Current Edge

Silence in notices written for another purpose, six months after the first go-lives, is not evidence of failure.

The resignation gets the attention; the assurance is what gets written down.

A span is a claim about what a manager no longer has to do.

The hub is visible because it is easy to draw; the return is invisible because nobody has agreed how to draw it.

It is therefore a reliable guide to what a programme costs to get wrong, and an unreliable guide to what it delivers when it goes right.

A claim that cannot be observed is hard to refute, and equally hard to rely on.

A buyback is the one use of cash a board can switch off without anyone filing a complaint.

A percentage is only as comparable as the question it answers.

A benefit total is a promise made at approval, and a promise is only as good as the follow-up.

A commitment is strongest when its owner can say which function, which workload and which automation rate it rests on.

The part of the case the committee approved is the part the filings are least likely to expose.

Solicited is not negotiated, and negotiated is not published.

A saving exists only when an overlapping tool is switched off, with a date, an owner and a contract end.

It is a hypothesis that has not been tested on the people it is meant to protect.

A rising AI share inside a falling total says that AI has become the preferred explanation for a smaller volume of cuts.

A buyer that treats both as the same backlog will negotiate the wrong protections.

The structure is a delivery mechanism for whatever advantage the units already have.

The slogan sorted them wrongly.

The sensitivity sits in the second contract, which has not been priced yet.

The title is a unit of labelling; the mandate is a unit of authority.

A baseline cannot be taken retrospectively once the old process is gone.

It has centralised the sentence about AI.

If the answer is yes, the programme is exposed to both risks with cover for one.

A saving that lands nowhere in particular will land nowhere.

An outcome price with no guarantee, no refund for disputed counts and no cap is usage pricing with better marketing.

Running the service and delivering its benefits are different accountabilities, and the first is routinely mistaken for the second.

The question was never build or buy in general.

The lab race supplies the alternatives; architecture decides whether a buyer can use them.

Neither fact excuses a redesign that removes the practice junior work used to provide.

A pilot result at the first firm is evidence of a pilot.

A portfolio is real on the day its executive takes a decision that a programme manager could not have taken, and survives it.

Approval then becomes a judgement on management's credibility rather than on evidence.

Labour binds last, after the other gates have opened, and only for sponsors who have not designed it down.

Risk moved, and so did the boundary of what counted as delivered.

Both readings are accurate, and the distance between them is the finding.

The distance between discussion and transition is the empirical core of this paper.

So the typical manager has not been handed twice the team.

The pipeline problem is not a vacancy to be filled but a capability to be built — and rebuilt continuously, because the forces eroding it are structural rather than cyclical.