Skip to content

What Is Similarity?

Jeff Schnitzer edited this page Mar 12, 2017 · 1 revision

May 23, 2011

http://www.quora.com/What-is-the-future-of-dating-and-matchmaking-services

I just launched Similarity.com, so you could say my answer is biased. On the other hand, I'm sufficiently convinced to spend money and time building a company around what I believe is the answer to this exact question.

I anticipate two trends:

1) Making effective use of the social graph.

While the exact number of people who meet online is a contested issue, the fact that the overwhelming majority meet offline is not. Whether meeting people through work, or school, or friends, social context is essential to dating. Even going out to bars, men bring their wingmen and girls bring their girlfriends.

Yet nearly every existing online dating website tries to subvert this natural way of things. Online dating must be anonymous! Performed in a secure undisclosed location, under the covers with the lights out! Websites tacitly convey that being single and available is something shameful that should be hidden. From whom?

I think partly this attitude comes from an older generation of users who are genuinely embarrassed by online dating. But it is also heavily reinforced by the need to defend existing business models - i.e., preventing users from contacting each other without payment. Paid dating sites go through a lot of effort to scare you into believing that sharing an email address is somehow more dangerous than sharing a drink with the cute stranger at the end of the bar.

Ironically, popular dating websites are becoming victims of their own success - by legitimizing online dating and bringing it mainstream, the next generation of users don't place the same paramount priority on anonymity and privacy. Oversharing is standard operating procedure, even if proper spelling and grammar is not. These users are ready to bring social context into dating online.

Match and eHarmony's established userbase will never accept this shift, so I believe the time is ripe for a new player to disrupt the industry. The first site that makes effective use of the social graph should be able to carve out a large stake in the younger, more outgoing audience.

It's worth calling out some failures here: OkCupid, despite having a solid product that appeals to a hip audience, is still rooted in the traditional anonymous model. Zoosk, despite having implemented Facebook Connect, doesn't actually use the social graph in any meaningful way. I do not suggest that these sites are commercial failures, just that they aren't approaching the problem in any new or novel way. This is why Zoosk is burning through $30 million of VC buying Google adwords at $5-$7 per click.

At the very minimum, a social dating site should follow Similarity.com's footsteps: Blur the line between your dating profile and your Facebook profile, allow matching to friends-of-friends, and encourage users to find matches for their friends.

2) Empowering the user

I think it's terribly arrogant for sites like eHarmony and Chemistry.com to declare that they will find your perfect match through the power of SCIENCE! It harkens back to the 1970s image of computer dating with punch cards and a big board of blinky lights.

I'm sure that there's plenty of statistical evidence that on average, certain kinds of people tend to produce happier matings with certain other kinds of people. But unless you plan to go on a few thousand dates, these statistical trends aren't meaningful to us as individuals and can't realistically hope to predict how two people will get along. This is why Match.com's primitive filtering is so successful - because after you separate the born-again Christians from the rabid atheists and the teetotalers from the bi-winners, what really works is a gigantic grid of pictures. And plenty of sites (ahem areyouinterested) seem to be very successful with just the the pictures.

It may seem strange that I am so dismissive of matching algorithms when Similarity.com is itself based on algorithmic matching. My complaint is that existing websites do nothing interesting with their algorithms. PhDs can argue about how many collective IQ points were burned to forge my matches, but at the end of the day they're just giving me a handful of profiles that I might or might not be interested in. The primary source of interactivity is waiting until next month for the matches to rollover - great news for the company being paid $60/month, but it leaves me feeling like a tourist.

Here are realistic scenarios that the One True Algorithm folks are blind to: "It's winter - I'd like to date someone that skiis." "My last boyfriend was pretty shy - I'd like to date someone more assertive this time." "I'd like to try dating someone from foreign culture X."

My point is that taste is situational and varies from month to month - ESPECIALLY with younger people who haven't created a blueprint for the exact partner they expect to marry within 72 hours of their profile going live. Rather than condescendingly telling you what you want, an algorithmic matching system should provide knobs and levers which empower you to discover what works and what doesn't. This is a whole lot more fun than a machine that dings and spits out a punch card with your Scientifically Approved Matches.

I can't claim that Similarity.com has gotten this perfectly right out the gate, but it's something we're developing. We already allow more customization of the match results than any other dating site that I'm aware of.

original