The Changing Vocabulary of "M&V"

jskromer
Oct 10, 2025
2 min read

I asked ChatGPT "how are observability engineering and digital twins related?"

They’re deeply related—almost symbiotic—but operate at different layers of the same reality.

1. Observability Engineering: Real-Time Insight

Observability engineering is about making complex systems measurable, understandable, and debuggable in real time.

It involves:

Instrumentation: embedding telemetry (metrics, logs, traces) into systems.
Data pipelines: aggregating and correlating data across distributed components.
Analysis and visualization: detecting anomalies, performance drift, or causal relationships.

In essence, observability is how a system reveals what it’s doing.

2. Digital Twins: Virtual Representation

A digital twin is a dynamic, data-driven model of a physical system—e.g., a building, turbine, or grid node—that evolves continuously based on incoming data.

Its power lies in simulation and prediction:

Comparing real performance vs. simulated baseline (counterfactual).
Testing “what-if” scenarios.
Quantifying uncertainty and forecasting outcomes.

In essence, a digital twin is how we reason about what the system means.

3. The Connection: Observability Feeds the Twin

Observability engineering provides the sensory nervous system for the digital twin:

Telemetry from the real asset populates the twin’s state variables.
Observability tools maintain synchronization between measured and modeled behavior.
Trace and metric data support model calibration and anomaly detection.

Conversely, the digital twin gives observability context:

Predictive models flag which signals matter.
Simulation results inform alert thresholds and root-cause analysis.
Historical counterfactuals define what “normal” or “optimal” should look like.

4. Practical Example

In a smart building:

Observability engineering captures HVAC sensor streams, occupancy logs, and equipment status.
A digital twin (e.g., EnergyPlus or IES or even Modelica) uses that data to simulate expected energy use.
The twin detects deviations—say, a 10% rise in fan energy—and observability tools trace it to a failed VFD sensor or control sequence.

5. Convergence: From Monitoring to Decision Support

The frontier is autonomous observability-driven twins:

Continuous learning from streaming data.
Bayesian or ML-based calibration to maintain fidelity.
Closed-loop control: the twin suggests or executes corrective actions.

What does this mean for "M&V"?

This convergence is foundational to self-managing infrastructure, AI-assisted M&V, and counterfactual verification—areas directly relevant to your “Counterfactual Designs” framework.

Okay - back to a human. What this says to me is that us "M&V" folks have to keep scannign the landscape for tools that will enhance our work - bringing confidence to the results of energy and carbon management.