For example, if their capabilities to identify shadowsocks traffic is not particularly specific, filtering would result in undesirable impact on other traffic. They can also have other out-of-band estimates for the extent of shadowsocks use (presence of the software on seized or searched equiment, observed chatter, informants, etc).
a: build a method of detection and prevention and
b: find and coerce developer to stop improving software,
#b is required assuming the developer(s) is considered to be an above average adversary. When there is no silver bullet solution a cat+mouse game is inevitable. That further increases the value of this action.
#a being done at same time as #b has an effect on the collective behavior of the adversary. I'm sure various members for the RIAA and MPAA are wondering how they could have dealt with "filesharing" in a similar manner during the Napster days. But in the end it only buys you time in a cat+mouse games. meh, im sure there is some sun tzu art of war blah blah somewhere saying the same. more poetically of course.
Hmm... okay, so they defeat shadowsocks by recognizing the packets.
> Then they stopped the development of shadowsocks.
But if they already had shadowsocks beat, why do they make a public show of shutting it down?
Sounds more like they recognize that they don't have the GFW technology to defeat shadowsocks on-going development over time. Which suggests all you need is a new developer.