Not sure what his current thoughts are on the evil score system but back in 2015 he expressed the following opinion:
<...>
It also seems to show that the evil score system is working as expected: the vast majority of the Internet is not being forced to pay, and in the isolated sections where a registration fee is required, prohibitively-large fees are very rare.
<...>
The system is imperfect but it's pretty difficult to make an automated IP blacklist system more effective without accidentally hitting regular users in the crossfire. There's quite a lot of IP addresses and, as with many other goods, a lot of them are available on the open market (be it as part of a collection of proxies, VPNs, etc.). Unless you blacklist
all of them (which would be a monumental task on its own), people (or bots) will find ways to slip through the cracks.