Oracles: the hard part of DeFi is not the math
Oracles: the hard part of DeFi is not the math
Short one from the good news desk.
The equations of automated market makers are simple and public. The dangerous question is what price the protocol believes. A lending market that reads its collateral price from one thin exchange can be pushed, and the liquidations that follow are not bad luck, they are a design flaw being collected on. Serious protocols average across sources, weight by depth, and add delays or deviation limits so a single candle cannot rewrite reality. When I size a position in any DeFi system, I spend more time on where the price comes from than on the fee schedule. Flash-loan attacks are, most of the time, oracle attacks wearing a costume.
Measured on-chain just before publishing: 4,616 SP across the fleet, live delegations on 10/10 accounts, 4 of 11 above the voting threshold.
If one number surprises you, the RPC read is one URL away.
Your piece "Oracles: the hard part of DeFi is not the math" stopped me, specifically the part with 4,616 SP.
All three chains descend from the same Graphene codebase, which is why the same keys, the same operation names, and the same account names work across them.
Would be curious how this looks a week from now.