Disqus has released an infographic of some analysis they’ve done on their comments to compare pseudonymous, eponymous (real name) and anonymous commenters. They looked at both quantity and quality and found that pseudonymous commenters are better for a community than either eponymous or anonymous commenters. To save you from having to wade through a rather pointless infographic, here are the key facts:
Disqus measured Quality and Quantity:
- Positive measures
- Number of times a comment is liked
- Number of times a comment is replied to
- Negative measures
- Number of times a comment is flagged
- Number of times a comment is marked as spam
- Number of times a comment is deleted
They found that, by these measures:
- Pseudonymous comments were
- 61% positive
- 28% neutral
- 11% negative
- Anonymous comments were:
- 34% positive
- 55% neutral
- 11% negative
- Real name comments were:
- 51% positive
- 40% neutral
- 9% negative
- Aggregate number of comments by identity
- Average number of comments by identity
They found that the percentage of comments by identity was:
- 61% pseudonymous
- 35% anonymous
- 4% real name
The average pseudonymous commenter contributed 6.5 times more than the average anonymous commenter and 4.7 times more than commenters identified via Facebook.
Now, this data is interesting, but although it’s not really a smoking gun, it certainly should give companies pause before they start trying to force people to use their real names instead of pseudonyms; they may well be encouraging a less civil environment rather than the more civil one they are trying to, or telling us that they are trying to, nurture.
I would like Disqus to repeat their work but be a bit more rigorous. For example, testing their data to ensure that they are accurately differentiating between pseudonymous, anonymous and eponymous commenters. After all, using Facebook to log in doesn’t guarantee that someone is eponymous, nor does not using it mean they are not. I’d also like them to test their quality measures against both sentiment analysis and a panel of real humans. The latter would be relatively easy to do via something like Mechanical Turk. Of course, if they’ve done this already they should publish the details in a methodology.
The whole argument about anonymity, pseudonymity and real names on the internet over the last year or so has been mainly people arguing from assertion, so it is nice to see some real data. And there can be no doubt that Disqus has a lot of comments to analyse, so this isn’t just some skewed sample from a tiny corner of the web. But we do need both to see more work in this area and more companies taking notice of the evidence instead of sticking to their well-oiled but misfiring guns.