Stories
Slash Boxes
Comments

SoylentNews is people

SoylentNews is powered by your submissions, so send in your scoop. Only 16 submissions in the queue.

Submission Preview

Link to Story

OKCupid: "We Experiment on Users, Everyone Does."

Accepted submission by Ethanol-fueled's Sweaty Laptop http://www.okcupid.com at 2014-08-03 00:07:29
Business
The Guardian's report, [theguardian.com] paraphrased:

Following in Facebook's footsteps, [washingtonpost.com] Dating service OKCupid has cheerfully admitted to manipulating what it shows users.

The three specific experiments which OKCupid co-founder and data scientist Christian Rudder posted on the site's official blog are summarized:
  1. "Love is Blind Day" — OKCupid temporarily removed all pictures from the site. Sure enough, fewer users were active, but first messages were responded to 44% more often.
  2. The second involved attempting to discover how much of a given user’s rating – the score ascribed to them by other users – was derived just from their picture. By presenting a small subset of users with their profile text hidden, Rudder found that just 10% of the typical user’s score is based on the text they write on their profile. The vast majority of the rating, it seems, is based purely on the picture.

  3. The third experiment was the most controversial — To test its theory, the site lied to a portion of users about how strongly they matched with other users, and observed how many single messages led to a full conversation. Sure enough, they found that two users who actually had a 90% match but were told that they had a 30% match were less likely to carry on talking than two users who actually had a 30% match but were told they had a 90% match. In other words, Rudder says, "if you have to choose only one or the other, the mere myth of compatibility works just as well as the truth."

While OkCupid's blog post was seemingly an effort to demystify the process of experimenting on users, for many, it simply underscored the tech industry's failure [thinkprogress.org] to "understand why some testing is ethical and some is not."

[Submitter's note: I have used OKCupid successfully for years before quitting recently due to suspicions about strange behavior of its users and mistrust in its technical capability, so I will provide context participating in the discussion should this submission be accepted. That OKCupid admitted to experimenting on its users came as no surprise to me.]


Original Submission