The college football playoff format, and how it got here

A team can finish a season unbeaten and be told it is seventh. That is not the college football playoff format breaking down. It is the format doing precisely what it was built to do, because what gets ranked here is not a record. It is a résumé, and a committee sitting in a hotel room has decided that the twelve opponents this particular unbeaten team happened to play were the wrong twelve opponents.

Nothing else in American sport works this way. Every other postseason is a table with a line drawn through it. Win enough games, finish above the line, get in. The argument, if there is one, is about a tiebreaker.

College football cannot do that, and the reason is arithmetic rather than politics. Everything else about the sport's postseason, the century of polls, the bowl contracts, the formula that replaced them, the committee that replaced the formula, follows from one structural fact that no amount of reform can touch.

Too many teams, too few games, and no way to compare them

The top division of college football carries well over a hundred teams. Each of them plays a twelve-game regular season, and a conference championship game on top of that if it qualifies.

Do the division. A team faces fewer than a tenth of the other teams in its own division. It never meets the rest, will never meet them, and in most cases has not met them within the professional lifetime of anyone currently on the roster.

• 130Teams in the worked division
• 12Regular season games each
• 117Opponents a team never plays
• 9Share of the field played, per cent

Invented round numbers chosen to show the arithmetic rather than to state the size of the real division, which changes with realignment and is published by the NCAA. Percentage rounded.

Compare that with the professional game. Thirty-two clubs play seventeen games each on a schedule built by a central office to a fixed rotation, so every club's opponents are known in advance to be roughly comparable, and the standings can therefore be read straight off. That is why turning a regular season record into a playoff bracket is an exercise in tiebreakers rather than in judgement. Baseball plays a season long enough to drown almost all the noise. Basketball plays eighty-two games against everybody.

College football plays twelve, against a schedule the team helped choose.

Roughly two thirds to three quarters of those games are conference games, and the conferences differ enormously in depth. The remainder is non-conference, self-selected, and negotiated years in advance by athletics directors who have every incentive to make it easy. So two unbeaten teams can finish a season having shared no common opponents at all, and having played schedules that are not comparable in any honest sense.

Transitivity does not rescue you either. In a large league, the chain of results eventually pins everybody down: A beat B, B beat C, and after eighty-two games those chains are dense enough to mean something. In a twelve-game season they are not. A single result is one afternoon, in one stadium, in one set of conditions, and stringing three of them together produces a claim no statistician would defend.

That is the problem. A hundred-odd teams, twelve data points each, and almost no overlap between them. The results are real but they do not add up to a ranking, and something outside the results has to supply the missing comparison.

Every format this sport has ever used is a different answer to that one question. A poll asks people. A formula asks a computer, then asks people, then averages the two. A committee asks a room. None of them are measuring. All of them are estimating, and the arguments that follow are arguments about the estimate.

There is a second consequence, and it is the one that made the sport rich. Because the season is short, each game carries an enormous share of the evidence. One loss in September is a twelfth of a team's case. That is why a Saturday in October can feel like a knockout tie, why the regular season sells the way it does, and why every proposal to expand the postseason is met with the objection that it will make the regular season matter less. The objection is correct. It is also the price of the fix.

Before the playoff, the champion was a vote and the bowls were exhibitions

For most of the sport's history nobody played for the national title, because there was no game to play. Writers voted in one poll, coaches voted in another, and the team at the top when the voting closed was the champion. When the two polls disagreed, and they did, the season produced two champions and both of them printed a banner.

The bowls had nothing to do with any of this. They were not a postseason. They were independent events, run by local organising committees in warm cities over the New Year holiday, built to fill hotel rooms and sell a television window in a dead part of the calendar. To guarantee themselves a draw, they signed contracts with conferences: this bowl takes the champion of that league, that bowl takes the champion of this one, every year, whoever it happens to be.

Those tie-ins were the whole business model. They were also the reason the sport could not stage a title game. If the two best teams in the country came from two conferences contracted to two different bowls, they were not going to meet. Not because anybody had decided against it, but because the calendar was already sold.

So the polls filled the gap. What they fixed is easy to state: they produced a champion at all, at no cost, without renegotiating a single contract or extending the season by a day. What they broke is subtler and it has never entirely gone away.

Polls are anchored. A preseason ballot gets published before anyone has played, based on last year's roster and this year's recruiting, and every subsequent ballot is an edit of the one before it. A team that starts fifth does not overtake a team that starts first by playing well. It overtakes it when that team loses. Voters do not rebuild their ballot from scratch each week, because nobody has the time, so the preseason guess propagates through the whole season and the sport spends four months arguing with a document written in August.

The committee era inherited the criticism and answered it with a rule: the rankings are rebuilt each week from the results so far, with no carry-over ballot. Whether human beings can genuinely do that, having ranked the same teams six days earlier, is a fair question. But the instruction is there, and it exists because of what polls did for seventy years.

The BCS put a formula on it, and then everyone audited the formula

The first attempts to fix the bowls were negotiated buy-outs. The Bowl Coalition, then the Bowl Alliance, persuaded some of the major conferences and bowls to loosen their tie-ins so that the top two teams could be steered into the same game. The Rose Bowl and its two conferences stayed out, which meant the fix worked except in the years it was most needed.

The Bowl Championship Series, from the 1998 season, got everybody in. The deal was straightforward: the top two teams in an agreed ranking would meet in a designated title game, rotating among the major bowls, and the rest of those bowls would be filled by a mix of automatic qualifiers and at-large picks.

The agreed ranking was the innovation, and it was a formula. Human polls on one side, a set of computer rankings on the other, combined into a single number to two decimal places. Early versions carried explicit terms for strength of schedule and for number of losses. Later versions folded those inside, averaged a panel of computer rankings, and discarded each team's highest and lowest computer score to stop any single model from swinging a result.

One detail from that period explains more than the rest of it put together. Margin of victory was removed from the computer inputs, because a formula that rewarded winning by more encouraged teams to run up scores on opponents they had already beaten. Removing it made the rankings measurably worse at predicting who would win a football game. It was removed anyway.

That is the tell. The BCS standings were never purely a measurement. They were a governing document, and a governing document has to be defensible in public as well as accurate, which is a different and usually incompatible test.

What the BCS fixed: in most seasons, the two teams with the strongest claim actually played each other. After a century, that was not a small thing.

What it broke, in order of seriousness. It made the ranking itself the entire argument, because when a champion is selected by a published formula every excluded party can audit the formula line by line and demonstrate exactly which component cost them a place. It formalised a two-tier structure, with automatic qualification for the champions of a set list of conferences and grudging, conditional access rules for everybody else. And with only two places, a third unbeaten team had no remedy whatsoever. The grievance changed from "we never got the chance" to "we never got the chance, and here is the arithmetic that excluded us", which is a considerably worse position for an administrator to defend.

The BCS was killed by being good at its job. Year after year it delivered one excellent game and one furious constituency, and eventually the constituencies outnumbered the games.

Four teams solved the top of the argument and moved it downwards

From the 2014 season the sport ran a four-team bracket. Semi-finals rotating among the major bowls, a final bid out to a host city, and the field chosen by a selection committee rather than by a formula.

Throwing out the formula was deliberate and the reasoning is worth stating properly. Any published formula becomes two things at once: a target to be optimised against, and a document to be litigated in public. Athletics directors will schedule to the formula, conferences will lobby to change its inputs, and every excluded team will produce a spreadsheet. A committee cannot be reverse-engineered in the same way. It can be criticised, but it cannot be gamed with the same precision, because nobody outside the room knows the weights and there are no weights to know.

What four teams fixed: a third unbeaten team now had somewhere to go, and the title was settled by two games on the field rather than by a poll.

What four teams broke: it moved the marginal argument from second place to fourth, which is a far harder call. The gap between the second and third best teams in a season is usually visible. The gap between the fourth and the fifth almost never is, and the sport spent a decade holding a national argument about a distinction too fine to be made.

Four places also closed the field in practice. A conference champion from outside the wealthiest leagues had close to no path, because the committee's stated criteria all point back at the quality of opponents faced, and a team in a weaker league cannot manufacture strong opponents in a schedule negotiated years earlier. Meanwhile a single September loss could effectively eliminate a very good team before the committee had met for the first time.

And it hollowed out everything below it. Once four teams were playing for something real, the other thirty-odd bowls were exhibitions again, and players with professional futures began declining to play in them. That has direct consequences for the calendar that feeds the professional game, which is why the tape a scout grades and the player who eventually reports can be the better part of a year apart by the time the draft comes round.

What the expanded college football playoff format changed, and what it cost

From the 2024 season the bracket expanded. The structure, as distinct from the numbers, works like this. A set of places is reserved for the highest-ranked conference champions. The remaining places are at-large, filled by the committee from its own rankings without any requirement that a team has won a title. The field is seeded, the top seeds receive byes, the first round is played at the higher seed's home stadium, the quarter-finals and semi-finals are played in bowls, and the final is played at a bid city.

The numbers attached to that structure are exactly the part you should not take from an article. The size of the field, the number of automatic places and the rule that assigns seeds have all been argued over, and at least one of them has already been revised once: the first expanded bracket seeded the top conference champions into the byes regardless of where the committee had ranked them, which produced brackets where a team ranked well down the list held a bye while a higher-ranked team played an extra game. That was replaced with straight seeding off the committee's ranking. Take the current numbers from the published format at the College Football Playoff's own site.

What expansion fixed is real and underrated. The marginal argument moved from the fourth place to the last one, where it costs much less. A team arguing that it should be twelfth rather than thirteenth is arguing about a road game in the first round, not about being locked out of the championship. A team ranked fourth that gets seeded fifth loses a bye, which is a penalty rather than an execution. Champions of leagues outside the biggest conferences get in, which restores a reason for those leagues' seasons to exist.

What it cost is also real. More games, played by players who until recently were not paid at all and are now paid under a system nobody has finished designing. A first round played in December, sometimes in genuinely severe weather. And a regular season that matters a little less at the top, because a one-loss team from a strong conference is now safe in a way it never used to be. That last one is the trade the sport made knowingly, and the people who warned about it were not wrong. They were outvoted.

Inside the selection committee: who votes, and how the votes actually work

This is the part that makes college football's postseason unlike any other, so it is worth being precise about.

The committee is a mixed group. Sitting athletics directors from the member conferences make up the largest block, joined by former coaches, former players, former administrators and a small number of others, sometimes including a sitting university leader or someone from the media side. Terms are staggered so that the whole group never turns over at once, and a chair is named who fronts the rankings publicly and answers for them.

Members are paid a stipend, not a salary. Most of them hold a full-time job at an institution with a direct financial interest in the outcome of the vote they are casting. That is not a scandal, it is a design constraint, and the recusal rules exist because of it.

The committee is instructed to weigh a defined set of things: championships won, strength of schedule, head-to-head results, results against common opponents, and other relevant factors including the unavailability of key players. Read the instruction carefully and notice its shape. Those are not weights in a formula. They are the considerations to apply when comparing teams whose records are comparable, which is to say they are tiebreakers deployed by judgement, in an order the committee decides case by case.

The voting is where the design gets genuinely clever, and almost nobody describes it accurately.

Members do not sit down and write out a list from one to twenty-five. They rank a handful of teams at a time, by secret ballot, in repeated rounds. First a pool is established: members list the teams they believe belong in the conversation, and a team named by enough members enters the pool. Then a round runs. Each member lists the best six from the pool, unranked, and the six teams appearing on the most lists become the ballot. Each member then ranks those six against each other. The points are totalled, with the highest and lowest scores for a team discarded to blunt any single member's outlier, and the top three are seated in the rankings. The three that were not seated stay live, three fresh teams are added, and the round runs again. Any member may call for a team already placed to be reconsidered and moved.

1. 1The pool is builtMembers list the teams they consider worth ranking. A team named by enough members enters the pool for that meeting. Nothing has been ranked yet.
2. 2Six teams are nominatedEach member lists the best six from the pool, unranked and by secret ballot. The six appearing on the most lists become the working ballot.
3. 3Those six are ranked against each otherEach member ranks the six. This is the whole point of the design: members compare a small named group directly rather than trying to hold a hundred teams in order at once.
4. 4Points are totalled and outliers trimmedThe highest and lowest scores a team receives are discarded, so one member cannot move a team on their own.
5. 5Three teams are seatedThe top three go into the rankings. The other three stay live and carry into the next round.
6. 6Three new candidates join themThe round repeats with the three unseated teams plus three fresh ones, working down the board in small comparisons.
7. 7Recusals are enforced throughoutA member connected to a school takes no part in any vote involving it, though they may answer questions about it.
8. 8Any team can be recalledA member may move that a team already placed be reconsidered and moved, which is what stops an early round locking in a mistake.
9. 9The chair publishes and explainsThe ranking is released with an explanation of the reasoning, which is a reconstruction of a secret ballot rather than a record of it.

The shape of the published selection protocol. The exact group sizes and the number of teams seated per round are set out by the CFP and have been adjusted; check the published procedure for the current values.

Why rank three at a time rather than twenty-five at once? Because people are bad at global orderings and reasonably good at pairwise comparison. Asked to produce a list of twenty-five, a voter will anchor on last week's list and shuffle a few names. Asked which of these six is best, they have to actually compare them. Every team's position therefore emerges from a vote against specific named rivals, not from an average of twenty-five private lists.

The recusal rules are narrower than most people assume and they are worth knowing exactly. A member must recuse from voting on a school where they or an immediate family member are employed or otherwise compensated. Recused members may answer factual questions about that school. They do not leave the room.

That last point is the honest weakness. Recusal removes the vote, not the influence. The discussion is where a ranking is actually made, and a member sits through the discussion of their own school and of its direct rivals. Nor does recusal follow a member's alma mater by itself, which is a defensible line to draw and an uncomfortable one.

The committee has access to a large volume of data and is explicitly not bound to any of it. No metric is authoritative. That is deliberate, and it is the clearest possible contrast with the way an open pyramid settles the same question. In a league with promotion and relegation, the table decides and nobody votes on it: finish in the places and you go up, and there is no committee to petition. College football cannot use that mechanism, because it has no table. It has a hundred and thirty separate twelve-game experiments and a room full of people trying to reconcile them.

Automatic bids reward a title, at-large places reward a résumé

Two defensible principles pull in opposite directions here, and the format has to hold both.

The first says that winning your conference should get you in. It is the only thing in the sport decided purely on the field, by teams playing each other, with no vote involved. Reserving places for the highest-ranked conference champions guarantees access to leagues that would never earn an at-large place on the committee's rankings, and without that guarantee those leagues' seasons become exhibition football from the first week of September. A competition in which most members are mathematically eliminated before kick-off is not really a competition.

The second says that the best teams should be there, and a conference title is weak evidence of being one of them. In a conference of eighteen teams where a member plays a third of the field, the champion is the team with the friendliest rotation as much as the best team in the league. That is not a slight on the champion. It is the sampling problem again, reappearing inside a single conference.

The tension shows up in the same place every year: the last automatic qualifier ranked far below the first team left out. Both complaints about that are correct simultaneously. The champion did the one thing the sport asks a team to do. The excluded team beat better opponents. There is no arrangement of the rules that makes both of those claims lose.

There is also a paradox in the conference championship games. A team that has already secured its place plays one more high-risk game against a good opponent, while a team that failed to reach its title game sits at home resting and cannot lose. The committee is instructed not to punish a team for losing a game it earned the right to play, and it says it does not. Whether that instruction survives a close call between a team that lost a title game and a team that never played one is a question the sport re-asks every December.

Seeding, byes, and the argument about where the games are played

With a field that is not a power of two, some teams skip the first round. Who receives those byes has been the single most contested rule in the format, and it is the clearest illustration of the automatic-bid tension in practice.

One philosophy: reserve the byes for conference champions, on the ground that winning your league should buy the best structural reward the bracket has. The other: give the byes to the highest-ranked teams, on the ground that seeding should reflect the ranking the committee just spent three months producing. The first version was tried and produced brackets that looked indefensible on paper, with teams ranked well down the list holding byes over teams ranked above them. It was replaced.

The general lesson generalises beyond this sport. Any rule that decouples seed from rank will eventually produce a bracket that looks absurd, and it will then be changed. Anyone reading this after another round of revisions should take the structure from here and the numbers from the published format.

Home sites are the best thing the expanded format introduced. The first round is played at the higher seed's stadium, which is the first time this sport's postseason has offered a genuine home game as a reward for finishing higher. A December playoff game in a full campus stadium is a materially different event from a neutral bowl with two travelling supports and a lot of empty seats, and it gives the regular season back some of the meaning that expansion took away.

It also created the weather argument, which resurfaces every winter. Northern campuses in December are cold, and there have been proposals to move first-round games to warm neutral sites for player welfare and travel convenience. The counter-argument is short: a home advantage available only in mild climates is not a home advantage, it is a subsidy.

The later rounds stay in bowls. Quarter-finals and semi-finals are contracted to a rotating set of them, and the final is bid out to a host city in the manner of a Super Bowl. For a supporter, a deep run therefore means a campus game, then a bowl city, then another bowl city, then a final, inside a few weeks and at their own expense. That is a real cost of expansion and it falls entirely on the people least able to spread it.

One rule that does not change between rounds: postseason games cannot end level. College football settles ties with its own alternating-possession procedure, each team starting a series in range of the goal, which is nothing like the professional overtime rules and their arguments about who gets the ball first. Two teams can trade scores for several rounds before it breaks. It is the one part of the postseason nobody has proposed changing.

1. 1Conference championship weekendThe last games that count toward anyone's résumé are played. Titles are settled on the field, which is the only part of this whole process nobody votes on.
2. 2The committee meets for the last timeThe ranking is rebuilt from the completed season rather than edited from the previous week's list, with every connected member recused from votes involving their own school.
3. 3The final ranking is publishedOne ordered list. Everything that follows is mechanical, applied to that list rather than argued about again.
4. 4Reserved places go to conference championsThe highest-ranked champions take the automatic places, which is how leagues that would never earn an at-large place get into the bracket at all.
5. 5The rest of the field is filled at largeThe next best teams on the ranking take the remaining places, whether or not they won anything. This is where a very good second-place team gets in ahead of a champion.
6. 6The bracket is seeded and byes are assignedSeeds follow the ranking and the top seeds skip the first round. This is the rule that was changed after the first expanded bracket, so check the published version.
7. 7First round at campus sitesThe higher seed hosts in its own stadium in December, weather included. The only genuine home game the postseason offers.
8. 8Quarter-finals and semi-finals in bowlsContracted bowl games on a rotation, at neutral sites, with the bowl name attached to what is now a playoff fixture.
9. 9The final at a bid cityAwarded to a host city years in advance. A champion decided on the field, from a field decided by vote.

The route through the format rather than its dimensions. The size of the field, the number of reserved places and the rule assigning byes are published by the CFP and have already been revised once.

Why the bowls did not simply die

There are roughly forty bowl games. The overwhelming majority have nothing whatsoever to do with the playoff, and they are still here.

They survive for reasons that have very little to do with football. A bowl is television inventory with a guaranteed audience in a stretch of the calendar with nothing else in it, and it pays conferences, which distribute the money to members. It also gives a six-win team an extra block of practices before the game, which is genuinely valuable for a young roster: a second spring camp with a fixture at the end of it. Coaches will defend that practice allowance long after they have stopped defending the game.

The playoff did not kill the bowls because the playoff needed them. The bowls owned the stadium relationships, the sponsor deals and the television windows, and the cheapest way to build a bracket was to rent those assets rather than to fight them. That is why the quarter-finals and semi-finals carry bowl names. The playoff bought the bowls' cooperation rather than their extinction, which is the same move the BCS made and the Bowl Alliance made before it. Every reform in this sport's history is a negotiation with the same handful of counterparties.

What has genuinely eroded is participation. Players with professional futures started skipping non-playoff bowls to protect their draft position, and then the transfer portal removed a second group who had already committed elsewhere. A December bowl now often features a team missing several starters and led by an interim coach whose predecessor left in November.

The bowls have responded by leaning into the exhibition, and there is nothing wrong with that. Not every game has to matter. It does mean the word "bowl" now covers two completely different products, and a reader should keep them separate: a playoff quarter-final that happens to be branded as a bowl, and a friendly between two 6-6 teams with a sponsor's name on it.

Conference realignment is the force underneath the whole format

None of the above is stable, because the conferences themselves are not stable.

The mechanism is media rights. A conference sells its inventory collectively, and the value depends on how much of it there is and how many people will watch. Adding a school that delivers a large television market or a national brand raises the payout for every existing member. So conferences recruit and schools move, and the regional logic that produced these leagues in the first place has been dismantled almost entirely.

The playoff accelerated it. If access is decided by a committee weighing strength of schedule, then belonging to a conference the committee respects is worth something concrete. A school outside those conferences can go unbeaten and still be ranked below a two-loss team from a stronger league, and the only remedy available to that school is to change conference. Realignment is therefore not merely a money grab. It is a rational response to a selection process that rewards the company you keep.

What it did to scheduling is the part that feeds straight back into the sampling problem. Conferences got large enough that a full round robin became impossible, so members now play a rotating subset of their own league. Two teams can finish a conference season on identical records having played substantially different sets of opponents. The one reliable comparison the sport used to have, common opponents inside a conference, has been weakened from within.

And it did two contradictory things to the value of a conference title. In a huge league, the title is harder to win and tells you less about who was best, because the champion played a third of the field. But because places in the bracket are reserved for champions, the title is worth more than it has ever been in access terms. Less informative, more valuable. That is a strange combination and it is entirely a product of the format.

The geographic cost is real and it is paid by the people with the least say. Cross-country travel for a midweek game in a non-revenue sport falls on athletes and on students. The television money is collected centrally.

The criticisms that actually land

Plenty of complaints about the format are noise. These five are not.

Strength of schedule is not measurable in this sample. Every strength-of-schedule figure is a model output, the models disagree with each other, and they disagree because there is not enough interconnection in the data to pin the conferences relative to one another. Non-conference games are the only bridges between leagues, there are very few of them, and most are deliberate mismatches. Any strength-of-schedule number is therefore an opinion with decimal places attached, which is the most dangerous form an opinion can take.

The scheduling incentive runs backwards. A team that schedules three difficult non-conference opponents raises its expected number of losses. The committee says it rewards schedule strength, but the reward is vague and the cost of a loss is sharp and immediate. So the rational athletics director buys home games against overmatched opponents, pays a guarantee, and banks a win. The buy game is a functioning market: the visiting school takes a fee that can fund a substantial part of its athletics department for the year and accepts a beating in exchange. Everyone involved knows exactly what is happening.

The resource gap converts directly into results. A small number of programmes generate revenue that dwarfs the rest, and now that players can be paid, that revenue turns into roster quality without an intermediate step. Professional leagues attack the same gap deliberately, with drafts that hand the best young players to the worst teams and, in basketball, a lottery designed to stop that reward becoming a reason to lose. College football has no mechanism running in that direction at all. Nothing in the structure redistributes anything downwards.

Availability is no longer stable in January. Players opt out of games that no longer serve them, transfer windows open in the middle of the postseason, and coaching changes happen while a team is still alive in the bracket. The committee is instructed to weigh the unavailability of key players, which is a quiet admission that a team's record was compiled by a roster that may no longer exist by the time it is being ranked.

The ballot cannot be audited. Votes are secret and the criteria carry no published weights, so the output cannot be checked against the process. The chair explains the ranking afterwards, and that explanation is a reconstruction rather than a record. There is no way for anyone outside the room to establish whether the stated reasons are the operative ones. That is the price of not having a formula, and it was paid on purpose by people who had watched a formula get torn apart for fifteen years.

• Wins over teams with a winning record
• Wins over teams with a losing record

Team Blue65

Team Red47

Team Gold29

Every number here is invented to show the method. Team Blue, Team Gold and Team Red are not real teams and no real season is being described. Opponents split by whether they finished above or below a .500 record.

Show the numbers

Worked example: two 11-1 records that are not the same résumé

Item	Wins over teams with a winning record	Wins over teams with a losing record
Team Blue	6	5
Team Red	4	7
Team Gold	2	9

Three identical records. Read them by opponent quality and they stop being identical, which is roughly what the committee claims to be doing when it puts one 11-1 team four places above another.

Money, the portal, and who is actually on the field

The format now sits on top of a compensation system that did not exist when the playoff was designed.

Players can earn from their name, image and likeness, and schools can share revenue directly with them under a settlement that caps how much each school may distribute. That cap is set annually and adjusts over the life of the agreement, so take the figure from the settlement documents rather than from a headline. The structural point is that it exists at all, and that it arrived through litigation rather than through bargaining.

That last distinction matters more than the number. The professional game solves the same problem with a hard cap agreed between a league and a union, which is why what a professional club may spend is knowable to the dollar in advance. College football arrived at revenue sharing without a union, without a collective agreement, and without anyone on the players' side of the table with the standing to sign a deal. The rules are therefore set by settlement terms and by whatever the next case establishes, which is a fundamentally less stable arrangement than the one it resembles.

For the playoff specifically, three consequences follow.

Reaching the bracket is now a recruiting argument with a price attached. Postseason exposure converts into value for players, value attracts players, and better players make the bracket more likely. The loop runs in one direction and it runs fastest for the programmes that were already ahead.

The transfer windows collide with the postseason. A player on a team still alive in the bracket faces a market with a fixed closing date while he is preparing for a quarter-final. Whatever the calendar says on paper, that is a genuine conflict for a twenty-year-old, and the further a team advances the worse it gets. Advancing in the bracket is the one thing that shortens a player's window to secure his own future.

And the replacement of scholarship limits with roster limits changed how squads are built. Instead of a fixed number of funded places plus an unlimited tail of walk-ons, there is a hard cap on bodies. A four-round bracket asks far more of depth than a single bowl game ever did, so the expanded format has quietly raised the value of the thirtieth-best player on a roster at exactly the moment the rules capped how many players a roster can hold.

How to read a résumé the way the committee says it does

If you want to argue about the bracket usefully rather than loudly, stop counting wins and start reading them. Six questions do most of the work.

How many of the wins were over teams that finished above .500? Split any record into wins over winning teams and wins over losing teams before you compare it to anything. Most identical records stop being identical immediately, and the ones that survive this test are the ones actually worth arguing about.

Where were those games played? A road win over a good team is stronger evidence than a home win over the same team, and the committee's own criteria say so. Count how many of a team's best wins came in its own stadium before you accept the résumé at face value.

Are there any genuine common opponents, and only those? Two teams that both played the same third team give you one honest data point. Use it. Chains longer than that, where A beat B who beat C who beat D, tell you nothing at all in a twelve-game sample, and anybody building an argument out of one is selling you something.

What do the losses look like, rather than how many there are? A one-score road defeat to an excellent team in late November is different evidence from a home defeat to a mediocre one in September. A record cannot express that difference. The committee is explicitly permitted to, and this is the place where its judgement most often earns its keep.

Who was available, on both sides? For the team you are judging and for the teams it beat. A win over a good opponent missing its quarterback is a lesser win, and the criteria now say as much.

Then ask the question underneath all of it. If these two teams played ten times, which one wins more often? Everything above is an attempt to approximate an answer to that, and the honest response in most close cases is that nobody knows, including the thirteen people whose job it is to decide.

That last admission is the real subject here. The sport keeps rebuilding its postseason because the underlying problem cannot be solved, only traded around. Twelve games cannot rank a hundred and thirty teams, and no committee, formula or poll can conjure the missing information out of a schedule that was never designed to produce it. Polls hid the problem, the BCS put a decimal point on it, four teams narrowed it, and expansion moved it somewhere cheaper. The next format will move it again.

Which makes the postseason argument permanent rather than fixable, and explains why it is a better lens on the way American football is organised than any other single thing in the sport. Every format is a statement about what the people running it think a season is evidence of. Read the bracket that way and the annual outrage becomes something more interesting: an argument about the limits of what results can prove.