Recently a geeky girl reader submitted a question about LinkedIn.  LinkedIn is a professional networking site that allows you to connect to and manage your professional contacts online.  It's sort of a modern day Rolodex, if you will.  It's handy in that you can connect directly to those people you know through professional association, like  current or past employers, networking groups and associations, school, conferences, etc.  If you don't have direct information for whomever you wish to add to your network, you can connect with them via a current colleague or friend.  You simply ask your colleague for an introduction through email and the respondent has the option of confirming or denying your request.  A basic/free membership to LinkedIn is, at the very least, an excellent tool for keeping an up-to-date list of your contacts and colleagues.  A professional membership, however, takes networking to a whole new level, giving you greater access to contacts with whom you may only have a peripheral association, and letting you take advantage of more advanced features like tools for recruiting and job postings and, on the other end of the equation, job searching.  I know some employers who only use LinkedIn for recruiting and have great luck with the service.  All of that being said, I think most people I know use LinkedIn as a way of automating the maintenance around keeping their contact network current.  Based on what I know of her, I imagine our reader with the question is exactly that type of user - a casual subscriber who maintains her professional contacts list using LinkedIn. 

Our geeky girl reader wondered about the People You May Know feature on LinkedIn.  This feature shows up in your sidebar and recommends possible connections.  Miraculously the recommendations are pretty darn accurate.  Often those people you may know are actually people you DO know.  The accuracy of these recommendations has increased LinkedIn's creepy rating by a factor of ten.  The service recommends people you may have never had any contact with through the website.  And sometimes those recommendations are so obscure AND accurate it borders on mind-blowing.  Our reader wanted to know how in the heck LinkedIn can figure out who we might know and how they can be so darn creepy about it -- so I set out to find the answer.  I searched the web over and, come to find out, the people at LinkedIn know just how creepy and accurate they are and they work hard to keep this secret side of their code, well, secret.  There are a lot of techies and bloggers supposing how they do it out there in the webosphere.  But there are no proven theories about how they get to their lists.  After my own research I did the only thing a self respecting geek girl could do (pffft), I reached out to a member of the geek girls men's auxiliary, one genius engineer and Director of Technology at Clockwork, Matt Gray.  I got way more than I bargained for in that exchange.  Matt sat both Meghan and me down and explained to us the intricacies of graph theory.  I'll never be able to do any of that justice here.   Instead I'll just give you my layman's take on the discussion and invite Matt and other geniuses to comment on this post and correct me if I over-simplify it to the point of being wrong.  Remember though, our goal here on the Geek Girl's Guide is to keep things simple.  And I aim to please. 

In very very basic terms you can think about each one of your connections on LinkedIn as a line -- from you to that person.  If you were to draw lines for every connection you'd have a heck of a graph.  Say one of your friends, we'll call her Rita, knows a guy named Bob.  Rita has a line from her name to Bob's name on her personal (and invisible, hypothetical, stupid but useful for this discussion) graph.  That line in and of itself is just a connection.  But say another of your friends, we'll call her Sarah, also knows Bob.  Suddenly the line from Rita to Bob and Sarah to Bob is bolder in terms of how it relates to your network.  Because not just one, but two of your connections know Bob, which increases the likelihood of you knowing Bob.  Throw another person from your network in there with a connection to Bob and, doggone it, Bob might just be somebody you might know and that's how he shows up on your list.  The theory is that LinkedIn has programmatically harnessed graph theory to weigh your connections in terms of how they relate to everyone in your network.  They automagically layer all of these graphs over each other and make recommendations of people you may know based on the number of people in your network that share connections and the likelihood that those connections should be yours too.  They probably go a step further and factor in the invitations people send outside of their LinkedIn networks.  If two or more of your contacts invite the same person to use LinkedIN it's likely you know that person too.  That's probably how people you know, who aren't currently using LinkedIn, end up on your People You May Know list. 

There you are.  An overly simplistic, but easy-to-understand explanation for why LinkedIn is so creepy in it's accuracy.  Thanks to Matt for his help in laying it all out.  Thanks to geeky reader Ann for the question.  Now go ahead, people.  Comment.  Tell me how dumb I am.  Tell me the right answer.  Tell me how I could have explained it better.  I'm open to all of it.