Signal, shares, and the deposit that is burned

Signalling is not a deposit that sits with your name on it. You are exchanging GRT for shares in a pool attached to one subgraph.

What you hold.

Deposit GRT against a subgraph and you receive Graph Curation Shares in return. Your claim on the pool is proportional to the shares you hold relative to all shares issued for that subgraph.

Query fees flowing to that subgraph’s curators accrue to the pool. Your shares are then worth more GRT than you paid for them. To exit, you burn your shares and receive GRT back at whatever the pool is worth at that moment.

deposit GRT burn the tax receive shares fees accrue to pool burn shares to exit

Two things follow from the share model that catch people out.

Your holding is denominated in shares, not GRT. The GRT figure you see against your position is the current redemption value, not a balance. It moves.

Other curators arriving does not dilute your claim. They buy their own shares at the current rate. What their arrival does is change how thinly the subgraph’s future fees are spread, which is a different thing and matters just as much.

The deposit charge.

Entering costs 1%, and this is the detail worth being precise about: it is burned. It does not go to the indexer, the developer, the Foundation, or to earlier curators. It leaves the supply.

That design choice does real work. If the entry charge were paid to existing curators, early positions would be paid by later arrivals, which produces exactly the incentive structure you would expect and would not want. Burning it means nobody in the system profits from your entry, so nobody in the system benefits from encouraging churn.

For you it is straightforwardly a cost. Deposit some amount and slightly less than that amount becomes shares. You have to be right by more than the charge before you are ahead.

Where the fees come from.

A subgraph’s curators share in the query fees generated on it. That flow depends on three things, none of which you control:

Whether anyone queries it. Your entire forecast.
Whether indexers actually serve it. Signal makes it attractive, it does not compel anyone. A subgraph with signal and no indexer allocated to it generates no fees.
What the network takes. 1% of query fees are burned, and indexers take their share subject to the rebate mechanism.

Point two is the one people miss. Signal is necessary for a subgraph to be served and it is not sufficient. If you are checking a position, check that somebody is actually allocated to it, not merely that you have signalled.

What your shares are not.

Not a claim on the subgraph. You do not own it, cannot change it, and cannot stop the developer changing it.
Not a claim on indexing rewards. Those go to indexers and their delegators. Curators are paid from query fees only.
Not transferable in the way a token position usually is. Your exit is by burning shares back to the pool.
Not protected against the subgraph being deprecated. If the developer stops maintaining it and queries move to a replacement, your signal is attached to the old one.

Before reading on: why burn the deposit charge instead of paying it to existing curators?

Because paying it to earlier curators would turn curation into a mechanism where early positions are funded by later arrivals.

Everybody’s incentive would shift from forecasting query demand towards recruiting the next curator, since that is where the reliable money would be. The signal would stop being information about demand and start being information about promotion, and indexers would be back to having no useful guide.

Burning it means the only way to profit from curation is for the subgraph to actually be queried. Every route to a return runs through the forecast being correct, which is the only property that makes signal worth reading.

Check yourself

When you signal, what do you hold?

The curation deposit charge is:

You have signalled on a subgraph with healthy signal but earn nothing. The most likely reason is: