I must create a system, or be enslaved by another man's. I will not reason and compare: my business is to create.

- William Blake

Monday, December 28, 2015

Granularity: the ideal level of detail

This post by Delta has been a great inspiration to me since I read it a long while ago. It uses Miller's Law to discuss how many options are "too many options". It is a great post - go read it if you haven't yet.

According to Wikipedia, Miller's article ("The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information") "is often interpreted to argue that the number of objects an average human can hold in working memory is 7 ± 2".

There is more to it than that, as you can see on the link, and the whole notion is somewhat disputed (there is mention of a "Magical Number four", for example), but Delta makes a very interesting argument based on the number 7 (and often less) to explain why so many people might find 3e "too complicated" when compared to OD&D. I feel the same -  I really like having lots of options, but I also think they can be overwhelming.

This whole thing got me thinking about what is the ideal level of detail for my games. Should I use a d100 range for traits? Should I ignore +1 modifiers? Should I use a d6 for skills, or is it not detailed enough?

Bear in mind that all of this might be coincidental, and has a lot more to do with personal taste than mathematics or psychology.

So, what is the ideal level of detail for me?

As you would imagine, the answer is often... seven.

Take B/X ability modifiers, for example; they range from -3 to +3. So, you have seven degrees of Strength, for example (-3, -2, -1, 0, +1, +2 , +3). Sure, there might be some difference between Strength 10 (+0) and Strength 11 (+0), but it is ignored most of the time (and I appreciate the illusion of difference between 10 and 11, but that is a subject for another day).

Curiously, when the Fate RPG was created, the creators played with the idea of having three tiers of "strong", according to Rob Donoghue: The idea was to let a character be strong, stronger or strongest. But it also allowed someone to be drunk, drunker, or drunkest.

There something very straightforward (and Orwellian, I guess) of using a couple of words when describing all possible human tiers of Strength: weakest, weaker, weak, (ordinary), strong, stronger, strongest.

This applies to lots of RPGs, which use three levels of "Strong", for example, but only one level of "weak", since heroic characters will seldom have many "weak" traits. Both the Storyteller system and Savage Worlds, for example, use a kind of Weak/Average/Strong/Stronger/Strongest progression, going from one to five "dots" or from d4 to d12.

The Fate RPG, on the other hand, would adopt an "adjective ladder", with more tiers - which causes some confusion for me, because sometimes I forget if "Fantastic" beats "Superb" or if "Epic" beats "Legendary". It would seem that for this level of granularity numbers would be better than words.

Now, let's stretch the concept a bit further.

If you have seven possibilities with the same likelihood of happening, each has a 14,2% chance of taking place.

Well, this happens to be the granularity I like when dealing with modifiers, which is probably why I dislike using +1 modifiers when rolling a d20. Using a d20, the closest to 14,2% is a +3 bonus - which is what you get when you invest in a NWP in AD&D 2nd Edition. Some versions of old-school D&D use +4/-4 as the "go to" modifier when rolling 1d20.

But what if you're are using a d6? Well, +1 in 1d6 adds about 16,7% chance; very close to the ideal level of granularity I need.

Fifth Edition D&D has substituted most modifiers with a "2d20 pick highest/lowest" system, and that was one of the most universally admired features of this new version, because many people thought that those small +1 modifiers were too fiddly and somewhat of a nuisance. This method influences the results in different ways, depending on the Target Number, but on average the difference is 3,5 - or 17,5%.

By the way, the average result of a d6 is 3.5. Which means that adding a d6 to a d20 or to 3d6 has similar results... and may explain why the classic method of "roll 3d6 under ability, 4d6 if the task is hard, etc., works so well.

There are exceptions, of course. I still use the d20 and +1 increments for combat, since in the long run having a +1 bonus will give you an edge more than 5% of every combat. Also, if you take the "plus or minus two" idea into account, increments of 10% to 20% instead of about 15% could work too (using +2 to +4 to a d20 roll), if you prefer even numbers.

Coincidentally, I've recently realized that I also like the idea of SEVEN OUTCOMES for most actions. It is more than most people are used to (unless they play Fantasy Flight Star Wars, I guess), and certainly more than you need in most situations, but dividing outcomes in worst, worse, bad, ordinary, good, better, best, has worked for me; I can easily think of each category for most uses of a skill or ability.

Not all outcomes have the same chance of happening, of course.

You can find an explanation lots of examples in this post.





9-120Neutral, as expected




You can go the Rules Cyclopedia way to make it simpler (five outcomes, still within "seven, plus or minus 2"):




6-8Neutral, as expected



Again, all of this might have no basis in science; it is one of those happy coincidences that work for me, but might work differently for your group.

So, what's is the point?

In conclusion, systems that bother with +5% bonuses are too fiddly for my tastes, and these modifiers are too often ignored by my players. If your tastes are similar, you can keep these ideas in mind when designing your own games or house rules for your favorite systems.


  1. Interesting post! Your approach reminds me a bit of the Monster Reaction Table in the RC (5 options) and is (incidentally) what I chose for my NPC Reaction Table (2d6, 7 results).

    That being said (and bearing in mind that it is about personal preferences), I'd like to add that, although interesting, it lacks the exploration of hierarchies in that "magical number 7" concept: if you are able to remember 7 hierarchies, each resulting in 7 subcategories, you'd have easy access to 49 pieces of information. And, consequently, the use of rule books (or DM Screens!) at the table, which would reduce your chart above to 1 item instead of 7 (because you know where to look for skill resolution results!). Same goes for using hierarchies on character sheets.

    I also have met players that criticized the lack of "depth" in rules light games like Epees & Sorcellerie and came to the conclusion that at least the impression of complexity helps reminding players that a game is run by rules and not by arbitrary decisions.

    So for me it would be about hierarchies and references. If I (as a DM) know where to look and have easy access to that information, I feel comfortable with very complex rules. For players it would be somewhat the same, but with the restriction of no more (or less) information as you'd have room on a Din A4 character sheet.

    Especially in game design it's crucial to think about access to and the necessity of information (with the important distinction between DM and players in mind) and if there also is a lower limit to those ideas (as in: how much is not enough).

    That is not to say that a rule of 7 doesn't have merit. I'd totally use 3 to 7 as the number of hierarchies to allow easy access to information. But how complex those rules and pieces of information themselves are, depends far more on how often they are used at the table and how fast I need (final) results.

    Nonetheless, interesting post and great food for thought!

    1. Thanks fro the comment! You make some interesting points. Good catch about the Monster Reaction Table in the RC, it was exactly what inspired me - take a look at this post:


      I think I get what you're saying about hierarchies, but could you give me some examples in RPGs?

      Although I think good rule books/DM screens can make for easy access to the rules, nowadays I prefer rules I can (mostly) keep in my head.

      Of course, this is all very subjective. I played GURPS for a long time and still think its an easy game to run, even with hundreds of skills, advantages, perks, etc. And even B/X can be too complex for my tastes when regarding XP per level, etc.

      "at least the impression of complexity helps reminding players that a game is run by rules and not by arbitrary decisions" - this is a great concept, and deserves some reflection. It just might explain why I prefer 3-18 abilities even if they are just -3/+3 most of the time.

    2. A few good examples would be from what I know to use: the RC. I know my way around the chapters (first hierarchy), so if I were to look for a specific skill, I know Chapter 5, skills in alphabetical order (second hierarchy). That's enough to find something you won't need all the time fast enough to make it relevant for the game. Same goes for Monsters (most of the time, anyway). Treasure tables are also good organized in that book and the basic idea to have all rules in one book is still one of the RC's biggest features. But the lack of structure in the same book gives lots of other good examples.

      Although, for instances, Monsters are fast to find, it'd been nice to have pages referenced from chapter 7 (Encounters) to chapter 14 (Monsters). Or better yet, have that whole chapter 14 right after 7 ...

      Spells are easy enough to find, to give a second example, but I find it somewhat counter-intuitive to have the cleric spells first, then the druidic spell (?!) and only after that the magic spell (I'd start with magic, then clerics, the druids ...).

      Some of it is visual, too. So just by skimming through the book, I have a good idea what happens on specific pages. I have seen this done better (Better Than Any Man is very well organized, for instance, same goes for Stonehell, to give two examples of books where good hierarchies are crucial to make them useful), but the RC does many things right in that regard.

      I am able to DM a game like you describe and will do so if I have to improvise or play online. But if I got time to prepare and all the rules I need at hand, I'll go for more crunch every time (with HackMaster 4e being the most complex game I liked to DM ...). And it's not as much that I don't like rules light systems, I just think that those games more often than not leave too much room for DM fiat, so I keep it as random as possible and will roll everything I can to make that work (people already made fun of me for rolling the weather).

      Now that I think about it, rituals are other good examples of how complex rules are used at the table. Time is another factor here, to make it a bit more complex. Say you have a rule that everyone rolls 3d6 for a combat round, the sum of that roll is the individual initiative, the separate dice are number and value of the dice you are able to use that round. Say furthermore that 1s are discarded and don't count for initiative, while sixes not only count for initiative, but also generate a new die that is rolled and also added (as long as it isn't a 1, of course). Because of the extra rules for 1s and sixes, the spread is that of a d20, btw. And that's a (short) example of how combat in the game I'm working on is played. What might sound complex right now, is something players will grasp after the first round and continue to use from round to round, because it's always the same. It's ritualized ...

      There are way more examples of how dynamics like that help using complex rules in a game, but that might be worth a post in itself and I have already written too much :)

    3. Yeah, I see what you mean. Good organization might make the most complex systems easier to grasp, and vice-versa. The RC is a great example.

      Also, familiarity is relevant somehow: if you're very familiar with a system, it is easy to use every important detail and to ignore the unimportant ones.

      There might be a good argument about allowing complexity in things that you don't use often (such as specific skills) and avoiding it when dealing with stuff you deal with all the time, such as HP.

      Ultimately, detail is a matter of taste, like salt, I would say; a little makes things tastier, a lot and it makes it uneatable.

      I like your idea to leave things to chance rather than DM fiat, this is the way I like things too. Since I have a hard time remembering all the rules and don't like checking books often, I try to use the same method for different situations.

      By the way, I see nothing wrong with rolling for weather; we already have a good table for wind in the RC, why not use the concept further!