Everyone needs a rating before they can get one

The trouble is what happens at the very bottom. To get a DUPR that means anything, you need a stack of logged matches against a variety of already-rated players — a dozen or two before the number is considered reliable. But a new player in a local rec league has none of that. They're playing the same handful of people, nothing's getting logged, and so the system just... doesn't have them. To get rated you need organized, recorded play; to get into organized play fairly, you more or less need a rating. It's a small chicken-and-egg, and a lot of people are stuck inside it. The ladder is real and it works — it just has no easy first rung.

The people running the league are stuck on that same rung from the other side. They've got to build teams that are fair and brackets that aren't lopsided, and they're doing it with almost no data — eyeballing skill, trusting self-ratings, running an impromptu skills day to guess where everyone belongs. It's the same missing piece the players feel, just from the organizer's chair. Nobody's being unreasonable; the tool for that first read simply doesn't really exist.

That missing first rung is what I've been poking at with VolleyForge. Not a rival rating — a layer underneath the one we already have. The idea is simple, even if the building isn't: give a new player a low-friction way to get an honest first read on where they stand, and give a small league a way to produce that read out of the games they're already playing. And because DUPR is the standard, the point isn't to wall that data off — it's to let clean, grassroots results flow up into the rating everyone already uses. A bottom rung that feeds the ladder, not another ladder.

The mechanism I keep coming back to is boring on purpose: the read on skill shouldn't be a thing you go and collect, it should be a byproduct of running the league at all. So the data model underneath VolleyForge is just the shape of a season — a league has seasons, a season has teams and players, and every night of play is a set of matches, each split into the individual games that decided it. The standings are never a number anyone types in. They're derived: computed from the matches actually played, at the moment you ask for them, and stored nowhere. If two teams played, the record exists; if they didn't, there is nothing to invent.

League → Season → Team, Player, Match → MatchGame
# standings are computed from Matches at read time — never stored

The scheduling works from the same instinct. A new league doesn't wait for me to hand-draw a bracket; it generates a round-robin by the circle method, rotating the field so everyone plays everyone and an odd roster draws a bye. The seeded demo is exactly that — eight teams, sixteen players, twenty-eight matches — a full round-robin that exists only because the generator drew it. Run enough honest nights like that and a picture of relative skill ends up sitting there in the results, waiting to be read, without anyone having stopped to grade a single person.

I'll be honest that the interesting part, to me, isn't a cleverer algorithm. It's the plumbing — running grassroots leagues well enough that a trustworthy read on skill just falls out of the play as a byproduct. That's the part nobody's really taken, probably because it's unglamorous. It's the same thing I keep finding myself drawn to elsewhere: the quiet, structural work underneath the thing everyone actually looks at.

It's early — more an attempt than a finished thing, and I want to be clear about that. I don't know yet whether it becomes a real product or stays a sharp idea I chased for a while. What I do know is that the gap is real, because I stood in it: a court full of people with obvious, varied skill, and a system that couldn't see almost any of it.

There's a reason I'm allergic to a rating that rides on a platform that lies, and it's that the first version of this one lied constantly. It came off a no-code builder, and under the nice screens almost nothing was real: testimonials from people who never existed, a cheerful 'League Created!' banner that created no league, an admin panel you could unlock by editing a value in your own browser. Every button fired a success message and persisted nothing. The whole thing was a demo wearing the costume of a product.

So the rebuild had one rule I wouldn't bend: a success message has to be earned. Every write now runs through a single path that only celebrates once the server actually confirms it — it waits for a real ok, refetches, and shows you the new state; on anything else it surfaces the real error. And I don't trust a screenshot of an API call to prove the screen works — the only proof I count is driving the real button in a real browser and watching the record change after a hard refresh. A rating is worthless if the plumbing beneath it will cheerfully tell you a comfortable story, so I built the plumbing to refuse.

A number that says how good you are is only worth having if the system under it can't be talked into lying. Honesty isn't a layer you add on top of the product — it's the foundation, or the number is just decoration.

Mostly I think the appeal is that it's a small, honest problem. Nobody's underserved in a tragic way; it's just a bit of friction everyone in that gym had quietly worked around. But those are often the ones worth solving — the problems obvious to anyone standing in the room that nobody has gotten around to fixing.

I keep thinking about the spread of skill I could see that first night, and how strange it is that the easiest thing in the world to notice is the hardest thing for the system to know. The score of any single game was never really the point; the point was the shape of the players around it — the thing the room read in a glance and the record couldn't hold. If this ever amounts to more than an idea I chased for a season, it'll be because it closed that one small distance: between what a court full of people plainly is, and what the ledger is willing to admit.