Useful positions, kept short.

Observations on the moments when technology choices stop being experiments and start becoming commitments.

Note 01
4 min read

The pilot is not the decision.

A pilot can prove that something works. It cannot decide by itself whether the organisation should depend on it.

Teams often treat a live system as a successful system. The change sounds small, but it collapses three separate questions: can it operate, does it create enough value and should the organisation accept the obligations that follow?

A disciplined pilot fixes the decision before the test. It names the value threshold, evidence required, unacceptable failure and accountable owner. Expansion is then a separate act—not the default reward for reaching the end of the trial.

Position: write the release decision before the pilot can inherit it.

Note 02
3 min read

Evidence before architecture.

The faster a team can build, the easier it becomes to answer the wrong question with impressive precision.

Architecture creates gravity. Once data pipelines, integrations and operating routines exist, the cost of changing the original premise rises. Technical progress can then become a reason to protect the decision it was meant to test.

The countermeasure is not slower engineering. It is a written evidence boundary before engineering begins: what would justify the next commitment, what would contradict it and what does not count as evidence at all.

Position: do not let implementation quality stand in for decision quality.

Note 03
4 min read

When a reversible choice becomes permanent.

Most lock-in is not announced. It accumulates through interfaces, operating habits, incentives and missing exit work.

A choice can remain contractually reversible while becoming operationally impossible to change. The warning signs are familiar: data is difficult to move, one supplier owns the control surface and internal expertise develops around a single path.

Reversibility must be designed and reviewed, not assumed. Name the last credible exit point, keep the cost visible and decide deliberately which dependencies are worth making permanent.

Position: optionality disappears unless someone owns it.

A real decision is more useful than a general brief.

Bring the decision, its deadline and its downside. Norelmarch will tell you whether a review is the right next move.

Open the private brief