That’s surprising, as the OpenClaw installation makes it pretty difficult to run without auth and explicit device pairing (I don’t even know if that’s possible).
I agree—it looks like the OP didn't provide any sources for these numbers either. That's why I would have hoped that the original maintainer had a better set of metrics to dispute them. It doesn't seem like he does though :(
The problem is that a lot of users of OpenClaw use a chatbot to set it up for them so it has a habit of killing safety features if it runs into roadblocks due to user requests. This makes installations super heterogeneous.