How This Is Collected, and What It Cannot Tell You

Collection

Once a week we fetch each ISO's own published interconnection queue report over plain public HTTP, store the response bytes with their SHA-256, parse one row per project, and seal it under that week's snapshot date. Sealed rows are never edited. We have 39 sealed snapshots, from 2026-06-11 to 2026-09-07.

Why the history is the point

Every one of these reports is free to download today. None of the ISOs publish an archive. Each new report replaces the last. So the current queue is a commodity and the record of how it changed is not, because after the fact it cannot be reconstructed by anybody, including us. That is the only thing on this site you cannot get from the source.

How we compare in-service dates

The proposed commercial-operation date arrives in five different formats across the queues we read (2030-01-02, 12/31/2028, 3/1/2028, 06-2028 and 11/2027), and some rows carry a word where a date should be, such as I/S or N/A. We parse it only to answer whether the date moved later or earlier, and we apply four rules:

• A value we cannot parse is reported as changed with both published values shown and no direction. Sorting I/S against a real date would invent a slip.
• A day count is shown only when both sides carry a day. 06-2028 is ordered as June but never differenced, because the ISO published a month rather than a day.
• A date published for the first time, or withdrawn, is counted separately and not as movement. There is no earlier plan for a first publication to have moved from.
• The same date restated in a different format is not a move. A raw text comparison would call it one.

Not every ISO publishes this field. Where a queue publishes it on few rows or none, the changes page says so beside the count, because a zero there would otherwise read as a queue whose schedules all held.

What we do not do

• We do not normalize status. Every ISO uses its own vocabulary, and ISO-NE publishes no status column at all in its report. Any shared 'active / withdrawn' scale would be our interpretation wearing the ISO's authority.
• We do not fill in blanks. Null capacity, absent county, missing commercial-operation date all render as not published. Every capacity total on this site shows how many projects it was computed from.
• We do not call a dropped row a withdrawal. We report that it is no longer listed.
• We do not treat our collection date as the data's date. Each page shows when the source file itself last changed, which for some ISOs is much older than our last read.

The one mapping we do make

The national technology table on the hub, and the per-technology state pages, group the ISOs' fuel codes into shared buckets. This is the complete mapping. Any code not listed here (including a blank fuel field) is counted as unclassified and is never assigned to a bucket on a guess.

Where each number came from