Why Software Delivery Is More Quantum Than You Think
In quantum mechanics, we can’t predict exactly where a particle will be. We describe a range of possible outcomes, each with a probability. The system exists in a set of possibilities until a measurement is made. Only then does a single outcome emerge.
This might seem far removed from software delivery. But the underlying idea is surprisingly relevant.
In complex systems, outcomes are not deterministic. They are probabilistic.
Traditional planning models assume the opposite. If we define scope upfront, map dependencies, and plan carefully enough, we should be able to predict what will be delivered. PI Planning, for example, operates on this assumption. Work is defined in advance, commitments are made, and a plan is constructed for the next 8–12 weeks.
In this sense, PI Planning resembles classical mechanics. If you know the initial conditions and the forces acting on a system, you can predict its future state. The system is assumed to be stable and predictable.
Like the early Bohr model of the atom, this provides a simplified view that can be useful for establishing structure and alignment. But it assumes a level of stability that rarely exists in real-world delivery systems.
Because in practice, delivery systems are not stable. Requirements evolve. dependencies shift. Work takes longer—or shorter—than expected. New information emerges continuously.
The system changes, but the plan does not.
Certainty at the start is an illusion.
What follows is predictable. As reality diverges from the plan, coordination increases. Teams re-sequence work. Dependencies are managed. Alignment must be continuously re-established. The system adapts, but only through increasing effort.
The problem is not poor planning.
The problem is applying deterministic thinking to a probabilistic system.
Quantum mechanics forced physicists to abandon deterministic models in favor of probability distributions. Modern forecasting applies the same shift to complex systems.
Instead of asking:
- What will we deliver?
We ask:
- What is the system likely to deliver?
This is not a weaker form of prediction. It is a more accurate one.
In a flow-based system, delivery performance can be observed and measured. Throughput tells us how much work is completed over time. Cycle time tells us how long work takes once started. Variability tells us how consistent that performance is.
Together, these form a probability distribution.
Instead of a single predicted outcome, we get a range:
- There is a 50% chance we deliver X features
- There is an 85% chance we deliver at least Y features
This reflects how the system actually behaves.
Before work is completed, multiple outcomes are possible. Features may complete early. They may take longer than expected. Some may be descoped or replaced. Only when the work is done does a single outcome emerge.
Delivery collapses possibility into reality.
This is where traditional planning creates tension. PI Planning frames delivery as a commitment. Teams agree upfront to deliver a defined scope over a fixed period, based on assumptions that are often incomplete.
As delivery progresses, new information emerges and plans change. The system adapts, but the commitment does not.
A flow-based model approaches this differently. Delivery is treated as a forecast, not a commitment. Forecasts are based on observed system performance and updated continuously as conditions change.
The conversation shifts.
From:
- Did we deliver what we committed?
To:
- What is the system likely to deliver, and how is that changing?
Commitments assume certainty. Forecasts reflect reality.
This is the real shift. We are not abandoning predictability. We are redefining it.
While classical physics predicts exact futures, quantum mechanics dictates only probable outcomes. Perhaps I’m stretching the quantum analogy a bit too far, but I think it’s fair to say that in complex systems, certainty is an illusion.