Cobo Makes Games blog

What one co-op playtest changed in our game

One two-player playtest of Mytheria, the notes it produced, how we sorted them, and the changes they led to, from loot pickup to the world map.

Mytheria's world map: dragging it across the Woodland and pinning a waypoint with a right-click
On this page
  1. How do you run a useful co-op playtest?
  2. The short answer
  3. Write it all down, and decide later
  4. Split bugs from decisions
  5. Sort by what breaks play first
  6. Treat numbers as guesses
  7. Prove every fix
  8. Make the map a tool
  9. A checklist for a co-op playtest

Testing a co-op game alone misses many of its problems. A rejoin that loses your bag, or a dawn that arrives on one screen before the other, only shows up with two people on two computers. One long two-player session of Mytheria, our pixel-art co-op game, produced about fifty notes. This is how we sorted them and what they changed.

How do you run a useful co-op playtest?

Play the real build on separate computers, the way players will, for long enough to reach the second and third day. Write every note down the moment it happens, in plain words, without deciding anything. Afterwards, split the notes into bugs to fix and decisions to make, sort the bugs by how badly they break play, and fix each one with a check that fails on the old version.

The short answer

What we do with a playtest:

  • Write each note down as it happens, exactly as it felt, typos and all.
  • During the session, play and take notes without fixing anything.
  • Split bugs from decisions: a lost bag is a bug, and "make it grindier" is a decision.
  • Fix what breaks play first, then balance, then the map, then polish.
  • Treat the numbers in the notes as guesses to try in play.
  • Give every fix a check that fails on the old code.
  • Replay the whole session on every new build, so the same bugs can't come back.

Write it all down, and decide later

During the session the only job is to play and write. A note like "the new day doesn't show up for everyone at the same time" is worth more than a fix made on the spot, because the fix can wait, while the exact feeling is hard to recall later. The notes stay word for word in their own file; the discussion goes in a separate one.

Split bugs from decisions

About half the notes were plain bugs, like a guest losing their items on rejoining or a weapon not firing in the home meadow. The other half were wishes and questions, like harder bosses or a grindier pace. Bugs go straight on the fix list, while decisions get written up as clear questions with a recommendation, and nothing that depends on them starts until they're answered.

Sort by what breaks play first

These are twelve of the notes and what each one became. Read them in the order they were written, then sort them into the order we worked through them:

Sort the notes, then open one to see what changed

Twelve of the notes, shortened, with what each became in the game.

The order matters because the notes depend on each other: nobody can judge the balance if a rejoin wipes their bag, or judge a map they can't drag. So the work went from what breaks play to progression and balance, then the maps, then bosses and danger zones, and finally world polish.

Treat numbers as guesses

The notes come with numbers: "half a second", "like five logs", "maybe 50 % longer". We wrote proposals next to them, a pickup delay of 0.1 seconds, a channel of 3.75 seconds, a quarter second of dash protection, and then tried each in play. Some stayed as proposed and others changed: the dash protection ended at 0.2 seconds, and the recharge at 0.76 instead of 0.9. A note tells you where the problem is, and play tells you what the number should be. Loot pickup itself is in loot drops that feel good to pick up, and the Waystones in fast travel design.

Prove every fix

Each bug got a check that fails on the old code and passes on the new: the game file set back to the old version for one run proves the check would have caught it. Then the whole session became a repeatable run on two real builds of the game, doing what the playtest did: building, travelling, mining, quitting mid-night and rejoining. Its final run passed all 50 of its checks. The same list-and-rank habit runs our visual polish too: see how we decide what to polish first.

Make the map a tool

The note that changed the most visible part of the game was about the map. A map you can only zoom is little more than a picture, while a map you can drag, read and mark is something you plan with. Now you drag it, zoom where you point, and pin waypoints that your partner sees too. The clip at the top of this post is that map in use.

A checklist for a co-op playtest

  • Play the real build on separate computers.
  • Play long enough to reach the second and third day.
  • Write every note as it happens, and decide later.
  • Split bugs from decisions.
  • Fix what breaks play first.
  • Treat every number in the notes as a guess to test.
  • Give every fix a check that fails on the old code.
  • Make the session something you can run again on every build.

Further reading

Wishlist Mytheria

Mytheria is a pixel-art co-op adventure for one to four players, coming to Steam for Windows, macOS and Steam Deck. A wishlist is the best way to follow it and to help us.

See the game Join the Discord

Mytheria

Coming to Steam

The store page isn't up yet. Leave your email and we'll tell you when you can wishlist Mytheria on Steam.

We'll send that one email and never share your address.

Windows · macOS · Steam Deck