Decision 12 · Week 12 · 20 October 2026
Is ARAM Balanced, or Merely Corrected?
Turning the course's analytical framework back onto Riot's design of the mode
The claim
A mode that requires continuous per-champion correction is maintained rather than balanced.
The decision: whether the mode is well designed
Constraints this week is about
- no pick
- no recall
- one lane
- forced contact
For eleven weeks this course has taught adaptation to a system. This week the system is the object of study rather than the environment.
The mechanism#
Champions in this mode are not the champions from Summoner’s Rift. They are those champions with per-champion adjustments applied on top: modifiers to damage dealt and damage taken, to healing and shielding, sometimes to ability haste. The list is maintained patch by patch. Champions enter it, leave it, and have their numbers moved.
| Case | Damage dealt | Damage taken | Healing |
|---|---|---|---|
| Champion AA small correction. Arrived close to the intended outcome and was nudged. | − | − | — |
| Champion BA large correction in three directions at once. The interaction, not the number, is what needed repair. | − − | + + | − − |
| Champion CCorrected upward. Designed around a resource this mode removes. | + | — | + |
| Champion DNot on the list. Transplanted unaltered and produced an acceptable outcome. | — | — | — |
Schematic. The cases are constructed to show the shape of the list, not its contents: no champion is named and no modifier is quoted, because the argument is about the existence and spread of corrections, not their values.
This is not a criticism. It is a description, and it is the most interesting design fact about the mode, because of what it concedes.
Every one of those modifiers is an admission that a champion, transplanted unaltered into this environment, produced an outcome the designers did not want. The list is a record of the gap between how a champion behaves where it was designed to live and how it behaves on a bridge it cannot leave.
The argument#
Ask what work the word balanced is doing.
If balanced means statistically balanced — win rates clustered near even across the roster — then a sufficiently long list of modifiers achieves it almost by construction. Any champion whose win rate drifts gets a number adjusted until it stops drifting. The distribution can be flattened indefinitely. This is a solvable problem and the mode largely solves it.
If balanced means experientially balanced — that being on either side of a matchup produces a game worth playing — then the modifier list is evidence against, not for. A champion that requires a large correction to reach an even win rate is a champion whose fundamental interaction with the environment is wrong. The correction hides the outcome; it does not repair the interaction. The player on the receiving end does not experience a corrected win rate. They experience the interaction.
Same evidenceOne roster, one modifier list, and a win-rate distribution clustered near even.
The mode is balanced
- Win rates are near even across the roster, which is the standard the word normally means.
- Drift is detected and corrected within a patch cycle, so the distribution stays flat.
- By this standard a longer modifier list is evidence of better balance, not worse.
The mode is corrected
- The size of a correction measures how wrong the underlying interaction was.
- A flattened win rate hides that interaction from the table but not from the player in it.
- By this standard a longer modifier list is a longer record of displacement.
Hence the distinction the week turns on:
That is the point of putting this week last. The reasoning applied here is the same reasoning applied to augment rankings in week 4: a population statistic answers a question about aggregates, and the question anyone actually has is about a particular case.
The counterargument#
Every live game is maintained. Continuous patching is what a live service is, and treating maintenance as evidence of failure would condemn the Rift too, which has been patched continuously for longer.
This is the strongest response and it is largely right. The argument above does not survive as maintenance is bad. It has to be narrowed:
The Rift is patched because the metagame moves — players find things, and the design responds. This mode is patched for a structurally different reason: champions are placed into an environment they were never designed for, and the patching corrects for that displacement. The first kind of maintenance is a design responding to its players. The second is a design paying rent on a decision made once, at the start, and never revisited.
Whether that rent is worth paying is a genuine question, and this course does not settle it. Both answers are defensible. What is not defensible is answering it by pointing at a flat win-rate distribution.
What the semester was actually about#
Eleven weeks of optimising inside a system. One week asking whether the system deserved the optimisation.
Both halves are the same skill. Reading a composition you did not draft and reading a mode you did not design are the same act performed at different scales: taking something handed to you, working out what it actually does rather than what it is described as doing, and deciding accordingly.
Takeaway#
You do not need better rolls. You need better decisions — and the last of them is deciding what the thing you have been getting good at is worth.
Teaching team
- Spike — Convener — Game Systems & Decision-Making Lecturer