Quality, not a quick win

Barrelhouse started as a quick win. On August 22 I changed the frame and wrote it at the top of the roadmap. This is going to be a quality game. I still plan to ship it much sooner than The Lien, my larger game, and I do not want to miss an opportunity for the sake of shipping quicker. Part of why it needed saying is on me, because when the project started I was the one asking for a quick win.

Why it went at the top

It went above everything else in the roadmap, and not into a decision log, because every document in the repo had been written under the old frame, and a quick win frame never announces itself. It shows up as an idea parked with a plausible reason, again and again, until the reason nobody wrote down is that it would take longer.

The notes held a worked example. Three times I had suggested animatronic figures, meaning each stage figure cut into a few pieces that move, instead of frames of animation. The idea was deferred three times and never weighed on its merits. Its recorded trigger was "after the audio round", and that trigger had quietly expired once the pilot track had a measured tempo for the motion to follow.

On its merits the idea is good, for a reason specific to this pipeline. Frame animation means generating several sprites per member, and nothing structural holds a generated set together beyond a fixed seed and an identical prompt prefix. Drift between two frames of one animation is far more visible than drift between two portraits. Moving slices of art that already exists and has already passed a smoke test carries no generation risk at all. The real cost is slicing twelve figures by hand, and whether a sliced pixel figure holds together at 3x can only be answered by looking at one. So the next step is a spike on one figure with four cuts, and animation is now my named top priority.

What changes in practice

Three rules came with the reset.

"That is more work" stops being a reason on its own. It was never written down as one. It worked as an unspoken tiebreak.

An idea gets deferred with a named trigger or it gets scheduled. "Later" is not a status. Every raised and unshipped idea now lives in a standing backlog at the top of the roadmap, with its status checked against the code rather than against anyone's memory, and a round does not get to skip that table.

Shipping sooner than The Lien is still the goal. The roadmap says this is not a license to gold plate. It is a license to take the better option when there is one, and to stop treating the fast option as the default.

What the same day built

Three rounds of refining the show's moments had come back boring on smoke reads, and on August 22 the protagonist round went looking at a question nobody had checked. Was choosing between band members a decision? It was not. A beat's outcome depended only on the member's instrument, so which member answered it made no difference beyond that.

That round put the player on stage from the first second as one of the twelve members, and the old empty "no band yet" state was deleted. Each member gained two authored stats, stamina and presence, set so the roster trades instead of ranking. Two guards in the suite refuse a roster where one member is best at everything, or where no pair trades. Both ideas sit in the new backlog marked shipped.

The same day the show became a song, which A show is a song covers. The backlog also picked up two rows that measurement forced. At the song's real number of turns the beat deck ran out within a single show, and nothing in a show was ever unaffordable, because stamina pools had been sized for a per second drain that no longer existed.

None of that is animation yet. The difference is that the question of how the figures move now gets decided on its merits, and not on how long it takes.