I've fought spam in other contexts and found that to be true. A change stopped spam until spammers learned what the change was, which made them adapt. The adaption might not be perfect, but it would happen, so prolonging the blocked period was essential. If one can do something that's effective and that the spammers don't understand for months, that's
great. If they can spam for days and then blocked for a longer period and that repeats, they'll go away and attack an easier target instead (well, some of them will look back later).
Let me try an argument other than "it's been my experience", though.
Suppose that it's not true. In that case, the anti-spam filter provides a publicly visible definition of spam in its code, and it's one that isn't grounded in the recipients' opinions. If the filter is to be effective, spammers must not be able to adapt their behaviour in a way that users consider spam but isn't according to the definition. Many people have tried to provide such a definition since 1995 and AIUI all have failed. I think the persisten failure indicates that it's extremely unlikely that one can be found, and therefore it's extremely unlikely that a publicly visible spam filter can be as effective as one that can be updated in secret.
Let me try an argument other than "it's been my experience", though.
Suppose that it's not true. In that case, the anti-spam filter provides a publicly visible definition of spam in its code, and it's one that isn't grounded in the recipients' opinions. If the filter is to be effective, spammers must not be able to adapt their behaviour in a way that users consider spam but isn't according to the definition. Many people have tried to provide such a definition since 1995 and AIUI all have failed. I think the persisten failure indicates that it's extremely unlikely that one can be found, and therefore it's extremely unlikely that a publicly visible spam filter can be as effective as one that can be updated in secret.