What breaks first when settling P2P skill-based matches across third-party games?

I’ve been exploring a narrow piece of backend middleware for settling peer-to-peer, skill-based matches played inside third-party games.

The scope is intentionally limited:

• no matchmaking
• no odds
• no payment custody
• no game client integration

The system just:

• locks match terms once both players accept
• accepts result submissions
• resolves disputes (dual confirmation or evidence-based review)
• outputs a deterministic settlement decision with an audit trail (“who should be paid and why”)

On paper, this seems straightforward. In practice, very few platforms offer anything like it.

For folks who’ve worked on marketplaces, gaming platforms, payments, or trust & safety: what actually breaks first at scale? Disputes overwhelming ops? Fraud vectors? Payment rail constraints? Regulatory risk? Something else entirely?

I’m explicitly looking for failure modes, not validation.