I have seen one discussion about increasing Curation rewards. I think the larger issue is about decentralization. Query fees for curators act like the proverbial canary in the coalmine. If they are low, or nonexistent in many cases, curators will cease signalling, and that can turn into a downward spiral that basically says, decentralization is not working. Though I am a curator, I think this indication of the success of decentralization is more important than the fees curators receive.
One possible source of help is to make clear guidelines about the share level required to receive query fees. It remains a mystery to me, and perhaps many curators, how much of a signal is worthwhile, and what the expected return would be.
I realize the network is in bootstrap mode and there are plenty of fundamental issues to work out. In consideration of that, it would be helpful to see a regular status update, and how various improvements are making progress on the road to decentralization. Thoughts?
1 Like
Hey Phil,
Curation incentives definitely need to be addressed for the health of the protocol. With less than 3m GRT in curation, it is easier to manipulate than the intended behavior of finding important subgraphs that will drive queries through the mainnet.
Currently curators have earned ~16,000 in query fees added to their bonding curve since the launch of the mainnet or about a 1% APY marketwide.
There is an issue related to the current gas costs for an indexer to claim the query fees (and thus add GRT to the curator reserve for the subgraph). When multiple indexers are allocated to 1 subgraph that sees 300 GRT in query fees, if the queries are evenly split, the subgraph may result in 0 query fees claimed (Rocket Pool for instance had ~110 GRT in query fees in November and the allocations active for that month reported 2.6 GRT in query fees).
I bring this up because the easiest boost to curator incentives would be to change the query fee cut (even if temporarily), however the more we reduce this the more likely we become that an indexer won’t claim them at all.
More advanced incentive remodeling may need to be done, but each of those would take time and likely contract updates whereas this would be a simple parameter update within existing logic.
1 Like
Fantastic, great insights. I just read that a GIP has reduced gas fees for deploying subgraphs, would this also help with claiming query fees? Could I ask why query fees need to be claimed in the first place, as opposed to simply being deposited?
GIP0019 reduces gas for the first signal. It helps publishers who deploy & signal and it helps curators who are the first to signal. The first time signal occurs it has to set up the ERC20 contract setup for the curation shares etc. This does not effect query fees.
Queries are routed to indexers who serve the data and capture a voucher saying that they served the query. When they close their allocation they submit these vouchers to say ‘this is what I produced in query traffic’. If the query volume doesn’t exceed the cost to report this on chain, most indexers will chose not to do so (since it costs more than it returns).
1 Like
This reveals yet another issue, thanks. The query volume certificate seems like unnecessary overhead. I would venture to guess the design of the decentralized approach mimicked the centralized approach where transaction costs are minimal, or at least don’t affect incentives.
It makes sense to put this on other chains that are lower cost, which seems to be a major priority. But is there any thought toward re-architecting what exists on Ethereum, or to move to an L2?
I’m not quite sure what you mean by this. It would be very inefficient to log each query on chain.
There is attention on moving to an L2 but it’s a very difficult task for a protocol as complex as ours. @rburdett highlights some of the challenges here: Roadmap to L2 utilization - #2 by rburdett
Also take a look at Brandon & Ariel’s responses.
This is why I feel like simply adjusting the query fee % that is deposited into the curation reserve is the best current adjustment to the curation sector. As we pass some of the other hurdles related to gas fees, then we can start perfecting curation incentives better.
Also, I think it may be worth while reducing the bonding curve (making it less steep). I believe the current ratio is 1:2 and if we were to adjust this to 3:4 or 2:3 it would likely orient curators more toward signaling accurately rather than signaling before others.