Possible venues of defense:

	- Authentication. Forces the identity of the spammers so that
	  they can be more easily tracked.
		See CMTP.

	- Flood prevention. Counters the "bulk" aspect.
		- Block based on true identity. What We Want. Will
			cause problems with mailing lists.
		- Block based on CPU expense. Can be thwarted by
			DDoS and in general, control of multiple
			computers; not equal for all
		- Block based on time. Parallelizable.
		- Block based on content. True hashes could easily
		  be prevented by varying the amount of space or
		  whatnot in the mail. Maybe some sort of fluid
		  blocking that slows down more the more similar a
		  message is to one sent in the past? Would require
		  immense storage.
		- Block based on time and IP. Still can't prevent
			DDoS, or any attack with multiple sources
			or multiple targets.
		Look around for general DDoS prevention ideas or
		papers. A fully fledged system would require some
		sort of peer to peer structure. Such a structure
		could handle all the other problems too.

	- Leges and automatic TOS countering. 
		Works on the meaning aspect. Also includes things like
		tracking down the originators easily. Terms Of Service
		is kind of an intra-ISP law.
		Once authentication is in place, this can be automated
		to a much greater degree. Creating standardized spam
		report formats is also of use.

	- Content filtering. Bayesian works fairly well, but we may
		be able to squeeze more out of it by Markov models
		(don't know how they work) or NNs, or even multi-gram
		Bayesian.
		The classifier should also distinguish higher order
		information, like the difference between HTML tag
		options, HTML comments, and text. This could be
		programmed manually. [blurring heuristic and content filtering?]
		Could possibly be backed up with letter statistical
		analysis tools, or as Minsky put it, mechanisms that
		deal with "numerous effects of small value".