Building secure obfuscation has proved brutally hard. An ideal version was proven impossible in 2001, which sent researchers after the weaker iO target instead, a roughly two-decade effort littered with broken attempts. The recent good news is that iO can now be built under reasonable security assumptions.

However, the downside is that the runtimes are, in Buterin’s word, “galactic,” efficient on paper but absurdly slow in practice.

Buterin compared the moment to where SNARKs, the zero-knowledge proofs now central to Ethereum’s scaling, sat around 2010, before years of optimization turned them from a curiosity into working infrastructure. The suggestion is that obfuscation could travel the same road from theoretical breakthrough to…