William Hung: Love Him or Hate Him

Today we released a new beta feature on Summize that let's you quickly see the products people love, hate, and disagree on.

Let's jump right into an example.  First, search for rock music cds:

Which retrieves a list 529,321 reviews on over 20,000 products!  That's a lot of reviews to read if we're just trying to find some new music to check out.

Now notice the beta feature in the right-hand column labeled "Read enough reviews?". Clicking "what people love" quickly re-orders the products to show you just the rock CDs that people love:

The majority greenness of the snips is a conspicuous indicator that most people agree with each other about the high quality of this music — other sites would give these CDs "five stars".  Conversely, if you click "what people hate" you will see the music everyone generally agrees is bad (i.e., "one star").

But check out the snippet for OK Computer in the screenshot above: it says, "people are all different and have different likes and opinions".  For many types of products, such as music, what the "crowd" thinks may not reflect your own tastes, or the product features that are most important to you.

So click on "what people disagree on":

The snips say it all: you can immediately see that user opinion is divided on these products, with William Hung's album topping the list as the CD people disagree about the most.  Other sites would average out this polarizing love and hate to arrive at an average "three star" rating, and bury the CD far down in the search results.

On Summize, however, we highlight this CD as a quintessential example of why we write reviews in the first place: to express our individual opinions. Disagreement does not make a product average; rather, products that foment the crowd's disagreement are often the products that we are most passionate about as individuals.  We predict you will either love or hate William Hung's... music — only 8% of people said it was just "OK".

This feature is still in beta (meaning we get it wrong sometimes), but as disagreement is such an essential ingredient of user reviews, we felt our site would not be complete without it.  Please give us your feedback as we continue to improve.

Comments

[Greg1]

Highlighting the "disagreements" is quite cool - but at first glance I'm not sure how I'd use it to find new music.

America can't agree about Ashlee Simpson's Autobiography? Yeah, I immediately see what camp I belong to. As you say, there's wisdom in the masses, but there's a lot of delusion as well. :)

I wonder if there is some cool way to correlate the differences of opinions to view products which are "clustered together."

Love the site.

[Greg Pass]

Greg1 -- thanks for the feedback.

Your wonderment about constructively correlating camps is prognostic. When the masses disagree, we then wonder which of the mini-masses we most trust -- whose opinions we most believe. Expect to see more along these lines...

[ejo]

William hung is an inspiration to us all, just check out this article

http://www.entrepreneurjoe.com/2007/08/21/the-great-william-hung-of-american-idol-fame/

[Pete]

What is the difference between the Amazon histogram and the summize heatmap? According to Amazon, the love/hate breakdown for SICP is roughly 53% vs. 34%. Amazon has 153 ratings for SICP, with a sharp split between 1's and 5's:

http://www.amazon.com/review/product/0262510871/ref=cm_cr_dp_all_summary?%5Fencoding=UTF8&showViewpoints=1&sortBy=bySubmissionDateDescending

Summize has 155 ratings (153 user reviews from Amazon) and is labeling most of these as favorable (67% vs 11%):

http://www.summize.com/product/structure-and-interpretation-of-computer-programs---2nd-edition--mit-electrical-engineering-and-computer-science-/paperback?rating=great