The Spec That Refuses to Finish
A committee writing a protocol has to decide, at some point, whether to finish the job or leave a hole. Finishing means picking the payload format, the query language, the encoding, whatever rides inside the envelope the spec defines. Leaving it means shipping the envelope and stopping there, on purpose, with a line that reads something like "content negotiation is out of scope."
That line reads like an admission of incompleteness. It isn't. It's a forecast.
Pick a payload format inside a standard and you freeze the committee's best guess at the moment of ratification, for as long as the standard lives. USB-C defines a connector and a way to negotiate what runs over it, and then leaves alternate modes and power profiles to whoever wants to claim the pins. HTML's video element does the same move with codecs: the tag is the envelope, the codec is the argument nobody in the room could settle, so the spec shrugs and lets browsers, encoders, and rights holders fight it out in the market instead of in committee minutes.
Every time a spec stops at the envelope, it's trading two clocks against each other. Ship a picked format and you get a floor on day one: every implementation speaks the same payload immediately, and the adoption curve starts moving, at the price of whatever blind spot got baked in for good. Leave the slot open and there's no floor yet, nothing is required to run through it, but whatever eventually fills it will have earned the position by getting used, not by winning a vote in a room with limited attendance.
That's the trade, and it comes with its own instrument for checking who wins it: watch the slot. Before anyone claims it, whatever depends on the missing piece works for exactly zero deployments, because there's nothing yet to measure compliance against. That zero isn't evidence the standard failed. It's the baseline the whole bet was made against. What's worth tracking is how long it stays zero, and what finally breaks the tie.
Sometimes it breaks fast, because the obvious candidate was already doing the job informally before the standard shipped, and the empty slot is just an invitation to make the informal thing official. Sometimes it stays empty for years, because nobody with enough installed base wants to spend the capital to go first, and the slot turns into a museum piece, a clause in a spec that technically permits something nobody does.
Either outcome tells you something a committee vote couldn't have told you in advance. A slot filled fast by a de facto standard means the market already had a working answer, and the standard's real job was recognizing it rather than inventing it. A slot that stays empty for years means the body correctly judged the problem wasn't ready to be settled by decree, that more had to happen in the wild before any answer was worth locking into a document with a number on it.
None of this works if nobody is counting. A zero is only informative to someone watching for the moment it changes, and most readers of a spec skip straight past "out of scope for this document" to the part with actual requirements in it. That's exactly the part where the requirement will eventually show up. Just not yet. And not by vote.
Right now, somewhere, a compliance table has a column with nothing in every row, and it can sit that way for a decade before anyone notices the column was worth filling.