Double Secret Elimination

On May 10th, in Game 4 of the Western Conference Semifinals, Victor Wembanyama threw an elbow into Naz Reid's throat. The officials called an offensive foul, reviewed the tape, upgraded it to a Flagrant 2, and ejected him. Two days later, the league office decided he would play Game 5. No suspension. A minimum fine of $2,000 — except they waived that too.

That decision was a flagrant 3 on the fans of Basketball. 💩

A Flagrant 2 is, by the league's own rulebook, the most severe in-game disciplinary call available. It exists as a category specifically to mark contact that is unnecessary and excessive — contact that crossed a line the league drew itself. When the league's response to one of its own players crossing that line is "no suspension, no fine, see you Tuesday," the only thing you've learned is that the line isn't real. The rules are made up on the spot by referees you won't see, on criteria nobody publishes, on the eve of an elimination-adjacent game in a series the league has an incentive in extending.

That's an integrity problem where the rules and tournament format give the league office the leverage to make decisions willy-fuckin'-nilly. It matters a lot more than you think.

So I can't make the league change the rules, but I'd like to share a vision that offers hope. The NBA playoff single-elimination conference bracket is a fossil. It was designed for a world that no longer exists. And the longer we keep it, the more leverage it concentrates in exactly the wrong places.

What hope looks like

I said I'd share a vision that offers hope. Here's where it comes from.

I played Killer Queen. For the uninitiated: it's a 5 on 5 arcade game, one of my favorite games, and the entire tournament scene is built on double elimination with bracket reset. A vast majority of major, regional, and local with more than eight teams. Not because of tradition — the scene is barely a decade old. Because the people who built it wanted fairness and honesty after early dishonesty from tournament organizers. The bracket should measure who is best. Not who got the easiest path to grand finals. Not who got the call on the eve of an elimination match. Not who the room in the back wanted to see advance.

I finished 4th at BumbleBash 4 in 2019. BumbleBash is Killer Queen's world championship — when it happens, it's the actual big one. I went in as a Minneapolis player on a roster from a city whose scene didn't exist yet when New York, Chicago, and Portland's top players had already been competing for couple years. Minneapolis got the game in August 2016. Seattle, Spring 2017. We took a losers-bracket elimination match against a Portland team captained by a two-time BumbleBash champion, Damish Shah. In the end we lost to the eventual 2nd and 3rd place teams: New York's and Portland's best. Chicago finally won it that year.

I know we got beat by better teams. I know it was a fair fight.

That sentence is what this whole post is about. That sentence is what I meant when I said this matters a lot more than you think. I finished 4th at the world championship of my game and walked away certain the result was honest, because the format made it honest. Nobody in a back room decided whether the rulebook applied to my bracket. Nobody had every incentive to extend my run for ratings. Nobody waived a fine on an opponent's elbow on the eve of an elimination match. We played until we had been beaten twice — by teams that were better than us — and then we shook hands and went home. The bracket measured something real, and we agreed with what it measured.

That is the experience the NBA cannot give Naz Reid, Ayo Dosunmu, Anthony Edwards, Jaden McDaniels, Julius Randle, Rudy Gobert and Donte DiVincenzo. It cannot give it to Wemby either, by the way — when Wemby wins a championship, there will be a non-trivial number of basketball fans who believe the league office helped him do it, and that belief will be structurally impossible to refute, because the format gives the office the leverage and the office uses the leverage. The players deserve a result they can stand behind. The fans deserve a result they can trust. Neither group gets one under the current format.

The thing single-elimination basketball cannot teach you, and double-elimination tournaments teach you in your first weekend, is how the format changes what a losing means. In single elim, a bad match is an execution. You played 82 regular-season games and one bad night sent you home. In double elim, a loss is data. It tells you what to fix before your next series. It tells the bracket you have one mistake's worth of margin left. It tells your opponents you've been knocked down and are coming back up the hard way, and they should be afraid of that, because the team that just survived three elimination matches in a row is a different team than the one that started the tournament. The format produces a narrative that single elimination structurally cannot produce — the story of the team that got beat, regrouped, came back, and won the whole enchilada. KQ grand finals has that story every time. Basketball has been structurally excluded from it for forty years.

Fighting game majors run double elim. EVO runs double elim. Smash, Tekken, Street Fighter, even Dance Dance Revolution— every great competitive community on the planet runs double elim, because somewhere in the early days of each of those scenes, somebody with discretion got to ask "what should the bracket do?" and the answer was always the same.

The NBA can change. The bracket is the bracket because of "tradition." It's a fossil from a broadcast era that ended fifteen years ago, and nobody at the league office has had the imagination, the spine, or the structural incentive to change it. The leverage concentrates in the wrong places, every year, and every year the league office uses it.

The leagues that thought about it chose double elim. The league that didn't is the NBA.

That's what I mean by hope. There is a format that produces honest results. It exists. People play under it every weekend, around the world, in scenes much smaller than the NBA, with much less money on the line, because they care more about the integrity of the result than the league office of professional basketball does. Hope is knowing the alternative is real and proven. The vision isn't speculative. It's just not here yet.

The spec

Vague visions die in the comments, so here's exactly what I'm proposing.

The format is 32 teams in a double-elimination bracket with bracket reset in the grand final. The whole league is in. No conferences. Seeding is 1 through 32 by regular-season record, ties broken by point differential — not head-to-head, because head-to-head creates incentives to manipulate specific regular-season games, and we are trying to remove those incentives.

Every team makes the postseason. This sounds like dilution. It is the opposite. Right now, the regular season has two jobs — decide who's in, and decide seeding — and it does both badly. By February, both questions are mostly answered, and the back half of the season turns into load-management theater. Putting all 32 teams in the bracket strips the regular season down to one job: seeding. And seeding now matters a great deal, because the 1 seed plays the 32 seed in a best-of-3 where the gap is real. Every regular-season game shifts your draw. Tanking gets harder. Load management gets impossible. The 82-game season finally pays off.

The play-in tournament disappears, because it was the league's tacit admission the regular season wasn't doing enough work. The mid-season Cup can disappear too, for the same reason. You don't need to manufacture tournaments in November when the real one in May is a 32-team double-elim bracket with the entire league in it. Call the first couple rounds "pools" like the Fighting Game Community— whatever!

Winner's bracket:

• Round 1, 32 → 16. Best-of-3. The whole league is in. Half goes home in a week.
• Round 2, 16 → 8. Best-of-3. The pace is relentless on purpose.
• Round 3, 8 → 4. Best-of-5. Real basketball starts here.
• Round 4, 4 → 2. Best-of-5.
• Winner's Final, 2 → 1. Best-of-7. The winner enters the Grand Final with zero series losses and a one-loss cushion. They earned it.

Loser's bracket (you drop in here when you lose in winners):

• LR1: Best-of-3. The sixteen teams that lost Round 1 play each other.
• LR2: Best-of-3. LR1 winners meet the eight teams that just lost Round 2.
• LR3: Best-of-5.
• LR4: Best-of-5. LR3 winners meet the four teams that just lost Round 3.
• LR5: Best-of-5.
• LR6: Best-of-5. LR5 winners meet the two teams that just lost Round 4.
• LR7: Best-of-5.
• Loser's Final: Best-of-7. The team that emerges from this gauntlet plays for the title.

Grand Final: best-of-7 with bracket reset. This is the mechanic most NBA fans haven't seen in a championship context, so I'll explain it. The Winner's Final winner enters with zero series losses. The Loser's Final winner enters with one. If the loser's-bracket team wins the first best-of-seven, both teams are now tied at one series loss apiece — tied on the only metric the bracket measures. So they play a second best-of-seven. The Cinderella has to beat the favorite twice in a row to take the title, because they've already lost once and the favorite hasn't. This is the integrity mechanism. It is the whole point. It exists because we have collectively decided that the team coming up through losers, however heroic their run, has not yet proven what the bracket exists to measure.

A few load-bearing properties of this design worth naming:

The format isn't much longer. With 32 teams and graduated series, worst-case total games comes in around 130–150, vs. the current playoffs' practical max of ~105. About a month more of postseason. The trade is that the regular season can shrink by an equivalent amount, because seeding is now the only thing it has to produce, and you do not need 82 games to produce a clean skill rating. Cut the regular season to 65 games and you've gained postseason content, kept total games roughly equal, and made every game in both halves of the year matter more.

Tanking or throwing games goes away. Every team is in the tournament, so there is no longer a reward for losing on purpose. The reward shifts entirely to winning more, because every additional regular-season win is a worse Round 1 opponent.

Load management disappears. You cannot rest a star in a best-of-three. The format punishes the strategy without the league office having to write a memo about it. This is good rule design. Incentives do the work instead of enforcement.

The discretion budget shrinks. No single series, before the Grand Final, can end a season. A bad call, a bad night, a Flagrant 2 that should have been a suspension and wasn't — these become setbacks, not executions. The league office has less to swing, which means there is less to suspect them of swinging, which means the open secret has less room to operate.

Seeding actually means something. The 1 seed plays the 32 seed in Round 1. The 2 seed plays the 31 seed. If you were the best team in the league for 65 games, your reward is the easiest path on paper and a winner's bracket route that requires five series wins instead of eight. The regular season carries weight again, because the postseason is shaped to honor it.

The objection nobody can answer

The honest objection to this proposal is logistics. Owners won't like it. The schedule is harder. Travel is hard. It goes on.

Well, sponsors can make apparel and build inventory around the new format. The league and networks could use new contracts rather than the old broadcast-tv era windows. The Players Association can renegotiate. The arena calendars will shift.

I don't care, and you shouldn't either. Those are problems for people who are paid to solve problems. They are not arguments about whether the format is correct. They are arguments about whether the format is convenient for the entities who have the most to gain from keeping it as it is — which, again, is exactly the kind of argument you should be suspicious of when the underlying question is integrity.

This is solvable by people with calendars, lawyers, and a quarterly planning cycle. None of them is an argument for keeping a fossil that allows a dude to elbow check another man in the neck.

git gud