What a signature line tells you that a party roster doesn’t

Every contested proceeding hands you a party list. It is the first thing you get, it is free, and most of the analysis that follows gets built on top of it. Here is the problem. A party list tells you who showed up. It does not tell you what anybody did.

I ran our research system against a closed Virginia rate case to find out how much that costs. Case PUR-2023-00101, the 2023 biennial review, final order February 2024. No client, no engagement, nothing riding on the outcome. Just a dense public record and one question: how much of what you would actually want to know is in the roster, and how much of it only exists in the filings?

Almost none of it was in the roster.

The settlement recorded three positions, not two

Ten parties signed the stipulation. Four more filed to say they do not join it and do not oppose it. At least one party carried on the Commission’s own docket appears in neither list. 1

That sentence is the whole game. A support-or-oppose model has to put those four parties somewhere. Group them with the signers and you have overstated your coalition to your own client. Group them with the opposition and you have invented an adversary and briefed against a party that was never against you. The parties told the Commission exactly where they stood, in writing, and a binary model throws the sentence away.

Non-opposition is a stated position. It is not a weaker kind of support and it is not a politer kind of opposition.

Two of the largest hyperscalers landed on opposite sides

Google signed the stipulation. Microsoft did not join it. Both companies are named on page three, in adjacent sentences, in different lists. 1

Anyone briefing this proceeding as what the data centers wanted got it wrong, and got it wrong in a way that stays invisible until somebody opens the signature paragraph. No press release carries this. No case summary carries it. It is a name in a list on one page of a fifty-nine page filing, and it reverses the read on the single largest interest group in the room.

The coalitions were not where a reasonable person would put them

• The Attorney General’s Division of Consumer Counsel signed the same stipulation as the utility and the Data Center Coalition. 1
• Appalachian Voices, an environmental organization, signed it as well, alongside both of them. 1
• The Virginia Poverty Law Center did not join the stipulation the Attorney General signed. 1

All three of those come out backwards if you reason from what a party is instead of from what it filed. The third one is the expensive one. Two organizations that any briefing would collapse into the consumer side sat on opposite sides of a signature line, in the same document, on the same day.

This is now the second state where I have watched the consumer advocate back the outcome the utility wanted. In Ohio, the Office of the Consumers’ Counsel signed the stipulation the Commission adopted in its order of July 9, 2025. 2 Twice is not a rule. It is enough to stop treating the opposite as an assumption you can build a strategy on.

The claim of ours that got struck

Here is the one I would rather not print.

The Commission’s participant index shows the two industrial user committees on this docket sharing three of the same individual counsel. That is real, it is in the Commission’s own index, and it is verifiable. 3 From it I formed a working hypothesis: parties sharing counsel are aligned, so shared representation can be used to infer a coalition where the filings are thin. That is a useful shortcut. It would let you map a fight faster than anybody reading documents.

The same two documents killed it the same day. One of those committees signed the stipulation. The other appears nowhere in its fifty-nine pages. Same three lawyers, opposite outcomes. 1,3

The hypothesis is now a row in our ledger with a verdict of “fails” and the reason recorded underneath it. We kept the row instead of deleting it, and that is deliberate. A struck row is the only evidence that the check ever ran.

A hypothesis that dies before publication is the method working. The same hypothesis dying after publication is a correction, and you do not get to choose which one it is.

What a roster actually is

A party list is an administrative artifact. In this system it functions close to a service list: it tells you who is entitled to receive documents. That is a real thing to know and it is not a position. A signature is a different object entirely. It is an act, taken on a date, recorded in a filing, by a named party that could have done otherwise.

Every finding above came out of signature paragraphs and filing text. Not one of them came out of the roster.

There is a smaller finding sitting underneath all of it. The Commission truncates its own party name field at about forty-two characters. Three of the four truncated names on this docket can be completed from document titles in the same source. The fourth cannot. The obvious human completion is right there, and the record does not support it, so it stays flagged and incomplete. 3

What it costs to get this wrong

On a closed case, nothing. Nobody is on the other side of this piece.

In a live fight it costs you the argument. You put a coalition map in a deck, opposing counsel opens the same public docket, and now the room is not discussing your issue any more. It is discussing whether you read the file. That is very hard to come back from in front of a client, and it happens on facts that were never load-bearing to the strategy in the first place.

The strategy in these fights is usually sound. It is the facts underneath it that fail.