“Big Data” Business Strategy for ScammersPosted: June 25, 2012
A terrific paper by Cormac Herley, Microsoft Research, came out entitled, “Why do Nigerian Scammers Say There are from Nigeria.” It turns out that 51% of scam emails mention Nigeria as the source of funds. Given that “Nigerian scammer” now make it regularly into joke punch-lines, why in the world would scammer continue to identify themselves in this way? The paper was mentioned in a news item here, if you want the executive summary version but, really, I can’t imagine readers of this blog not finding the actual paper worthwhile and fun (it contains a terrific little model of scamming).
In a nutshell, the number of people who are gullible enough to fall for an online scam is tiny compared to the population that has to be sampled. This creates a huge false positive problem, that is, people who respond in some way and, hence, require an expenditure of scammer resources but who ultimately do not follow follow through on being duped.
As the author explains, in these situations, false positives (people identified as viable marks but who do not ultimately fall for the scam) must be balanced against false negatives (people who would fall for the scam but who are not targeted by the scammer). Since targeting is essentailly costless, the main concern of scammers is the false positive: someone who responds to an initial email with replies, phone calls, etc. – that require scammer resources to field – but who eventually fails to take the bait. Apparently, it does not take too many false positives before the scam becomes unprofitable. What makes this problem a serious issue is that the size of the vulnerable population relative to the population that is sampled (i.e., with an initial email) is minuscule.
Scammer solution? Give every possible hint – including self-identifying yourself as being from Nigeria – that you are a stereotypical scammer without actually saying so. Anyone replying to such an offer must be incredibly naive and uninformed (to say the least). False positives under this strategy drop considerably!
UPDATE: Josh Gans was blogging about this last week over at Digitopoly. He’s not convinced of the explanation though. To the extent there are “vigilante” types who are willing to expend resources to mess with scammers, the Easy-ID strategy could incur additional costs. As an interesting side note, in discussing this with Josh, he at one point suggested the idea that when legit firms come across scammers, they should counterattack by flooding them with, e.g., millions of fake/worthless credit card numbers (setting of something like a false positive atom bomb). Just one snag: US laws protect scammers from these kinds of malicious attacks.