How a connection point reaches the overdue list
Every megawatt of permitted battery capacity in all nineteen zones has been allocated. There is no queue to join, so the only way in is a position somebody gives back. The question worth answering is where somebody else’s position has already failed to arrive on time. This page is how that is read, and where it is weak.
What the list is
Every project holding firm gate 2 capacity whose contracted connection date has passed, grouped by the connection point it is contracted to, ordered by megawatts. Each row carries the capacity, the date the register published, and how long ago that date went by. Nothing on it has been checked against a landowner, a planning authority or a network study.
Why there is no longer a score
This list used to be forty substations ranked on a model worth one hundred points across four questions. Measured across those forty, seventy eight per cent of the marks actually awarded went to two of the questions: how much capacity was queued at the point, and how many applicants held it. Neither is evidence that anything is coming back. The two questions that measured whether a position was in trouble carried twenty two per cent between them, and twenty two of the forty listed points scored nothing on either. The point ranked first scored nil on the question asking whether a position was already late. A model that ranks size while claiming to rank fragility is worse than no model, because it looks like an answer.
The list also read a block of scored substations built outside the build and never rebuilt. It named fourteen overdue points where the live register rows give twenty. It now reads the project rows the capture carries forward twice a week, so it moves when the register moves.
Why a date is better evidence than a score
A connection date is published by the system operator, agreed with the applicant, and either in the past or not. It needs no weighting and it cannot be argued with. It is the only signal in this tool that a reader does not have to trust us to produce.
The second table is not the same kind of thing
Beneath the overdue list is a delivery reading across the firm gate 2 book, banded severe, high, moderate and low. That is an estimate. It reads the consent stage a project has reached against the years remaining before its contracted date. Nothing in it has failed and nothing in it is late. It is shown separately, and labelled, so the two are never read as one list.
What is wrong with it
A late connection date is evidence of lateness and not of abandonment. A developer may be late deliberately while it waits for something else, and a date can move by agreement without anything having gone wrong. The register publishes the date but not the reason. Eighteen of the twenty points currently on the list are in Scotland and most are small grid supply point connections, so the list says more about where small Scottish projects slip than about the market as a whole. And nothing here has watched a single position actually fail, because the change record only began on 20 August 2026. Ask again in a year, when the ledger can say which of these came free.