Standing caution. This log generalises from two people, occasionally three, across eleven games. A large part of what it describes could be our own dynamic rather than a property of these designs, and the section at the end says exactly what would overturn it.
What is being counted
A friendly-fire event is any action by one player that damages, knocks down, incapacitates or otherwise penalises another. Water from a hose, a swung tool, a vehicle collision, a spell with an area effect, a door closed at the wrong moment.
For each game we recorded two things: which player the game's own logs, scoreboards or failure messages attributed the event to, and which player actually initiated it. Those two are frequently not the same, and the gap between them is what this log is about.
| Game | Mechanic | Attributed to | Effect on play |
|---|---|---|---|
| Smoke Order | Pressurised water | Player hit | Searcher stopped entering first |
| Hollow Court | Area spells | Player hit | Melee players kept distance |
| Quarry Deep | Falling debris | Player hit | Players stopped working the same seam |
| Relay Station | Electrical discharge | Player hit | One player refused the switch role |
| Ash Ferry | Shared health pool | Both equally | Neither could tell what happened |
| Second Shift | Collisions, dropped items | Player who dropped | None, penalty trivial |
| Cold Store | Crate collisions | Player hit | Mild, comedy absorbed it |
| Tower Watch | Misaimed defences | Player hit | Ground player stopped advancing |
| Fen Crossing | Shared momentum | Player hit | Session ended at 12 minutes |
| Night Freight | Vehicle contact | Neither | None |
| Split Signal | Rotation conflicts | Player hit | Both stopped acting decisively |
Why the careful player absorbs the fault
Not malice and not laziness on the designers' part. It follows from how these events are detected. A collision, a water jet or an area effect is easiest to resolve at the point of impact, and the object at the point of impact is the player who was hit.
Attributing it correctly means tracking causation backwards — which player initiated the action, which is more work and, for some mechanics, genuinely ambiguous. Eight of these eleven games take the easy route.
The behavioural consequence is consistent across every one of the eight. The player accumulating penalties changes their play to stop accumulating them, and in every case that means doing less: standing further back, entering later, refusing a role, keeping away from a shared work area.
In all eight games, the player being wrongly faulted responded by contributing less. Not one of them responded by talking to us about it, and in six cases neither of us noticed until we read the session notes afterwards.
The three that do it differently
One game attributes correctly. Second Shift logs a dropped item against the player who dropped it, and the result is that nobody adjusts their behaviour, partly because attribution is right and partly because the penalty is small enough not to matter. Those two things reinforce each other.
Two games attribute to nobody. Night Freight simply does not record vehicle contact, and Ash Ferry uses a shared health pool where damage is indivisible. Ash Ferry is the more interesting case: with no way to tell who caused what, our sessions had no blame at all and also no learning. We got no better at avoiding the thing that kept killing us because we could not identify it.
The relationship with penalty size
Attribution matters more as the penalty grows, which sounds obvious and produced the log's most useful practical finding.
Cold Store attributes incorrectly and caused almost no friction, because being knocked over by a crate costs three seconds and looks funny. Fen Crossing attributes incorrectly and ended a partnership in twelve minutes, because the same error cost an entire traversal sequence.
| Penalty size | Games | Friction observed |
|---|---|---|
| Trivial — seconds | 3 | None in any session |
| Moderate — a task lost | 4 | Behaviour changed in 3 of 4 |
| Severe — a run lost | 4 | Behaviour changed in 4 of 4 |
So a game can attribute friendly fire wrongly and get away with it entirely, provided being wronged costs almost nothing. Several of these designs appear to have arrived at that intuitively, and the ones that combine bad attribution with severe penalties are the ones that lost us fastest.
What would reverse this log
A third regular player, played consistently across several of these games. Two people fall into a fixed pattern — one of us is naturally the careful one and one is naturally the fast one — and a great deal of what is written above could be a description of us rather than of these systems.
If a third player joined and the wrongly-faulted role moved with the personality rather than with the mechanic, the finding here is about people. If it stayed with the mechanic, the finding holds. We do not currently know, and this paragraph exists so that nobody reads the table above as settled.
Current state of the log
Eleven games with friendly fire. Eight attribute the event to the player who was hit, one to the player who caused it, two to nobody. In all eight cases of misattribution the wrongly-faulted player reduced their own contribution, usually without discussing it. Severity of penalty determines whether misattribution matters: trivial penalties produced no friction in any session, severe ones produced friction in every session.