Main idea

A Now / Next / Later roadmap is worth what it turns down: the Now is bounded by the team's real capacity, the Next stays short, and even the Later — supposed to represent the field of possibilities — stops being useful if it holds fifty directions. That work of restriction is immediately undone if everything the roadmap sets aside falls into a queue that has no edge.

The rejected topic then hasn't been rejected: it has changed column. The organization has produced the gesture of arbitration without producing its effect, and it keeps the conviction of having chosen.

Why it matters

This says where to look to know whether prioritization actually happened. Not in the roadmap, which is always short by construction, but in what surrounds it: a tight plan backed by an unlimited queue describes the same load as an unlimited plan.

It also gives a simple design rule: the constraint has to propagate downstream, otherwise it only applies to the display.

Nuances and limits

The limit doesn't have to be a number of cards. A maximum lifespan, an exit condition, a review that removes as much as it adds all produce the same effect without an arbitrary quota.

And a wide queue stays workable as long as it is neither reread, nor prioritized, nor promised — but it is then an archive, and that has to be owned as such.

Open questions

• Which form of limit best withstands the pressure of incoming requests: a number, a duration, or an exit condition?