Friday, 16 June 2017

Struggling with Attributes

Moving Away From The Big 6*3d6

In Explore I’ve gone through several iterations of attributes, moving further and further away from the traditional big six with 3d6.

Originally I started with the big six but wanted more balanced attributes without resorting to the usual solutions of escalating scores / point buy acrobatics / cookie-cutter stat arrays – I used this method to randomly generate six values 3-18 via 3d6 which always add up to 63 whilst keeping the same distribution as 3D6.

The next iteration was to replace the mental attributes (INT, WIS and CHR) with Memory and Intuition, so as to remove Logic and Personality attributes to make those issues purely down to roleplaying – and to split DEX into Agility and Reflexes.

Next I wanted some correlation between related stats. That’s tricky to resolve, but dropping the 3-18 and instead working only with the bonuses -3 to +3 made it possible. (It was a shame to lose the balanced 3D6 system though). I arranged the attributes in pairs – Strength/Constitution, Agility/Reflexes, Memory/Intuition – where the total sum of bonuses is 0, and each pair differs by no more than 2. Thus if you have STR 1 you cannot have a CON of -2 or -3. This was generated via rolling two D% and using this lookup table.

Using this system I then derived Height and Weight from the stats, which also gave me the consistency between the physical character and their stats.

This is the version I’ve used for the last two years. It works reasonably well, and creates varied, balanced and consistent characters. However when considering Climbing, which is about power-to-weight ratio, I encountered difficulties - when I put the numbers together something wasn't making sense. At this point I realised that if in the real world Height and Weight influence Strength, it would be simpler if it worked that way round in the game.

Taking a step back I bit the bullet and stopped blogging the revised rules, and decided to allow myself to follow any train of thought without worrying about the consequences, to see what I could come up with. Not for the first time I was glad I don't have a Kickstarter to complete!

Sacred Cows

Freed up to consider anything this quickly lead to the germination of a new system, but finalising it to something I'm happy with has been something I’ve been wrestling with on and off for the last five months. As you’ll see, mechanically it is only a slight evolution of the previous system, but the approach is very different. In particular it keeps the key benefit of the previous system - closely related skills still have bonuses which can differ by no more than 2 - and doesn't require any alterations to the rest of the rules (aside from a couple of +3s added to target numbers). In fact, apart from making character creation simpler and more transparent, it doesn't really impact on gameplay.

In the process I’ve ended up with a lot more attributes – something which I’ve previously been quite opposed to – but the reasons behind my opposition to that have evaporated. In addition I’ve removed randomness from character generation – something I’ve had in since day one. Both of these were a bit of a sacred cow, but it was making character creation difficult.

My goals are:
  • Attributes are broad brush bonuses to represent aspects of your character to give consistent characters.
  • Aspects of your character which you role play (problem solving and conversations) are not covered by attributes.
  • I want clarity of what an attribute bonus means – it should tell me something specific about the character.
  • I would like you to be able to choose attribute bonuses based upon what you would like your character to be like - that is without having to make dissociated decisions* during character creation.
  • There should be as small an overlap between attributes as possible.
  • It should be reasonably clear which attributes should give bonuses to skills.
  • I want to keep the same size range for bonuses to skills from attributes – seven steps from worst to best – as that has worked well so far.
  • I want only bonuses not penalties – my group was particularly enthusiastic about this change, despite it being only psychological.
  • It should be possible to excel at activities by having bonuses in different attributes. For example, you can be strong by being tall, or by being well built, or muscled.
  • The skills which have similar bonuses due to taking bonuses from similar attributes should be plausible. For example, acrobatics should be strongly linked to dancing, but not to lock picking.
In some respects it now resembles Rolemaster – which was a surprise as I'm a refugee from Rolemaster – and Size is of course an attribute in Runequest. Nothing is new under the sun – but every journey is different. On the other hand it still strongly resembles the prior system - strongly related skills (those sharing two out of three attributes) can only defer by at most 2 - exactly as before - but this "related skills" concept is now quite wide-ranging.

In the next post I'll unveil the attributes and what skills they apply to, and illustrate with a nifty diagram how I found problems and refined them away (well, I like it anyway!).

*It is often claimed that all character creation is dissociated - what I mean by dissociated character creation is the common situation where you make a choice about an aspect of the character purely based upon the mechanical effectiveness of the choice in the game rules, rather than your character concept.


  1. This is really clever.

    I've been working on a narrative character creation concept for my 2E AD&D rebuild (starting with Method I, but changing the stats based on players' answers to questions about their past) and grafting a linked attributes system onto that could be pretty cool.

    1. Thanks. I'll be posting details of the actual system shortly. I don't see any reason why it (or something very similar) couldn't be applied to AD&D. It's just a case of mapping the attributes to the things that get bonuses in the game, and the diagram I use for validating the mapping could be redone as well thus validating your choices. (The diagram I'm referring to should also be in the next post).