Coupled Autonomy (Principal-Agent Problems IV.)

Delegation and Dependence for Optimised Flex-Assets

This post continues an earlier series (Principal–Agent Problems), exploring how individuality and interdependence collide in asset optimisation. Comments, counterexamples, and critiques are very welcome. The aim is not to settle the question, but to trace some contours. If you disagree let me know.

A few weeks ago I asked a simple question under an optimiser’s public performance post. (You’ll find it if you want to.) This led to a short exchange that, in hindsight, exposes a deeper tension running through the energy transition.

Everyone in this ecosystem wants to be special.
Asset owners want their megawatts to stand out.
Optimisers want to prove their algorithms are unique.
Traders want to show they can outperform the market.

But the system that connects them—the markets, the forecasts, the control signals—binds them together far more tightly than most of us like to admit.

We talk about “individual optimisation,” yet the whole machine only works because of the connections.

No one builds these assets for charity. Commercial viability matters. But if renewable systems are to work at scale, the illusion of strict independence will have to give way to an honest focus on interdependence.

This post explores the wider tension between independence and interdependence. It is not specific to any particular market participant.

---

The optimiser’s post showed impressive BESS performance for their best asset.

I asked whether this optimisation was pool-based or asset-specific.
The answer: each asset is optimised individually.

That sounds straightforward.

Every battery has its own constraints—duration, efficiency, warranty, ramp limits, grid connection. Treating them individually feels both technically necessary and commercially fair.

But look a little closer, and you start to see something else.
Even “individual” optimisation lives within a web of implicit coupling: technical, informational, contractual, and layered through multiple levels of delegation.