The problem is that from GitHub's perspective, or that of GitHub's ISP, this would do little to avert the attack. That's because the "attackers" in this case appear to be users outside of China's great firewall who have accessed a service inside of China.

Consequently, it is primarily Chinese-speaking users worldwide and web services that provide Chinese language services (and would be likely to use Baidu's or another Chinese service's scripts) that end up looking like the attackers. In this case, it appears most of the requests came from Taiwan, Hong Kong, and the US.

China could of course filter this further. With only 1.75% of requests being turned into DDOS attacks, they could throttle this up to ~100% and filter the IPs they target to, say, only US IP addresses. Now the attackers appear to be regular US citizens.

What will you do then? Black-hole the US? Beg consumer ISPs to black-hole China?