From the age of Amazon, CNET and eBay, many people become comfortable in evaluating things in absolute terms and relying on plethora of reviews for peer experiences with products or services. Five-star system became de-facto standard in rating. Why is it broken now?
What’s the point of 5* rating system when 90% of ratings fall into 3*-4* category?
Five star rating? History.
If I am cranky and grumpy, I can easily award my newest Amazon purchase with 3* instead of 4*, although the truth may lie somewhere in between. In another mood, I could as easily give it 4*. Obvious problem? Not consistent.
Ratings are extremely error prone and inconsistent, but not if they’re comparable with something else! If every time I was prompted to do a rating for a product I was prompted to compare with similar products which is better or worse – that would make much more sense! Product A and product B have both 3*. Perhaps former has more reviewers, but is that something that can be actionable for me as a buyer? On the other hand, if I knew that Product A is better than product B, it would make a very consistent rating, and therefore much more actionable for potential buyer! Not to mention element of fun (remember hotornot.com?).
Of course, this paradigm is very cumbersome, as you need a reference point for each item, and in lots of ways it just doesn’t make much sense. When I want to purchase a bottle of wine, and use any iPhone app available out there, it will give me shitty five star rating system, and with 90% of wines falling into 3* or 4* I might as well go to fortune teller instead of using app for purchase decision. Consequently, comparison would make sense only if I’m comparing two bottles in front of me, and still – they need a solid reference system outside the 5* scope. Tough cookie.
Recommendation – a 5 star rating
What goes around this problem is recommendation system. Recommendation is basically a 5* rating, and doesn’t need a reference point, thus people love recommending stuff.
Unfortunatelly, what really ruined that – is when Facebook became platform and not “just” social network. It turned out people are more likely to engage with content or product if few of their friends are also engaged with it. Then everybody started blabbing about dopamine and addictions, and as a result, you can’t browse anymore normally without being harassed by “your friends read this” or “your friends liked that” on every single website out there.
My Facebook friend likes a book? Good for him.
There’s one thing I’m sure about. Once developers start using verbs (actions) introduced in Facebook’s social graph on F8, I won’t be able to as much as look at tonight’s cable TV program without something jumping out on me with “John, Peter and Susanna will watch this show”. And just imagine yourself trying to pick a bottle on wine.com with “Denise and Michelle bought this wine!” popping in front of your eyes. Yay. Great for them. But why do I care?!
“Hot” startups like stamped.com attempt to rationale this problem by saying “What does 3 stars from 70 strangers mean? Is that good or bad? Can I trust this person’s opinion?“. So they introduce “five star only” rating (a.k.a. stamp) from your network of friends. On the other hand, oink.com improves the resolution of location-based reviews to specific item level, which helps people browse through reviews of individual in a restaurant, instead of reading 4sq tips.
Experts to the Rescue
One thing that strikes me is how the hell are both Stamped and Oink gonna help me choose a meal in restaurant just because a friend of mine rated it well (stamped it) or someone recommended Cesar salad? Why does his/her taste matters? Just because she’s a friend? I mean… I have lots of friends who can chew a chicken head and be happy with it, shoud I take into consideration their recommendation about a salad or steak? And why is their “tip” is more valuable than regular FourSquare tip? I don’t think this will solve any problems, TBH. It will just impose different level of noise as we have today.
How can you go with expert-on-something in social network way, so you can get both relevant and scalable?
Look what StackOverflow did, and then Quora afterwards? They don’t give a flying fuck about friends and social network relations. All that matters is how competent are you on specific topic. And who puts you in charge? People who’s trust you “buy” with relevant opinions on specific topics. You’re not an expert person, you’re just expert on specific topics.
Facebook – Recommendations from your social network, but done the right way!
Ok, we’re clear that I don’t buy into Facebook’s vision of finding the right thing for me. My network of friends means absolutely nothing, unless I could somehow tag my connections with specific attributes.
So if I could tag somehow Michelle and Denise (call it “badge” or attribute) with “wine/spirits”, then I guess their recommendation of a bottle I’m evaluating is very much worth something, even if we do not have similar taste. Or I can “tag” John as a my plumber and softly recommend him to anyone in need of a plumber, although I don’t want him as my friend, nor do I care about what books he reads!
In pretty much the same fashion Facebook introduced verbs to connect people with objects, they’re in desperate need of soft connections between people.
Now comes the good part – with exposing those attributes (call them “soft connections”) to second level connections (friends of friends) I will exponentially increase my network to huge number of manually approved “experts” on various topics that will allow me to finally benefit from social-network recommendations! Bingo, we got both relevant and scalable!
I mean… this one would be a killer, right :) Unfortunately, if you’re thinking what I’m thinking – from perspective of a potential startup, it will be a very, very challenging to implemented such service/platform on top of Facebook graph. Of course, if anybody feels like funding this venture, drop me a line :))