Honestly I bet it's a learning algorithm that they used to identify bots, rather than something a human decided.
Spam detection algorithms usually have a training set. One of the features could have been "uses_feature_x", which, according to the training set, is known to have a high probability of being a spam bot (because humans rarely use those features)
Spam detection algorithms usually have a training set. One of the features could have been "uses_feature_x", which, according to the training set, is known to have a high probability of being a spam bot (because humans rarely use those features)