PoW Bot Defense Is the New Normal, But Nobody Loves It

A post about shipping WebAssembly support in Anubis, a proof-of-work bot defense tool, generated a real debate about whether PoW is actually a good solution to the bot problem. The thread noted that the compute burned is entirely wasted, unlike crypto where at least a token exists at the end. Someone linked a browser extension that can bust PoW challenges automatically, which undercuts the premise. A counterproposal came up: pre-computing PoW credits so users aren't stuck waiting mid-task.

The deeper frustration in the thread is that the bot problem has gotten materially worse since LLMs made it cheap to generate plausible content and fill out forms at scale. PoW is the current community answer, but it punishes real users with slow hardware disproportionately. The fact that Anubis took a year to ship WebAssembly support, and that this is being followed closely, tells you how much demand exists for any viable bot defense that doesn't rely on Cloudflare.

This connects to the Cloudflare threads that keep showing up: Spain's football streaming blocks are taking down Docker Hub as collateral damage, and Cloudflare's bot detection is increasingly blocking legitimate users with non-mainstream browsers. The dependency on Cloudflare as the default bot defense layer is itself becoming an infrastructure risk.