the importance to get an true picture of the priority that should be placed on changes, and we need to take that bias into consideration in determining how much value the overall game could achieve from more attention to a given area. In effect, we need to take the bias out of the sum column and add it to the difference column.

That means we’re calculating Sum – R and Diff + R, and the results come out as shown. The higher the Sum-R, the more valuable the activity is to my GMing style, regardless of how much I do or don’t enjoy it; this is a road map to how my priorities should be adjusted, in an ideal world, to play to my personal strengths and GMing style. While the heavy focus on metaplot is not a surprise, the next highest score is assigned to Character Mechanics, which is a bit of a surprise, then background and encounters – both of which are only mediocre on my radar.

The higher Diff+R, the more benefit will be achieved overall by more pages of work, again in an ideal world. Here, story is king, followed by personality development and script, then metaplot – and then a number of items of comparatively low value. Again, none of these are universally applicable, these are values specific to my style of campaigns.

Clearly, we need to put these two values together. Simply adding them won’t do it: some values will cancel out. While a more complicated operation might be needed, my intuition says to multiply the two together and take the square root. The result should by a priority weighted by benefit. Finally, we need to adjust for efficiency, by multiplying by the value derived earlier.

The ideal game prep

So what does it all mean? Well, ultimately, there are three groups of values in the final column above: A couple that are 9’s, or close enough to it; a number that are between 4 and 5, near enough; and a couple that are ranked 0. But those are my results, and are unlikely to corrospond to anyone else’s.

What the results add up to is a priority table for dividing up extra prep time. So, to get the ideal prep time, let’s divide the standard prep into two componants: the fundamental and the extra. According to the statistics gathered early on, the average prep time I use is about 4 hours. So let’s reduce that in half-hour increments and allocate the time thus made available according to this priority.

That sort of allocation is easy, it’s something that most GMs learn how to do from compiling tables: you simply convert the results to a percentage of the total and split the allocated resource – be it a die roll range of results, or, as in this case, a quantity of time – into that percentage of the total.

So here are the results (I’ve only shown the calculations for the first couple of coulns). Four plans, each progressively shifting the balance from the current time spent (on average) to a 50:50 balance with the new scheme. Which one is best? Well, my gut instinct tells me that Plan 3 is the way to go. There really isn’t all that much change between plans 1 and 2 and the current balance – we’re talking 7 minutes less spent on the story, 2 minutes less on the metaplot, and so on. The differences are actually fairly negligable.

But that’s the point: these numbers are reflective of an already-successful ongoing campaign, which means that it has to have more hits than misses. If the campaign were in trouble, I might be tempted to make more radical changes, and that’s what the results would indicate. I would round off the odd values – the 33 minutes would go up to 35, and the 29 minutes to 30. If you can find 4 hours, you can find 4 hours and ten minutes – even if the last ten minutes are at the gaming table; if there is a discernable improvement in the adventures, the players won’t begrudge an extra couple of minutes.

Any of these plans would retain the elements of my GM style that work; the differences are in those parts that I occasionally struggle with, and how high a priority they should be. And, of course, having made the recommended changes to my usage of prep time, I can always repeat the calculations using new values to assess the results. Improvement should be an ongoing process, not a one-time event.

Conclusions

One thing that struck me while I was playing around with numbers here was that there were relatively few surprises. In most cases, these were trends that I had already detected in my game prep. I was going to say that this was perhaps the most surprising outcome of the lot – but then I realised that it makes sense. As a GM, you tend to pick up on things that don’t work quite as well as you had hoped or expected, and tend to perform session post-mortems in your head anyway (especially when preparing for the next session). The more experience you have, the more adept you will usually become at detecting these hints and signs. None of us ever gets perfect at it, but we make corrections to our approach all the time, based on experience of what works and what doesn’t.

But there have been a few genuine surprises along the way. Some of the relative efficiencies were unexpected, and the emergance of more work on the NPCs as the most useful thing to prioritise for overall gain was a surprise. That’s useful information.

Postscript: Where to from here?

Okay, I’ll reiterate that I don’t really know what I’m doing here. The ideas seem sound to me, and the results appear sensible, but that’s not necessarily any indication, and I have absolutely zero evidence that this works. But, assuming that it’s all correct (by some miracle), the next step would be to rate individual sessions and link them with the actual prep time done. Get some hard statistical data for what works and whether there is a consistant pattern to the failures. Lose the assumption that the differences between a mediocre game session and a flop are the same as the differences between a success and mediocrity – or at least get some hard data. I would like to compare an average session with a failure with a success, instead of a success with a failure, and develop two different plans – one to avoid the clunkers and one to excel more frequently – because they might prioritise different activities.

I’ve done all I can (and more) – I guess it’s now over to someone with greater expertise in statistics and time-and-motion studies…