Learning From Experience

I recently posted a description in a comment to a blog post of how I handle experience and level gains that I thought I should include on my own site.  In considering how to write it, I realized it might be worth exploring a few other models of character development and learning as well.

Some Published Character Development Models

All character development models have some degree of abstraction to them, as a simple necessity.  It is impractical or impossible to fully model in a game how people learn.  However, different games model character development in different ways.  Let’s start by looking them over.

Dungeons and Dragons

Dungeons and Dragons, in all its incarnations, is largely gamist, with varying amounts of simulationist and narrative elements.

In all cases you gain experience points from defeating opponents (and in some editions, taking their stuff).  These experience points are a measure of how capable you are, as measured by the opponents defeated, and allow you to get better at what you do.  This is so even when ‘what you do’ has little to do with how you gained the experience points (kill a monster, get better at beating up monster… or writing magic scrolls?  What?).

Various editions have tried to find ways to improve the simulationist aspects.  AD&D 1e required training on gaining levels, AD&D 2e let different classes get different amounts of experience for class-appropriate activity (thieves get experience points for stealing stuff, fighters get experience points for defeating monsters (in single combat?  I don’t have my books here), spell casters might get experience points for casting spells and making magic items.  In D&D 3.x it wasn’t in the rules as written, but many DMs and campaigns required that any significant gains might require training or some other explanation.

“You’re a fourth-level barbarian… just when did you learn to become a wizard?”

“I, uh, watched Aristhomes prepare his spells every morning and it started to make sense?”

In most editions I remember seeing suggestions of ‘story awards’, small amounts of experience points gained for completing campaign goals.  I have also seen suggestions of ‘roleplaying awards’ for notable roleplaying at the table.

Still, overall D&D has classically used a gamist model.  Experience points are largely applied in ways that seem dissociated with how they are gained.  It has also led to grinding and extinctionist/completionist behavior (go looking for fights and kill everything in the dungeon to get all the experience points possible), though this might be somewhat limited by the guideline that a PC could only level “once per adventure”, to the point of being held at “XP needed for next level minus one XP” until he starts a new level.

Chaosium’s Basic Roleplaying

I consider Chaosium’s Basic Roleplaying system to be an attempt at simulationism.  Character growth, by and large, is basic primarily around the activities done.  When you use a particular skill you gain the ability to improve it later.  I don’t remember how, or even if, you could try to improve something you can’t actually do (how do you improve your POW score?).

Like D&D this leads to some odd behavior, such as repeatedly changing weapons in a fight so each skill can be exercised.  While I will admit the monastery fight scene in Crouching Tiger, Hidden Dragon was pretty cool because of the number of weapon changes, and Jackie Chan can be a lot of fun to watch because he changes between so many improvised weapons in his fight scenes, it looks pretty silly when it happens all the time.  Especially when it leads to someone standing over an almost beaten foe thinking through his inventory of weapons to find one he hasn’t used today.

HERO System

Perhaps oddly, I consider HERO System to be a largely narrative model.  Advancement isn’t particularly related to any particular activity, though you are likely to perform certain activities (fighting) in most sessions you gain Character Points for.  These Character Points may be applied to just about anything about your character.  However, the GM also has the option of granting ‘specific Character Points’ in the form of small rewards such as Favors or other Perks, usually derived directly from story elements of the session being rewarded.  Save the wizard’s familiar, he promises that should you call on him he will try to help you.

The player-chosen application of Character Points is somewhat dissociated from how they are gained, in that you can do one thing but get better at something unrelated, but between the Character Points being given for taking part in the story and (often) other rewards being derived directly from the results of the story, I’d consider this a largely narrative model.

Weird, given how the rest of the game works.

How I Do It, Basically

I mostly play d20-based games (for the last 11 years mostly D&D 3.x, though I’ve explored some related options such as Iron Heroes, Mutants and Masterminds, and Pathfinder — which is to say, D&D with different wrappers.  However, I ditched the D&D experience point system more or less entirely years ago for a couple of reasons.

  1. I didn’t like the specific dissociation between how experience points are gained and how the character grows.  I don’t mind abstraction, but ‘kill stuff to get better at writing scrolls’, especially for a nominally ‘bookish wizard’, never felt right to me.
  2. Any kind of ‘experience point cost’ to things really didn’t work as it was presumably supposed to.  The costs were either minimal and had no real effect (seriously, spending 6 XP to write a scroll of  a second-level spell?  Who cares?”) or were enough to slow someone down (about one session worth) in gaining the next level… during which session he gains experience points faster than the people who did level because he is lower level against the same challenges and thus pretty much catches up, or possibly even gets ahead of them.
  3. It’s frankly a pain in the ass to keep track of.

Instead, I handle character development in a largely narrative fashion.  I don’t care how many monsters you beat up, I don’t care how much stuff you take, I’m primarily interested in the completed actions and story points.  Note that ‘completed’ is not the same as ‘successful’; I don’t much care if the princess lives or dies, as far as character advancement is concerned.  The people in the game world do, though.

This is important.  I want to encourage the PCs to see things through to resolution, one way or another, so I provide the meta-game reward (levels) on resolution.  ‘Success’ (mission completed with happy results) or ‘failure’ (mission completed with unhappy results) is an in-game consideration and thus has in-game consequences.  Abandoning a scenario likely counts as a failure (princess still isn’t rescued) and thus may have in-game consequences, but can also be expected to have meta-game consequences (no level for you!).

As a mostly contrived example with some obvious flaws:

  • Successfully rescuing the princess is good (and gets you rewarded). XP for you, and probably her father’s gratitude.
  • Failing to rescue the princess because she died is bad, but does end the scenario. XP for you, but possibly her father’s enmity for your failure, and possibly that of the duke who abducted her because you caused his plan to force the king’s hand to fail.
  • Failing to rescue the princess because the Evil Duke managed to outfight you and escape with her is unfortunate, but may end this scenario, in which case: XP for you. And possibly a recurring villain (or villains, if she decides to turn and join him…) for another time.
  • If you screw around, beating up the Evil Duke’s guards instead of focusing on the goal, and he just walks away with her, no XP for you because you didn’t complete the scenario. It completed on its own and you were irrelevant.  Her father’s reaction may be somewhat neutral (she is still alive, and may in fact be in a better position to influence his opponent, her abductor), the Evil Duke may decide you are a potential future threat and target you for removal, or he might decide you’re inept and not worth the worry, depending.

Basically, you have to see yourself to an actual endpoint of the scenario and be relevant in the result.  I don’t much care how you get there.  If you sneak in, grab the princess and teleport out without seeing anyone else, that’s good enough for me.  And I clearly underestimated your abilities and will plan accordingly next time.  Mind you, you probably also missed out on a number of other results, including possibly treasure, preventing or setting up other situations, and so on, but you did achieve the goal and rescue the princess, completing the scenario.

Designing for Play

Above I describe how PCs advance in my D&D(-like) campaigns.  Here I describe how I apply this when preparing for my players.

I like to follow the ‘node-based design‘ guidelines I have described elsewhere, where I outline each scene, scenario, and campaign and the relationships between the various pieces.

Note that as much as I talk about designing the play elements, I typically keep things at a fairly high level unless and until it looks like they will come up.  If I design a campaign I will identify the individual scenarios that are likely to come up, definitely identify key scenarios, the likely consequences of each and the relationships between them, then stop.  I really only drill down to higher detail when it is likely to be used.


D&D talks a great deal about encounters (often combat, possibly social, traps, and so on).  In D&D 3.x the XP gain is often dependent on the specific creatures met.

Too much detail for me.  I like to think at the ‘scene’ level, where a particular part of the story may play out.  These usually resolve a minor plot point in a larger story, and in a scenario there are only a handful of key scenes.


A scenario tells a particular story or resolves a moderate plot point in a larger campaign.

If a particular scenario is a ‘standalone’ scenario there should be enough scenes to warrant the involved PCs gaining a level at its completion.  For these scenarios I usually plan about 15-20 scenes total, probably with 3-5 key scenes.  The remainder are largely optional — it is likely necessary to play through a minimum of 8-10 scenes in order to complete the scenario, but I don’t specify which additional scenes or their order of completion.

In a larger campaign an individual scenario is likely to be smaller because it doesn’t need to stand alone.  There might be 5-8 scenes total, with 2-3 key scenes, and the scenario might be completed in as few as 3-5 scenes (depending on the size of the scenario).


A campaign tells a larger story, often consisting of several related smaller ones, or resolves a major plot point in the campaign.

I have found in my groups that 3-6 scenarios relating to a single topic is about as much of that fun as they want to have before they do something else (possibly with the same characters, possibly not) and I plan accordingly.  I usually outline about 4-5 scenarios and expect the PCs to run through at least three to complete the arc, and up to six if they are really enjoying it.

Just as my scenarios are designed in a fairly non-linear fashion, the same techniques can allow me to design my campaigns to play out in a non-linear fashion.  As mentioned above, scenarios designed for a campaign tend to be somewhat smaller than standalone scenarios, in order to allow for sufficient choice for the players.


A setting is any set of related campaigns and scenarios, where ‘related’ might be as weak as ‘happen in the same place with much the same cast of PCs’ (as you might see in a sandbox game).

I don’t necessarily plan these, since things often change enough at the end of a scenario or campaign that it’s not worth my time or effort to devise the relationships between them.  I tend to note ideas for campaigns and scenarios and some potential relationships that may lead to them (in order to not lose the ideas, for a start, but also so I can lay groundwork for them to hopefully engage the players before they come up), but overall I don’t usually put as much effort into planning here.

Application in Echelon

The above works pretty well in D&D 3.x.  PCs gain levels when they ‘complete story elements’, successfully or not.  The number of scenes (encounters or other challenges) is somewhat variable, and I like that.  A clever plan to steal from the dragon should be rewarded at least as much as killing the dragon and taking its stuff.

The numbers above also work out rather well for Echelon.  If I’m sandboxing I can have each scenario be standalone and lead to a level gained, with every fourth or fifth level being a greater scenario that will move the PCs to the next tier.  In fact, it might take two scenarios to do so — I like the idea of having some extra time to enjoy their new capstone ability before moving to the next tier.

Given the level breaks, the top-level Expert gets some extra time to enjoy being at the top of his tier before becoming a ‘minor niner’ Heroic character.

For campaigns, perhaps the primary difference is in how the scenarios are related.  Instead of ‘random’ scenarios, specific goals are completed.  An ‘expert campaign’ could end with the PCs gaining minor titles amongst the gentry, while a ‘heroic campaign’ might see them move into the lower nobility as a reward for their actions.  There is a built-in structure to distinct improvement and advancement that fits rather well.

Closing Comments

This has largely removed the concept of grinding for XP. They went to the tomb of the legendary warrior to borrow (seriously, they asked permission and brought it back) his sword to fight the adversary because it helped their goal, they skipped a couple of side quests because they didn’t need what could have been gained (as missed out on some things they could have used), and so on.

All in all it seems to work pretty well.


  1. GreyKnight

    I have a similar conception of “levelling up after completing a quest”. If it’s a side-quest that seems too minor to be worth a level then perks and favours of the kind you mention under HERO never go amiss. If you’re completely stuck for ideas then extra loot will always go down well! :-)

    Training skills by using them seems like the most sensible approach, but the resulting behaviours are not quite right. In Crawl it’s nicknamed “victory dancing” as you spend time after killing a monster spamming the desired skill to move experience points into it. I wonder if there’s a way to make usage-based training work properly?

  2. Nagual

    I have a solution — NO XP — level up when:
    A. Campaign events suggest/require it
    B. The players got their tactics well worked out and encounters at the current level have become stale

    I use action points as the primary “reward” system — for good role-play, completing adventure objectives, etc. No XP means I can be more creative in my scenario design — no more “13 encounters = level up” bullshit! Plus the action point system gives the players a little more control over the pace of an adventure.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Back to Top