When it comes to Lunch.com there are some pretty awesome things about it that make it a unique site. It's far more interesting because the site is very much geared to bringing people together. One of the best innovations the site has incorporated is the similarity network. Through this network you can see how similar you are to other Lunch.com members. It's great in terms of discovering what two people may like or dislike and what they can't agree on. There's something quite unique about it. It's one of those things that just about every other website doesn't have. It's also what really makes it unique for Lunch.
The first thing of note with the similarity ratings is that it's based on what you and whoever you're comparing your similarity to have rated. The more you rate, the more accurate the system is. If you both happen to rate several of the same things, then the similarity network actually works out just fine. If your accuracy is "High" within the similarity network then your similarity rating (which is represented by a percentage) is formulated. It will be formulated whether or not you have a "High" "Medium" or "Low" accuracy. The only way to improve your accuracy and your similarity rating is to rate more things. Particularly within the areas where you know you can rate more.
The interesting thing about Lunch is that you are allowed to rate topics without actually reviewing them. Whether micro or full blown. So even if you can't think of what to say, you can still at least rate a particular datapoint. This is where the similarity network comes in. The ratings, and it's also where things with the similarity network can get just a little wonky, and we'll talk about that shortly. First, let's talk about the strengths of the similarity network.
Within each profile photo on someone's review or page you see a number in the corner. A percentage number. This is how similar you are to that particular person. You can then click on that number where it will take you to a page filled with tags. These tags are used to define the datapoints. You can click on any of these tags and view the ratings that you and whoever you are comparing to have rated. This also allows you to see just what they rated and how. The best part about this is you may say to yourself, "Oh, wow, I haven't rated that yet!" Let's face it... rating stuff is just as fun as writing a review itself. So it isn't just about seeing how "similar" you are to someone else, it's also about discovering what you just might have an interest in and what you may not. For example, I can easily go into any friends similarity, look at a tag where we agree and see what I haven't rated, but what they have and perhaps I might find curoisity and decide to explore it.
Finding similar interest does help to broaden your own in many ways. It's hard to admit, but when you see someone who shares say... a 92% similarity with you... it's hard not to go digging in and see what he or she has rated but you haven't. It's hard not to explore then and see where his interest lie that aren't your own. You might see and think that perhaps if he likes something you've NEVER even heard of... you just might like it too. And that's awesome and brilliant.
It doesn't work so well if the two of you haven't rated things which share a similar tag. For example, if I rated more food items I might improve my similarity with @devora. If I rated more foreign films I might improve my similarity with @woopak_the_thrill. In turn, if some of you rated more video games you might improve your similarity with me. If you rated more movies you might also improve your similarity with me. It's an interesting system. And your accuracy will be tracked for specific tags.
The only real detraction from the similarity network is how each person as an individual operates. @Scotman's idea of a +4 may be very different from my idea of a +4. Unfortunately the similarity network can't really compensate for that. The other thing that the similarity rating system can't really compensate for regarding this is that some people rate harsher than others. When I first came to Lunch.com I was not afraid to go negative and now it takes a lot for me to go negative. In short my +1 might actually mean the same as someone else's -5. In your similarity network this setup will come up as: "Areas where you don't agree." But... you actually might agree, just one person didn't go as harsh as another. So it's probably something where you agree, but it won't come up in your similarity network that way. It's not exactly something a computer can compensate for. It doesn't make as big of a difference in your accuracy as you might think, but it can be interesting to find that while you "don't agree," on something you'll talk to the person and find out you actually do agree, just one person decided they'd be harsher while you might not have. The similarity network can't compensate for how one person looks at one particular rating. As I said a +4 to me might be someone else's +5 or even someone else's +3. A +2 for me might be somsone's -2. Such as the case when given a point scale in any fashion. Even in the comments of one of my reviews I got a comment that said something along the lines of, "Your review reads more like a +4 than a +3."
It's really the only thing about the similarity network that might be a little wonky. I have a couple of friends where apparently we agree on certain things but the similarity network tags show otherwise. So it's interesting, but you still get some fairly high percentage ratings where you are similar. It's just that you can't compensate for what ones definition of "Very Good," versus "Great" means. You can't compensate for what someone's rating means.
On the other hand, you can make sure that the difference between a +2 and a +3 isn't astronomical. And that's where the Similarity Network does work to a good degree. If you give something a +3 and someone else gives it a +4... you go down as being similar. Although a +2 versus a +4 may not go down as such.
Nevertheless the similarity does work fairly well. It's not designed so that you'll be 100% similar it's designed so that you are able to find common ground with other members on the site. Common ground isn't "exact same ground." In short, even if you see similarity as being rather low... say... 60% that's probably not as bad as you think. It may mean you have little in common... but look at what you do. You can always build off of those particular interest. I've run into a few members where we disagree in one particular area, but happen to find a lot of common ground in others. So rather than looking at say 65% with some kind of questioning grimace, instead I decided to look and see where exactly that similarity came from... not where the differences came from. It's unrealistic to think people with low similarities will disagree on everything... and it's even more unrealistic to think that those with high similarity ratings will agree on everything.
You're searching for common ground... not a date. The similarity network really drives that home. If you begin digging and searching for commong round instead of worrying about where you may differ you'll discover that you may have more in common with other members than you originally thought. But even better? If you want to find out what someone's other interest are, there is actually an invitation on your similarity comparisons page where you can invite them to rate more so that you can find out how you stack up otherwise. You're allowed to stick in your own personal message too. There are many ways to improve your similarity and the fact that you can go in and see specifically where you don't agree is actually pretty helpful.
It's just that you need to get a personal touch on who it is you're also comparing to. Like I said, ones definition of what a +4 means could be different from someone else's. So it isn't about that percentage, it's also about putting that personal touch in there as well.
The point of the similarity network isn't to be as similar as you can possibly be, but to be able to dig in and find out just where those similarities lie. It's great to improve it when you can, but being similar isn't as important as looking at where it is you do find common ground.
What did you think of this review?
Fun to Read
About the reviewer
Sean A. Rhodes (Sean_Rhodes)
I'm a more analytical person. I believe that the purpose of the review is not for me to give you my opinion but for me to give you an analysis and help you decide if you want to get it. If you reading … more
Consider the Source
Use Trust Points to see how much you can rely on this review.
The Lunch Similarity Network points you to the best information through the right people. Your Similarity Network is a group of people who most closely share your likes, dislikes, and interests in the Lunch community. It's the basic idea behind Lunch – that it's more useful to get a few trusted reviews from people who share your interests than thousands of reviews from random, unaccountable sources. Through people in your Similarity Network, you'll find new things you'll like or updates on existing interests.
Lunch believes that you'll have the most meaningful exchanges with people who get your sense of humor, taste in books, love of snowboarding or penchant for Pac-Man. The more you rate, review, and contribute information to Lunch, the better we can connect you to people and information you'll like.