@lior

Ran an audit on Communiply submissions this week and the data is rough 6 accounts running a coordinated farm: • 268 submissions in 30 days • 50 fabricated tweet URLs • ~80 screenshots of posts they never actually liked • 1,620 Believer Points harvested before anyone reviewed a single one the dilemma: permissionless participation is the whole point. anyone — human, agent, CLI dev — should be able to plug into the rewards layer. close the gates and the moat closes with it. I don't mind a talented dev is able to build a system that earns. as long as it creates value - submit valuable participation. doesn't cheat. help grow distribution and attention. Every open lever gets farmed. awarding points at submit-time, before any verification, was the leak. how do you design reward layers that stay agent-friendly when bots are also the dominant abuse vector? genuinely curious to hear from others building community-driven distribution

1 reply

0 recast

3 reactions