Taofather is an argument about Bittensor subnets, settled by the chain instead of by shouting. Somebody says a subnet is weak. Somebody else says it holds. Sunday tells you who was right, and the pot goes to that side.
1. The city
There are 128 subnets on Bittensor, each with its own price, its own liquidity, and its own crowd of people convinced it is either the future or a rug. There is also TAO itself, priced against the dollar, which is the biggest table in the room. That is 129 places to be right or wrong, and every one of them is a market you can call.
You can also set any two subnets against each other and call which one comes out ahead. That is another 8,128 arguments, and most of them have never been settled by anything but shouting.
Subnet prices come from the chain itself, read off each subnet's own reserves. TAO against the dollar comes from Kraken, sampled every minute, one venue named here and never quietly averaged with another. Both feed the same ten minute window, and if that window is too thin to trust, the call is refused rather than approximated.
The game does not trade any of them. Nothing is bought, nothing is sold, no pool is touched. It reads their real prices off a full archive node and uses them to settle arguments.
2. A job
There are two kinds. A job on one mark, where a Boss calls a single market up or down, and a job head to head, where he names two subnets and backs one to beat the other.
One mark
A Boss calls a job on one market and picks a side:
- Hit means it goes down.
- Protect means it goes up.
He writes one line saying why, sets the size he is putting behind it, and pays the fee. The job goes on the board and the books open.
Ride or cross
Until the books shut, anyone can join. You either ride with the Boss, taking his side, or you cross him and take the other. You choose your own size. You pay a fee on that size. Nothing else moves.
A job with nobody on the other side is not an argument. If one side is empty when the books shut, the job is dead and it pays nobody. Nobody wins unopposed.
You lose your fee when you lose an argument, and when you walk away from one. If we could not price the market at the reckoning, that is ours and not yours, and what reached the pot comes back to you.
Your price is your own
The moment you put your name down, the market's price at that moment is recorded as your entry. The Boss gets his entry when he posts; you get yours when you join. Nobody inherits anybody else's number.
This is the rule that stops the last-minute edge. If everyone shared the Boss's entry, someone joining at hour eleven could see the market had already moved and take a free ride on it. With your own entry, joining late buys you nothing but less time.
You are not asked to guess at it. The board shows the entry price before you commit, on the form when you put a job out and on the sheet when you take a side, and it is the same ten minute mean your entry will record. It moves until your fee lands, which is seconds after you sign.
Head to head
The second kind of job puts two subnets in a room. The Boss names a favourite and an underdog and says the favourite beats the underdog this week. Crossing him means backing the underdog. There is no hit and no protect, because nobody is calling either subnet dead. One of them does better than the other, and that is the whole argument.
You take two entry prices instead of one, both at your own moment, and the job takes two exit prices at the reckoning. What you are scored on is the gap between them: how far your subnet moved against the other one, measured from where you came in.
That gap is all that counts, and not the direction either of them went. If the subnet you backed falls ten percent and the other falls thirty, you were right by twenty and you get paid. You can win a pairing in a week where both of them bled, which is a thing no single mark can do.
Because the score is the difference between two moves, writing the job the other way round gives the same number with the opposite sign. There is no advantage in which subnet gets named first, and no order to get wrong.
Both subnets have to price or the job is void and every fee comes back, the same as any other call we could not settle. A pairing is its own argument: you can hold a pairing and a single mark on one of the same subnets at once, because they are different jobs with different pots. What you cannot do is back one subnet over another in one open job and reverse yourself in a second.
The reason it exists: a single mark asks whether a market goes up or down, so a week where everything moves together leaves a board where every job resolves the same way. A pairing asks which one is better. That question survives a flat week, a red week, and most of the arguments people on Bittensor are actually having.
3. The reckoning
Every job in the city settles at the same moment: Sunday, 00:00 UTC. The exit price is the ten-minute average ending at that instant, so nobody can settle a job by pushing a market for one block.
How the pot is split
Your own entry price is checked against the exit. If the market moved the way you called it, you won, whatever your side did. If it did not, you did not.
The pot then splits among everyone who was right, weighted by how much they put up and how far the market moved in their favour from where they came in:
your score = your size × how far it moved your way, as a percentage of your entry
your share = your score ÷ the total score of everyone who was right
Two people can back the same call and be paid very differently. Calling a subnet before it falls fifteen percent is worth far more than scraping in on the last half percent, even at the same size.
A head to head job is scored the same way, with one substitution. Each leg’s move is measured from your own entry on that leg, and what moved your way is the gap between them:
backing the favourite your move = the favourite’s move − the underdog’s move
backing the underdog your move = the underdog’s move − the favourite’s move
Everything after that is identical: size times move, divided by the total of everyone who was right. A pairing does not get its own pot rules, its own fee, or its own minimum.
Why early beats late
Nothing in the game rewards waiting. Your entry is set the moment you place, so the longer you are in, the more of the move belongs to you. Someone who joins in the last hour before the books shut can only be scored on whatever is left of the week, which is usually very little. There is no advantage to hanging back and watching, because the price you would be watching is the price you already gave up by not placing.
Conversely, coming in early is a risk as well as an edge: more time in means more time to be proved wrong, and a call that looked right on Tuesday can be worthless by Sunday.
A late settlement is still priced at the reckoning
If the price data is incomplete when Sunday comes, a job waits rather than settling on numbers we do not trust. When it does settle, it settles on the price at 00:00 UTC, not on the price at the moment the machine caught up. Settlement time is a property of the job, not of when we got to it, and if lateness changed the outcome then an outage would quietly move money between players.
When the books shut
Every job's books shut at the same moment: 24 hours before the reckoning, which is Saturday 00:00 UTC. A job put out on Monday has most of the week for somebody to take the other side; one put out on Friday evening has a few hours. A job put out inside that last day belongs to the reckoning after, not the one a day away, and the board tells the Boss which reckoning he is in before he signs.
A caution about which day. Both instants are the start of their day in UTC, not the end, so for much of the world they fall on the evening before. A week in September, read two ways:
| UTC | New York | London | Singapore | |
|---|---|---|---|---|
| Books shut | Sat 19 Sep 00:00 | Fri 18 Sep, 8pm | Sat 19 Sep, 1am | Sat 19 Sep, 8am |
| The reckoning | Sun 20 Sep 00:00 | Sat 19 Sep, 8pm | Sun 20 Sep, 1am | Sun 20 Sep, 8am |
The two are always exactly a day apart. Which day you call them depends on where you are standing, which is why the board prints both clocks beside each other everywhere it names a time, and why it is worth reading the one in your own time rather than the day's name.
The word midnight does not appear anywhere on this site or on the board. It means the start of a day to some people and the end of it to others, and there is money resting on the difference.
The day between the books shutting and the reckoning is deliberate. A job joinable an hour before settlement would let somebody enter on a nearly finished week, which is not a call, it is a read. The gap is what makes every side take a position on something still unresolved.
Eligibility gates posting, not joining
A market has to be callable for a Boss to post a job on it. Once a job is up, joining stays open even if the market stops being callable partway through the window. The alternative punishes the wrong person: the Boss would watch his job void for want of an opponent and lose his fee for something he had no hand in.
4. What it costs, and where it goes
Posting or joining costs a fee charged as a small percentage of the size you put behind that call. Not of your holdings. Not of your wallet. Of that one call.
| Where it goes | Share of the fee |
|---|---|
| Burned, by your own wallet, gone from supply | 15% |
| Sent to that job's pot | 85% |
| Reaching the pot, after the token's 3% transfer tax | 82.45% |
The fee is 1% of the size you put behind that call. Nothing is charged on your balance, and nothing is charged again at settlement.
You need $MOB in your own wallet before any of this applies to you. Getting in takes four steps.
What it looks like
Say you hold 400,000 $MOB and you put 50,000 behind a call.
| Line | $MOB |
|---|---|
| Size behind the call, scored at settlement | 50,000 |
| Fee you sign for, 1% of that size | 500 |
| of which burned by your own wallet | 75 |
| of which sent to that job's pot | 425 |
| of which reaches it, after the 3% transfer tax | 412.25 |
| Tokens that leave your wallet in total | 500 |
| Tokens still yours the whole time | 399,500 |
A smaller call costs proportionally less. Put 5,000 behind it and the fee is 50, of which 7.5 burns, 42.5 is sent, and about 41.2 reaches the pot after the token's transfer tax. There is no minimum you have to match, no charge on your balance, and nothing further is taken at settlement.
What winning pays
The pot is what reached it: every fee both sides paid, less the fifteen percent each of them burned and less the token's transfer tax on the way in. Whoever read it right splits that. So your return depends on one thing: how many people were on your side relative to the other.
| If the players proved right hold | Back on the fee |
|---|---|
| Half the scoring weight in the job | 1.6× |
| Four tenths of it | 2.0× |
| Three tenths of it | 2.7× |
Those figures are after the burn and after the token's 3% transfer tax on the way in and out. Being right along with everyone else pays least; being right when most of the room called it the other way pays most. That is deliberate, and it is why a board full of one-sided jobs would be a boring one.
Lose and the fee is gone. There is no refund, no consolation, and no way to withdraw once the books shut. The same holds if a payment lands too late to be placed, or does not match what you were asked to sign: nothing is placed and the fee is spent. The board tells you so before you sign, not after.
The size itself never moves. It stays in your wallet from the moment you place until the moment you are paid, and it is read again at settlement, so selling out from under a call clips what you are scored on.
What the multiple means
Every job shows what each side pays. That figure is measured against the fee, not the size, because the fee is the only thing that ever leaves your wallet. A side reading 1.65x means that for every $MOB you paid in fees, about 1.65 comes back if that side is right. If it is wrong, the fee is spent.
Two things follow from the pot being made only of fees. Neither side can pay much above 1.7x when both sides are evenly weighted, because the burn and the transfer tax take their share before anybody eats. And a side nobody is on pays far more than a crowded one, which is the board's way of saying where the argument is thin.
Before you take a side, the board tells you what you would get rather than leaving you to work it out: enter a size and it shows roughly what comes back if that side wins, with your own size counted in. It says about, and it means it. Every player enters at their own price, so what each of them takes out depends on how far the market moved from their own entry, and nobody knows that until the reckoning.
Where the pot sits
Fees do not go to us. Every job's pot is held in a wallet called the count room, named on the board like the rest, and it holds no $MOB of its own. Money arrives when players pay their fee, it sits there while the books are open, and it goes back out to whoever was right when the job settles. Nothing is kept back, and what is in there at any moment is the sum of every open job's pot, which anybody can add up.
5. The token
$MOB launched with a fixed supply of one billion, cut to 900,000,000 by two burns before a single job was posted, with liquidity locked permanently, minting authority revoked and freeze authority revoked. Nothing can print more of it and nobody can freeze your account. It has only fallen since, and what is left right now is on The Body Count, read off the chain rather than typed here. It trades against Bittensor TAO on Solana rather than against the dollar, and every transfer carries a 3% tax, taken in $MOB and paid back out to holders in TAO, in proportion to what they hold. Hold $MOB, earn TAO, whether you ever place a call or not.
Supply only shrinks
A slice of every fee is burned, on top of the 100,000,000 already destroyed, and a job nobody crossed burns every fee that was paid into it. There is no faucet, no emission, no treasury unlock, no vesting cliff, so the number of $MOB in existence can only fall. That is a statement about supply, not about price: what the token is worth is whatever the market says, and burning is not a substitute for anyone wanting it.
Every one of those burns is a transaction anybody can open. They are listed with their signatures on The Body Count, alongside the running total and what the chain says the supply is at this moment. If our arithmetic and the chain ever disagree, that page says so in red rather than quietly picking the flattering number.
How fast depends entirely on how busy the city gets. The model below assumes the launch settings: a fee of 1% of size, 15% of it burned.
| Scenario | Mobsters | Calls / week | Avg size | Burned / year | % of supply |
|---|---|---|---|---|---|
| Quiet streets | 100 | 80,000 | 8,000,000 | 624,000 | 0.07% |
| Word gets around | 1,000 | 150,000 | 150,000,000 | 11,700,000 | 1.30% |
| The city runs on it | 5,000 | 60,000 | 300,000,000 | 23,400,000 | 2.60% |
Read the first row as carefully as the third. A quiet game burns almost nothing, and anyone telling you a new token is structurally deflationary at a hundred players is selling something. The burn matters at volume or it does not matter at all. These are arithmetic on stated assumptions, not projections, not promises, and not a reason to buy anything.
The burn share is deliberately small. It is the game's only rake, and every point of it comes out of what winners are paid. A heavier burn would make a better supply chart and a worse game, and a game nobody plays burns nothing at all.
The Family's cut
The Family bought at launch in the open, in the same first transaction as everyone else, and bought more than the rules called for on purpose: a larger first buy puts more into the pool from the first block, which means less slippage for everyone who came after. It then burned the excess back out, and later bought more off the market and burned that too.
100,000,000 $MOB has been destroyed, a tenth of everything that ever existed. 42,200,000 was burned out of the launch buy (first transaction), and 57,800,000 was later bought off the market and burned too (second). Supply is 900,000,000. Check both against the mint yourself rather than taking our word for it.
What remains sits in three wallets, all published on the board. A fourth, the count room, holds no $MOB of its own: it is where every job's pot sits between the books shutting and the reckoning paying out.
Funding the war chest and seed money meant sending $MOB, and $MOB charges its own 3% transfer tax on every send, ours included. 1,500,000 $MOB was taken out of those two transfers and went into the rewards pool. Note what moves and what gets paid: the tax is taken in $MOB, and holders are paid in TAO. So the two wallets hold 24,250,000 each rather than a round 25,000,000. Against the 900,000,000 that survived the burns, the Family holds 98,546,598, or 11%. Live balances:
| Wallet | Holds | Of supply |
|---|---|---|
| The Taofather's tribute | 50,046,598 | 5.6% |
| The war chest | 24,250,000 | 2.7% |
| Seed money | 24,250,000 | 2.7% |
| Wallet | Supply | What it is for |
|---|---|---|
| The Taofather's tribute | 5.0% | The builder's stake. Bought at launch in public, held at a published address, and no promises attached to it beyond that. |
| The war chest | 2.5% | Operating costs, paid as they come up. |
| Seed money | 2.5% | Posting jobs, and taking the other side of jobs nobody else will, so the board is never empty and nothing settles unopposed. |
SEED MONEY TAKES THE OTHER SIDE
A pot is made of nothing but the fees paid into it. That works when two sides disagree: whoever is wrong funds whoever is right. It does not work when a job is one-sided. If nobody crosses the Boss, the job dies and pays nobody. If everybody agrees and everybody turns out to be right, they split their own fees back and every one of them is slightly down, because there was nobody to be right against.
That is not a flaw in the arithmetic. It is what a shared pot does with nobody to win from, and it is the state a young board spends most of its time in.
So seed money takes the other side. When a side is empty or badly outweighed, the Family crosses it, pays the same one percent as anyone, and takes its entry at its own moment like any other player. Its jobs and its positions are marked on the board, so nobody is in any doubt about who they are arguing with.
Call it what it is: the Family paying to make the game worth playing while there are not yet enough players to play each other. It is capped by the same rule as everyone, one percent of what it puts behind a call, and it stops when both sides fill without it. Any pot it wins goes straight back into seeding.
Two things it is not. It is not a market maker and there is no market to make: the size never moves, so nothing is bought or sold. And it is not the house playing against you. The Family is on the other side of your job only when the alternative is no job at all.
What the war chest actually pays for
- Servers, the archive node, and the price feed that settles every job.
- Domains, hosting, and monitoring.
- Security review of the fee and payout code before it handles anyone's transactions.
- Exchange, aggregator and data listings if the token gets far enough to need them.
- Prize top-ups for one-off events, announced in advance, never quietly.
It is not a marketing slush fund and it is not a trading account. It will never take a side in a job, never post a call, and never trade against players. Every transaction it makes is on Solana at a published address, so what it spends is visible whether we describe it or not.
What seed money actually does
Early on, a board with two jobs on it is worse than no board. Seed money posts calls and takes sides so there is something to argue with. It pays the same 1% fee as anyone else, wins and loses like anyone else, and anything it wins goes straight back into seeding rather than out to us. Every one of its calls is marked on the board as the Family's, so you always know who you are riding with or crossing.
There is no separate creator fee on trading. The Family earns what every other holder earns, from the same 3% transfer tax, paid in TAO, in proportion to what it holds.
6. Why this cannot rug you
Read this section skeptically. It is the one that matters.
There is nothing to take
No deposits. No vault. No token approval to sign. The size you put behind a call stays in your wallet from the moment you place until the moment the job settles. The only transaction you ever sign is the fee, and your wallet will show you exactly what it does: burn a small amount, send the rest to that job's pot address. If a page ever asks you to approve unlimited spending or to send your balance somewhere, it is not us.
There is nothing to inflate
Supply is fixed at one billion, minting authority is dead, and liquidity is locked permanently by the launchpad. There is no team allocation beyond the 10% bought in public at launch, no vesting schedule, no unlock cliff, no treasury that can print. Every number in this document can be checked against the mint account on any block explorer.
Everything is on the chain
Every fee, every burn, and every pot payout is a Solana transaction anyone can look up. Every settled job publishes both sides, both totals, every entry price, the exit price and the payout, and the dossier stays up permanently. If the site ever disagrees with the chain, believe the chain and say so publicly.
The one risk that remains
The Family can sell its own tokens, and so can anyone else holding them. No design removes that, and anyone telling you otherwise is lying to you. What we do instead is name every Family wallet before it holds anything, so all three are public from the first block and every movement in or out is visible on Solana as it happens. We are not going to promise you advance notice of what we do with our own tokens, because a promise like that is worth nothing and gets broken the first time it is inconvenient. Watch the addresses instead.
7. Names and standing
A wallet is a string of characters and nobody remembers one. So a player can claim a name, once their wallet has proved it is theirs by signing, and that name shows on every job they touch. The wallet is always printed beside it. A name decorates an identity here; it never replaces one, and nothing is ever generated for a wallet that has not claimed one.
The ladder
Standing is earned by being right, and by showing up. Every rank pairs a number of settled jobs with a place in the standings, so nobody gets called a Consigliere off one lucky call.
| Rank | What it means |
|---|---|
| Associate | Hangs around. Nothing settled yet. |
| Soldier | Made his bones. Three jobs settled and ahead on the money. |
| Capo | Runs a crew. Ten settled and in the upper half of the standings. |
| Consigliere | Gets listened to. Twenty settled and near the top. |
| Underboss | Second chair. Forty settled and among the best in the city. |
| The Commission | Five seats, the highest earners, held week to week. |
| The Taofather | Nobody outranks him. Not earned, not ranked, not for sale. |
Made is made. Once you are a Soldier you stay one. Everything above it can be lost, and it is lost only after two weeks below the line, not one, because the Family does not break a man over a single bad week. When it happens, that is what it is called: broken.
Standing moves at the reckoning, with everything else. Not continuously, and never because somebody joined a job.
The blessing
Every week the Taofather picks one job off the board and puts his mark on it. He picks it himself and says why in one line, and that line is printed on the job.
A blessed job sits at the top of the wall for the week, above everything, however anyone sorts or filters the board. If it settles, it counts as two jobs toward the rank of everyone in it, on either side, win or lose. Being in the right argument is worth as much as being right.
Three things it is not. It counts toward rank, not profit: the standings rank by what players actually took home, and the blessing does not touch that. It does nothing for a job that voids, because a job nobody crossed never settles and a blessing leaves no mark on a job that was never argued. And it adds nothing to the pot.
The Taofather never blesses a job the Family posted.
8. The record
When a job settles, its dossier is published and never comes down: both sides, both totals, every entry price, the exit price, the move, the verdict, and who took the pot. Before it settles, the things that could be gamed stay sealed.
Over enough jobs this becomes the only thing here worth keeping: a public record of which subnets the crowd calls correctly and which ones it never does, priced by how much people were willing to put behind their opinion.
Anyone can move a market
Settlement uses what the market actually did, and that includes what someone deliberately did to it. A player with enough money can push a price, and a shallow market takes less money to push than a deep one. We do not gate which markets can be called and we do not rank them by how safe they are, because a warning that only appears on thin markets would wrongly suggest the others cannot be moved.
What the board does instead is show you the depth. Every job carries the market's TAO reserve next to it, so before you ride or cross you can work out roughly what it would cost someone to move the settlement price, and price that risk into your own decision. Calling a hit on a shallow market is allowed, and it tells the room something about the person calling it.
9. What can go wrong
- Nobody plays. The most likely outcome for any new game. A thin board means small pots and slow weeks, and the burn stays near zero.
- Thin markets. Only markets above a liquidity floor are eligible, so a subnet cannot be shoved around cheaply on settlement night.
- Data outages. If the price feed goes stale, jobs hold rather than settle on numbers we do not trust.
- You are wrong. Frequently. That is the game, and the dossier keeps the receipt.