It's the difference between iterated and one-time games.
Shum has a decision to Publish or Not Publish (P or NP). CCP has the decision to Retaliate or Not Retaliate (R or NR). In a one-time game, Shun chooses P and CCP chooses NR, because P/NR > P/R for CCP.
A one-time game may be a good model if there's only one thing you wish to keep out of the public eye (e.g. photos of Barbara Streisand's house). But CCP will play this game many times: establishing that they're able and willing to harm their critics, even at cost to themselves, has value. It encourages future critics to never publish in the first place.
Shum has a decision to Publish or Not Publish (P or NP). CCP has the decision to Retaliate or Not Retaliate (R or NR). In a one-time game, Shun chooses P and CCP chooses NR, because P/NR > P/R for CCP.
A one-time game may be a good model if there's only one thing you wish to keep out of the public eye (e.g. photos of Barbara Streisand's house). But CCP will play this game many times: establishing that they're able and willing to harm their critics, even at cost to themselves, has value. It encourages future critics to never publish in the first place.