Ideate

Settle on a solution in a meeting and the candidate count never passes two or three. Once someone puts an idea forward, the next person improves it rather than opening a new direction, and half an hour later the conclusion is a polished version of whatever came up first.

Pagedeck had that meeting. The problem was that people who joined recently could not catch up on past decisions, and within twenty minutes the room had settled on building a change-history screen. Two candidates had been raised. Two months later, checking with users, almost nobody opened that screen. What was needed was not a history but a single line saying which version is current.

With only two candidates there is no way to know whether the chosen one is right. This stage exists to hold judgment back and fill the count first.

Nobody rules on good or bad; the only goal is raising the number of candidates. Three rules hold it together.

1. Keep judgment out. One "can we even build that?" and the pile stops before it reaches ten. Feasibility belongs to prototyping, the stage of building something quickly, so ideas that make no sense go on the wall as they are.
2. Fill the count first. Aim for around forty. The first ten are what anyone thinks of; genuinely different directions tend to arrive after the twenty-fifth.
3. Build on other people's ideas. Rather than inventing a new direction every time, twist the idea next to yours into a variant. That is where the count picks up speed.

Starting out loud tilts the room toward whoever speaks first, so the usual order is eight minutes of everyone sketching on paper, then laying the sheets out one at a time. Five people with eight sketches each makes forty.

The input is the single sentence produced by define, the stage before this one. A broad sentence scatters the candidates. The output is not all forty but three directions that differ from each other. Pick three similar ideas and the next stage tests the same thing three times.

"We're still ideating, so put up the ones that can't be built too."

Feasibility judgments are not welcome in this stage.

"There are six candidates and you're already choosing."

Judgment arrived before the count was filled.

"All three are the same direction. Let's add another branch."

The candidates going into the next stage have to differ from each other.

At Pagedeck, Noa runs the session. She pins the problem sentence to the wall and starts an eight-minute timer, and she invites engineering and sales as well, because a room needs people who propose things that can't be built for the directions to spread. Theo's ideas cluster around what is easy to implement and Sofia's sit close to what customers asked for in words. Putting both up is what produces a candidate that isn't the average of the two.