The first time I read a Web3 game post-mortem that blamed bot farming, I assumed it was a technical failure. Then I read a few more and started seeing a different pattern. It wasn't that the bots were smarter than the team. It was that the reward system couldn't tell the difference between someone playing and someone extracting.
Pixels ran straight into this. After migrating to Ronin in late 2023, the game exploded from 4,000 to 180,000 daily active users in two days. From the outside that looked like a breakthrough. From the inside, it was when the pressure started compressing everything at once.
The design wasn't broken. The problem was subtler: the reward system was broadcasting the same signal to two completely different groups of people. Players looking for an experience, and bots looking for arbitrage. Same action, same reward, no way to distinguish intent from behavior. I'd call that a blind signal problem. And blind signals don't show their damage immediately. The economy bleeds quietly until the pattern becomes undeniable. By the time it was, $BERRY couldn't be saved. The team scrapped it entirely and rebuilt from scratch.
Stacked came out of the question that followed that rebuild: how does a reward system actually recognize the difference between a player who will stay and one who just came to extract? That answer doesn't live in a whitepaper. It lives in 200 million transactions, in fraud patterns logged from real attacks, in the behavioral data that only accumulates when you've been through enough cycles of growth and collapse to know what you're looking at.
Blind signal is an industry-wide problem, not a Pixels-specific one. Most reward systems being built right now are making the same mistake, just at a scale where the consequences aren't visible yet.
The first wave of P2E didn't fail because it ran out of tokens. It failed because it never knew who it was rewarding.