[JesseK]

I think the restriction would simply force the spammers to make (for example) posts of spam before the PM spam. However, a minimum post threshold can determine the variables for the folloing algorithm, much like fraud detection sensitivity increases for a credit profile when a credit card # is stolen:

For low post-count users, perhaps delay all PM's by [N] seconds to see if multiple PMs from the same user are being sent out. For each new PM sent before the delay expires, increase the delay of all PM's by that user by [N] more seconds. Once a quantity [X] of PM's are sent out by a single user before the delay [N] expires, delay them indefinitely until a mod can visually review them. For PM's sent to multiple users from a 'new' account, put a threshold [Y] of review: e.g. if sent to more than 5 users, it will be delayed until reviewed (4 users would still be delayed via the algorithm described above). Put a disclaimer in the PM process that PM's may be delayed and reviewed if sent en mass.

As a user increases post count, decrease the sensitivity (lower the delays, increase the user threshold) of the variables X, Y and N. In that way, current known accounts (e.g. EricH) are allowed to send multiple [assumed legititmate] PM's without scrutiny while also allowing new legitimate users to do the same once a certain amount of 'trust' is built.

Perhaps it could be a vB feature in the future?

The other metric that could be used is reputation; typically those with 50+ posts have received some sort of reputation, either positive, neutral, or negative (though the actual magnitude of rep shouldn't play into the variables since the concept is just a user authentication algorithm).