Just finished reading your thread on testnets. The cost-to-reward angle makes sense, but I'd be curious how you filter for projects that actually ship mainnet rewards versus the ones that stay in perpetual test phase.