The protocol contracts and the GRT you stake, delegate or signal with operate on Arbitrum One. Not Ethereum mainnet. This single fact invalidates a surprising proportion of the material written about The Graph between 2020 and 2023, including material published by this Academy’s earlier incarnation.
Why the protocol moved.
The protocol is not a token contract with a website. It is a contract system that processes a constant stream of small, frequent operations: opening and closing allocations, submitting proofs, redeeming payment vouchers, adjusting signal, changing delegation.
On Ethereum mainnet each of those is a transaction at mainnet gas prices. That does not merely make things expensive, it changes behaviour. If closing an allocation costs meaningful money, indexers close fewer of them. If adjusting signal costs meaningful money, curators leave signal where it is. The protocol ends up with participants economising on the very actions it needs them to take freely.
Moving to an L2 changes the arithmetic rather than the design. Operations that were once carefully rationed become routine.
What actually changed for each role.
Curators got the largest structural change. On L1, curation used a bonding curve: early signal on a subgraph was cheaper than later signal, and unsignalling moved back down the curve. On Arbitrum the curve is Flat. Signal and unsignal cost the same regardless of who arrived first. The early-mover advantage is gone, and with it a whole genre of strategy advice that is now simply wrong.
Indexers and delegators kept the same mechanics with a different cost floor. The interesting consequence is behavioural: cheap transactions make frequent allocation management viable, and make the difference between an attentive operator and an inattentive one show up in the returns.
Developers and consumers deal with an Arbitrum One deployment, Arbitrum addresses, and an Arbitrum block explorer. GRT bridged from mainnet is the same token in the sense that matters, but it is a different contract address on a different chain, and pasting a mainnet address into an Arbitrum interface will not end well.
How to date the material you are reading.
Since a great deal of stale writing is still in circulation, and much of it is well written and confident, it helps to have quick tells. Any of these means the page is describing a protocol that no longer exists:
- It tells you to deploy to the Hosted Service.
- It describes Cobb-Douglas rebate pools, or tells indexers to think about what other indexers will stake.
- It describes curation bonding curves as something to time your entry against.
- It mentions a delegation tax you will pay on deposit.
- It says allocations must be closed within a fixed number of epochs to collect rewards.
Each of those was true once. None of them is now, and the delegator, curator and indexer paths each deal with the current version.
Curation on Arbitrum One uses which pricing model?
The flat curve is the substantive L2 change for curators. It removes the early-mover advantage, and with it the strategy of racing to signal first. Advice about timing your entry on the curve is describing L1.
Which of these tells you a guide predates the current protocol?
The Hosted Service is gone. Cobb-Douglas rebates, bonding-curve timing, a delegation tax on deposit, and a hard epoch cap on allocations are the other reliable tells.
0 of 2 answered