How Quora Can Avoid Becoming A Popularity Contest

The danger of Quora seeing a drop in quality and becoming a popularity contest has been in the spotlight recently.


Quora is like voting for high school queen and king. Popularity trumps intelligence and wisdom.less than a minute ago via Twitter for iPhone

In my previous post on Quora, I said that it would be difficult for a few influencers to dominate the site. The process of admins moderating the site has apparently not scaled well, compared to the overall spike in usage. See this Quora thread for a few examples of wrongly upvoted answers. This specific threat to the quality of Quora is also discussed in this thread.

Is it possible for Quora to avoid becoming a popularity contest? I certainly think so. Quora can utilize a Person Rank algorithm and completely revamp its voting mechanism to get rid of this problem.

Person Rank (User Quality) Algorithm

In response to the challenges it faces as it scales up, Quora co-founder Charlie Cheever posted about some of the initiatives they have planned. An interesting element is the development of an algorithm to assess user quality (or Person Rank):

We’re developing an algorithm to determine user quality.  The algorithm is somewhat similar to PageRank but since people are different from pages on the web and the signals that are available on Quora are different from those on the web, it’s not exactly the same problem.  We’ll use this to help decide what to show in feeds, when to send notifications, and how to rank answers.

Signals to Leverage for Person Rank Algorithm

What signals should Quora use while developing this algorithm? In my detailed response to this question on Quora, I have listed several signals grouped under these categories:

  • Positive Signals for Community Contributions
  • Signals for Content Quality
  • Positive Signals for User-to-User Behavior
  • Negative Signals for User-to-User Behavior
  • Positive Signals for Overall Quora Usage

Of these, only the User-to-User Behavior is a group of signals that reflect popularity. The other signals are reflective purely of quality of contribution to Quora. (My answer also highlights those signals that are difficult to game or fake). These and other such signals can be appropriately weighted to arrive at a representative Person Rank for user quality.

New Voting Mechanism for Answers

Once a Person Rank score is in place, the answer ranking need not be based simply upon the number of upvotes, but should also take into account the user quality of who is voting.

Thus, a popular influencer’s answer may be upvoted by his or her significantly large following, but the answer may still not rank at the top because fewer other people with higher user quality have upvoted another answer.

Quora need not display the number of votes for each answer, and can replace it with a “Quality Score”. This will also reinforce the emphasis on quality, not quantity, in Quora’s UI.

Only time will tell if Quora is able to scale well keeping pace with mainstream user adoption. With approaches like these, I still think Quora has a good chance for maintaining content quality and not degrade to the levels of a popularity contest.

This entry was posted in Social Web and tagged , . Bookmark the permalink.
  • http://twitter.com/semilshah Semil Shah

    To me, this raises a few interesting questions:
    1. What type of NLP technology will monitor quality of actual content contributions?
    2. Will focusing on this type of ranking either create a backlash among users who feel slighted, or focus energy away from the site’s core strength?

  • http://www.skepticgeek.com Mahendra

    Semil,

    1. I don’t think NLP is required. An algorithm similar to Wikipedia can work (http://wikitrust.soe.ucsc.edu/an-overview-of-the-algorithms) (I have linked to Seb’s answer on existing algorithms from my answer on Quora)
    2. Backlash is inevitable for major changes like these, but I think this would focus energy towards the site’s core strength – quality.

  • http://www.louisgray.com/live/ Louis Gray

    What’s needed for all these streams is not just popularity, or even a single score for influence, but personal relevance. :)

  • http://www.thenetworkgarden.com hypermark

    Excellent analysis. I think that the key to Quora not losing its mojo is specifically anchoring on the User and the Quality of their answers. Quality can be defined according to multiple parameters: crispness of answer; qualitative uniqueness; virality; authority; related contexts, etc. The better job Quora does of rewarding users for committing earnest intent, the clearer path they have to building something really special.

  • http://www.thenetworkgarden.com hypermark

    Excellent analysis. I think that the key to Quora not losing its mojo is specifically anchoring on the User and the Quality of their answers. Quality can be defined according to multiple parameters: crispness of answer; qualitative uniqueness; virality; authority; related contexts, etc. The better job Quora does of rewarding users for committing earnest intent, the clearer path they have to building something really special.

  • http://www.skepticgeek.com Mahendra

    There are two different elements here:
    1. What to show in the stream / feed
    2. How to rank answers

    I think personal relevance is much needed for #1, not so sure about #2.

  • http://www.louisgray.com/live/ Louis Gray

    At least #1. Let’s see what we can do.

  • Tony Charles

    Hi!

    I was wondering if you were currently accepting advertising on your site?
    We are looking for sites such as yours to palce contextual and relevant advertising on.
    We already work with a large number of happy publishers with very positive feedback.
    Please reply to this email and let me know that you are interested and I will get some
    more information across to you.

    Kind regards,

    Tony Charles