Model: no estimates, estimates, beyond estimates

December 10th, 2020

In Model: 3 stages of relational dependencethere are 3 stages of dependency.

  1. Dependent

  2. Independent

  3. Interdependent

The first stage and the third are often confused

It's even said that people going from independent to Interdependent "fear they're going backwards".

While this is only natural progression.



Model: Agile Fluency with relation to estimates

  1. Lvl1 estimates not possible

  2. Lvl2 estimates possible / easy

  3. Lvl3 estimates not valuable, just work on the next most valuable thing (team knows what that is, because they have the market knowledge)

At fluency 1 there is not enough stability in the system to make estimates. Hidden technical debt can pop up when working, delaying otherwise simple things.


At fluency 2 there is wel maintained code. Little or no hidden bugs or nasty surprises. Previous work is representative of future. So it's possible to do pretty accurate estimations.


At fluency 3 the team is always working on the next, most valuable thing. They know what that is, at any time, because that knowledge and expertise is in the team. (and if they don't know what's the next most valuable thing, finding out what is, is the next most valuable thing) Estimates are still possible, assuming fluency 2 is faster than fluency 3. But the time spent on making estimates, is more seen as wasteful, because they could have started on the next most valuable thing.

I'm not entirely sure where to put a team that has fluency 3 without fluency 2. But they are described as more of a temporary thing anyway, in the model. But if I'd had to guess, they're a combination of 'it's not possible' + 'it's not valuable'.



Like relational dependence and interdependence  are often confused. Maybe the NoEstimates movement has the same confusion?

I've heard arguments like

  1. It's not possible to estimate

  2. But we have managed to in this one team

  3. There is no value in doing it, it's just a waste of time!

Before, I classified those arguments in 2 camps:

  • Don't do estimates

  • Do estimates

But maybe it's more useful to look at them as a maturity model of 3 steps. (like the relational dependence model is) One where we know the first and third steps are lookalikes, even though very different.

This post was referenced in: