Some more context: The marker system existed as a pragmatic early design choice, not as an intentional loophole. Initially, Featured Apps could only earn rewards when their transactions involved Canton Coin (CC) transfers. That excluded a lot of the institutional activity Canton was built for — settling trades in other assets, locking or transferring real-world assets (RWAs), minting non-CC tokens, license renewals, etc. Forcing artificial CC transfers into those workflows just to generate rewards would have added complexity for developers.[](<raw.githubusercontent.com/gl…>) CIP-0047 introduced `FeaturedAppActivityMarker` contracts so apps could self-report “economically meaningful activity” and earn rewards without needing a CC transfer. Governance (via the Tokenomics Committee) was supposed to enforce that markers only went on real value-adding events, not intermediate steps or spam. In practice this proved easy to game and inefficient: - Reward weight was not tightly linked to actual network traffic consumed. - Apps could insert multiple markers or structure transactions to inflate activity counts. - Every marker pulled the DSO (Super Validator organization) into the transaction, bloating size, increasing validation load, and reducing throughput. - Policing what counted as valid required ongoing human review, which doesn’t scale and creates pressure toward more centralized control. CIP-0104 replaced it with automatic measurement of real traffic burn from sequencer and mediator data. That removes the self-reporting surface, cuts the overhead, and ties rewards more directly to actual network usage. The Q3 volume drop is largely the result of that cleanup — the earlier high transaction counts included a material amount of marker-driven activity that the network deliberately reduced.