Some changes to the algorithms that might solve this problem.
1. Increase the threshold of reports that actually marks the thread as spam and hides it.
2. Factor date joined and post count into the power of the Mark as SPAM button. More experienced users will make less judgmental errors and can prevent sockpuppet accounts from abusing the feature.
Of course, we’ll all need to respect the guidelines of the spam feature. No algorithm is foolproof and heavily depends on the participation of users. Quick reminder on how to use the SPAM button (and how *not* to use it)