When I was in London I had a great time learning all about the Scamalytics and Introanalytics services from Nick Tsinonis. Its great to see them getting traction in the dating industry. Cupid PLC, White Label Dating and now Dating Factory are using the Recsys platform. More info.

Nick wrote a comment to my original post about their the partnership with Cupid LLC.

A few Friday morning pre-coffee thoughts on dating site scammers…

I begrudgingly respect talented scammers that know how to get people to give up money. Identifying scammers is a lot easier than matching people, although as Nick will tell you, scammer identification gets complicated when you start working with large-scale sites, especially when you have to back-trace through hundreds of thousands of profiles to regressively identify scammers.

What is the definition of scammer? Being clueless about appropriate behavior on a dating site or being an outright jerk is not the same as being a scammer. What is the behavior threshold for considering a user a scammer? How does a particular company decide to define a scammer in terms of behavior? Community guidelines differ from site to site. Serious dating sites may have more stringent rules for identifying scammers than casual or adult dating sites.

Advanced anti-scam, or should we call it Fake Profile Identification Systems, work something like this. Create a set of base rules like: Sends more than 100 emails a day, uses certain phrases in emails, gets flagged N number of times, IP address of originating messages. and lots of other tests, you run everyone through the system, see which tests are most valuable and return the least false positives, keep the good ones and continue to add new tests as scammers evolve, adding to the corpus (or concordance?) of trigger phrases and actions over time.

The first thing scammers will tell you is the simpler the scam, the better the return. Those Nigerian 419 email scams you see like “I am Mr. Borkorto from Liberia and I need to transfer $7,689.000,00 to your bank account” are worded exactly so a certain type of gullible person will respond to the email. It doesn’t matter that the majority of people would throw that email in the spam filter immediately. It’s the 0.00023% of people who will fall for it that the scammers are after.

Often, a scammer emails one or more people the exact same email content (novice) or content that is slightly modified (experts). A threshold is associated with the volume of emails and the similarity quotient between a set of emails. Good scammers evolve their “pitch” based on results accordingly. This arms-race is what fascinates me about scammers. The same could be said for marketers.

On the flip side of Scamalytics, I’ve never heard of a cloud-based scamming service, call it White Label Scammers. Probably exists somewhere, talk about a huge business opportunity.

Anti-scam practices in the dating industry are becoming increasingly necessary as the industry becomes more of a target for scammers. Here’s to the continued evolution of best practices focusing on keeping databases clean and people safe.