If you have read anything about curation written before the move to Arbitrum One, you have probably read about bonding curves. That material describes a mechanism the protocol no longer uses, and following it will lose you money in a way that is difficult to notice.
What a bonding curve did.
On Ethereum mainnet, the price of a curation share rose as more shares were issued for a subgraph. Signal early and each share cost less. Signal after a rush and each share cost more. Unsignalling moved back down the same curve.
The intent was to reward the curator who identified demand before anyone else, which is a defensible thing to want to reward.
What it produced in practice was a game about position rather than about data. The profitable move was often to identify subgraphs that were about to attract signal, rather than subgraphs that were about to attract queries. Those are different skills, and only one of them is useful to the network.
It also made exits treacherous. Unsignalling from a position you entered late meant selling down a curve, and a rush for the exit made everyone’s exit worse. Curators found themselves reasoning about other curators’ behaviour instead of about query demand, which is precisely the cognitive load the protocol has been removing from its other roles as well.
What Arbitrum does instead.
The curve is Flat. Per the L2 transfer guide, “the curation bonding curves on L2 are flat, which makes it easier for other Curators to curate on subgraphs, increasing the rewards for Indexers.”
Signal and unsignal at a consistent cost regardless of who arrived first. There is no discount for being early and no penalty for being late.
What this changes about how you should think.
Timing is no longer a source of return. If the only reason to signal on something is that others are about to, there is now no mechanism through which that pays. Your return comes from query fees on data that gets queried, and from nothing else.
Exits are no longer a coordination problem. You are not racing other curators down a curve. You can leave a position because your view changed, without modelling whether everyone else is about to leave too.
Being late is not a reason not to act. Under a bonding curve, arriving after a run-up could make an otherwise sound forecast unprofitable. Flat pricing removes that. If the data will be queried, signal on it, whenever you noticed.
Research beats reflexes. The remaining edge is entirely in judging what will actually be queried, which rewards understanding an application’s data needs. That is slower, less exciting, and a considerably healthier basis for a market.
The cost that remains.
Flat does not mean free. Entering still costs 1%, burned. Since that charge lands on entry rather than accruing over time, the economics still favour deciding once and staying put over reshuffling positions.
The difference is what you are being charged for. Under a bonding curve you paid for arriving late. Now you pay for arriving at all, once, at the same rate as everyone else.
Before reading on: flat curves make signalling easier, so why does the L2 guide describe that as increasing rewards for indexers?
Because indexing rewards depend on signal, and the bonding curve was suppressing it.
Under a curve, a curator considering a subgraph that already had substantial signal faced a worse price than the curators before them. That discouraged additional signal on subgraphs that were already identified as valuable, exactly where more indexing capacity would have been most useful.
Flatten it and later curators face the same terms as earlier ones. More signal accumulates on subgraphs people actually believe in, more indexing rewards flow towards them, and indexers have a clearer picture of where capacity is wanted.
The change reads as a curator convenience. It is really a fix to the signal quality the whole allocation mechanism depends on.
On Arbitrum One, signalling early on a subgraph gets you:
The curve is flat. There is no early-mover discount and no late-mover penalty, which removes timing as a source of return entirely.
A guide tells you to watch for other curators exiting before you do. What does this tell you about the guide?
Racing other curators down a curve was an L1 concern. With flat pricing your exit value does not depend on the order of exits, so modelling other curators' behaviour is wasted effort.
Under flat curation, where does a curator's edge come from?
The only remaining route to a return is query fees on data that gets queried. That rewards understanding what applications actually need, which is slower and considerably healthier as a basis for the mechanism.
0 of 3 answered