Externalities, asymmetric information, game theory, network effects, innovation diffusion:
understanding the countermeasures arrayed against spam, involves a
venerable catalog of analysis tools in the economics of information.
Since this is just a blog post, not a book, I'll try and keep it
short and sweet.
externalities
Spam is the environmental pollution of the knowledge economy.
Everybody's welfare is degraded by a few, who make a buck.
Most of the cost of spam is borne by
internet service providers (bandwith, processing) and
end users (lost productivity, computer crime).
Spammers face a skewed price mechanism: their balance
sheets show all the profits, but only a fraction of the real costs
of their activity. Predictably, overproduction results.
asymmetric information
Spam shows up at an ISP's mail servers in a situation of asymmetrical
information: the sender knows it's spam, the mail server is unsure whether
this is spam, or maybe a legitimate email.
Now it gets interesting. The ISP uses signals to overcome it's information
disadvantage.
First of all, the ISP checks the sender's reputation against
public and internal blacklists. Is this a known spammer? Shut the door.
Second, greylisting kicks in (see earlier post). If the sender
is neither blacklisted nor whitelisted, we're going to extract a signal by
initially refusing to process the email. If the sender retries,
that signals he's more likely to be a legitimate mail sender.
game theory
It would be very easy to fake this signal for a spammer - he just has
to submit his emails more than once. Yet he doesn't. It is easier
to just target another victim mail server that has no greylisting filter. Yet.
This introduces a cost differential between those ISPs that have greylisting,
and those that don't. The latter spend more resources on content scanning.
After a while, they'll introduce greylisting too, to cut down on spam.
Ultimately, when a sufficient proportion of ISPs uses greylisting,
the spammers will have to learn how to fake a "sincerity" signal
and reduce the effectiveness of greylisting.
innovation diffusion
As shown above, the effectiveness of a specific ISP's anti-spam
countermeasures is dependent on whether this ISP is
ahead of the curve or a laggard in introducing countermeasures.
The continuous battle between improved anti-spam measures and new
spamming tricks, creates a never ending pressure on any ISP to
keep improving the anti-spam infrastructure.
network effects
Network externalities arise from the cooperation between ISPs to combat spam.
I'll refrain from going into networking externalities and innovation diffusion
in an open source context, since that's a complete topic in it's own right.
Suffice to notice, in this context, that being able to do
# apt-get install postfix-policyd
reduces the cost of deploying a greylisting anti-spam layer
to less than one man-hour, while creating
significant
benefits in terms of reduced systems resource consumption
and reduced spam pollution of customer's inboxes.