Disqus Research: Pseudonyms Drive Communities | Hacker News Reader

Disqus Research: Pseudonyms Drive Communities

disqus.com

134 pointssimanyay48 comments

They did nothing to demonstrate non-Facebook logins are actually pseudonymous. I use my real name without using Facebook. I believe many do as well. Even if not true it could be true and they haven't shown otherwise.

They presumed the effect they are trying to measure. It is easy to draw a conclusion when you start with it as a premise!

I think it's more useful to think of pseudonyms as a superset of read identities--they're names that do not have to be your real name, but they do not have to not be your name either.

Since most sites would not proscribe using your real name as a username, this definition makes the most sense.

I think the conclusions this infographic draws are absurd. Let me explain by way of analogy:

Reddit and HackerNews are two sites on the internet, each with their own community of commenters. Let us assume for the sake of argument that Reddit users enjoy inane off-topic humor in their comment threads, and HackerNews users enjoy thoughtful posts with occasional dry wit. Let us also assume that Reddit users can post on HackerNews using their Reddit credentials, and vice versa.

After some analysis of their comment threads, Reddit makes the following claim: Reddit users drive communities! After all, nearly all of the comments on the site are inane banter from Reddit users. People from HackerNews rarely comment on the site, and Reddit's indicators of comment quality show that Reddit users generally post higher quality comments. They get so many upvotes!

But on the other hand, HackerNews claims that HN users drive communities! After all, nearly all of the comments are thoughtful and insightful posts from HackerNews users. People from Reddit hardly ever comment on HN articles, and Paul Graham's indicators of quality show that HN users generally post higher quality comments. Reddit comments tend to get downvoted to oblivion.

Anyways, I just wanted to show by this example that if you take a service that caters to a particular demographic, and ask its users to rate each other, of course you will find that the demographic comes out in a positive light.

I too think this page is too light on details.

I can see one major bias (albeit inferred) - they consider anyone posting using Facebook to be using a "real name" - and assuming that anyone posting using a "real-like" name outside of Facebook is a pseudonym.

If this is true, they're really only measuring and comparing the quality and quantity of Facebook-identified commenters vs other commenters.

Only Facebook commenters have "enforced" real names, in that their Facebook name is used. Signing in with a Disqus account or Twitter uses the names that a user chose on that service, which typically is not a real name but some sort of pseudonym.

So while it's true that it's basiaclly comparing Facebook-identified commenters to other commenters, only Facebook-identified commenters are required (assuming they sign in with their actual account) to have a real name.

This page seems long on cute fuzzy pictures and short on content. They don't say what discussion forums they sampled: most likely, their own. They don't describe their methodology in detail (I looked in vain for a link to or citation of a proper experimental writeup.) And what did they learn?

As far as I can tell, they learned that people who bother to sign up for forum accounts ("pseudonyms", in their terminology) make more return visits/contributions than people who use least-effort ways of logging in (existing Facebook account, or anonymous contribution).

Well, that's a surprise.

We sample data from the entire network of Disqus (which is over a million sites).

I may be wrong (as I'm not on the marketing side of things), but I believe when we're saying pseudonym we're saying "not Facebook". Things that let you choose whatever name you want, and not require your real name/singular identity.

Thanks for following up. I think you're confirming my assumption on the definition of "pseudonym", but correct me if you disagree.

Oh, and since I seem to have zeeg's attention ... why is it that every site I care about works just fine on my browser (Chromium/incognito mode/AdBlockPlus) but Disqus comments never load (I just get a "Disqus" spinning/loading GIF for ever)? I have stopped commenting on BoingBoing because it switched to Disqus and I don't care enough to debug.

Are you blocking JavaScript somehow? Or AdBlocking the disqus media maybe?

If you don't have JS disabled like zeeg mentioned, AdBlockPlus is a red flag. Can you try incognito mode with no extensions enabled? If that doesn't fix this reach out to our support team at http://disqus.com/support and we'd be happy to lend a hand.

Sorry, we put up the image first but haven't gotten a blog post up yet. We'll write about this and put up our thoughts soon.

We're going to blog about it which should shed more light on methodology -- TC just linked to the infographic. But yeah this was across the entire Disqus network, ~600 MM monthly UVs and ~60 MM commenters, and I believe "pseudonyms" include proprietary site logins + social media handles besides FB.

The major insight was that pseudonyms yield significantly more quantity and quality on the network. May seem obvious to you but wasn't so for many, including a number of us who've been with Disqus for years. The conventional wisdom was that anonymous comments would rank highest in quantity and FB comments would score highest in quality.

If I may offer a suggestion, perhaps mention that pseudonyms generally tend to be long-held identities - a lot of people think that pseudonyms are just throw-away identities used to gossip about your employer safely, instead of identities that persist for months, years or even decades. (A lot of people would recognize the handle CmdrTaco, but how many know his real name?)

(p.s. your backyard is awesome)

"pseudonyms generally tend to be long-held identities - a lot of people think that pseudonyms are just throw-away identities used to gossip about your employer safely, instead of identities that persist for months, years or even decades."

A fact too often forgotten. This is the same ignorance that keeps me from using Google+ while I'm prolific on Twitter.

Honestly I don't see how you could possibly have a reasonable methodology for the claims you make in the infographic.

How do you avoid selection bias, particularly in regards to frequency of comments? For example, a source of technical bias could be that it is a simpler process to log into a Disqus pseudonym than to authenticate with Facebook when posting. Usability is incredibly important in attracting repeat users, so if it is easier to post under a disqus username then it is not surprising that pseudonymous users post more frequently. Another source of obvious selection bias is the disqus branding: I expect people who have Disqus accounts are both more likely to frequent blogs, and are more likely to recognize that they can easily comment on a blog with a Disqus comment section. By way of contrast, Facebook users may not even recognize that they have the credentials to easily comment, and may be reluctant to share their credentials with your third party service except in cases where they have a strong opinion about a piece of content.

I can see the data on comment quality being more objective and interesting, but even here there is probably selection bias: if most of the indicators of quality come from Disqus accounts' appraisals, then you may be measuring whatever bias Disqus users have as regards comment quality from Facebook commenters.

There's no selection bias here (other than the obvious, blogs using disqus), since they are using aggregate data. They make the false claim that pseudonyms contribute more though which does not follow from their data. They should show comments per user per login source data to make such a conclusion. Their current data only hint that using a disqus account is easier than facebook or anonymous.

I don't think there's such "comment racism" in disqus. I 'd argue most disqus users barely notice they are using disqus.

The data presented in this post doesn't support the conclusion. The data do not show that pseudonyms are the cause of more quantity and higher quality, they just show a correlation. I hope that the full blog post that's been promised addresses this.

How would you go about showing causation?

TLDR; people are more likely to comment when their true identity is masked.

This is, IMO, likely due to the fact that people can't be held accountable for their comments if nobody knows their true identity.

Personally, I have made it a goal to only use alias variations including my name for the sole purpose of being held accountable for my actions online; good or bad. It's a better moral compass than acting on behalf of some made up "pseudonym" (i.e. alter-ego). If I wouldn't say it in public, odds are I shouldn't be saying it online.

Note: In the case of my HN alias, I've always included my full name in the profile page. The alias references a domain I started years back and was originally intended to help push traffic. It slowly morphed into my personal account as I bridged the gap between reader and contributor.

"If I wouldn't say it in public, odds are I shouldn't be saying it online."

I've been working on social sites for years, and I disagree with completely. There are many reasons, but here's one. What I think when I'm 20 is not a good indicator of who I am when I'm 40. I remember I posted with my real on some pantheist site when I was young, about why I was a pantheist, and for many years that was one of the top results for a search for my name. I regretted it, and luckily it's falling off of google.

Or, here's another example. I've made a small little community site (yakkstr.com) and lately there has been a lot of talk about non monogamous relationships and our feelings about them. I doubt people would be honest about this if they were using their real names rather than pseudonyms, and the value of the conversation is much greater when people are being honest.

You regret a number of things you've done in your lifetime. It's a part of being human. Do you think Zuck regrets his IM logs:

   Zuck: Yeah so if you ever need info about anyone at Harvard

Damn straight he does. He'd love to take it back. It's in the past, and I think we can all agree that it was an immature comment by an early 20 something and he has since matured.

I think you missed my main point. There is real value to pseudonyms, even if accountability is also good.

If one could be discriminated against for religious views then it's probably a good idea to not use your real name. And the list goes on and on.

Considering that he's built a multi-billion dollar empire off his view of un-privacy, it's not at all obvious that he regrets that comment or has matured past it. He may have grown that youthful apparent-indiscretion into his adult worldview. He may well be just waiting for us to catch up to his point of view.

TLDR; people are more likely to comment when their true identity is masked.

No, the data doesn't support that conclusion at all.

The data shows that most comments are made under pseudonyms. While it would be correct to say that the hypothesis that people are more likely to comment when their identity is masked would produce this data, the hypothesis that more people post using pseudonyms (though possibly less frequently) also supports this data. Therefore, the data as given can be said to support neither conclusion.

You are missing the paragraph titled "Average Comments Per User By Identity". That paragraph at least supports the conclusion that people who post under pseudonyms post more frequently, which refutes your alternate hypothesis.

Of course, there's no indication of methodology, and there's nothing to suggest that those who post under a true identity would post more frequently if they were pseudonymous.

This is, IMO, likely due to the fact that people can't be held accountable for their comments if nobody knows their true identity.

(I'm only a single data point but..) I'm more likely to comment on something the more I can use my real identity as a by-line. Demonstrating one's involvement or knowledge in a certain area can be valuable (think HN or Stack Overflow).

To me, it seems the sort of comments that would be encouraged by anonymity are those that are particularly controversial, negative, or that break confidences.

Are they trying to convince their future investors that they will survive, despite of the Facebook Comments social plugin? Their Alexa traffic data seems to contradict this.

Using directly-measured data, the Disqus network holds the #1 U.S. Network rank in Quantcast: http://www.quantcast.com/p-94WKwgUwZHlfo

I think this is especially shown in reddit. The idea that you have no idea who a username is but you know that it is the same person each time... I think this may be the factor that brings actual communication.

- In the graph, 61% appears to be inline with 51% .

- Would be interesting to see deviation bars on those values - my hunch is that real identity is not different from pseudo.

- It's false to suggest that pseudonymous users contribute more comments. You need comments per user data for that. Your graph just shows that pseudonymous accounts are easier to use.

- Why is everyone here attacking their data. Disqus is huge, this is an interesting topic and it would be interesting to have more data from them

- While writing your blog post, please also report separately the number of responses a comment evokes, or the length of the comment. It would also be interesting to have the number of edits per comment.

- Also post comments per user for different login sources

Note that the "Quality Signals by Identity"-graph is broken. 61% != 51% (well, at least not on a %-scale).

The graphic for "Quality Signals by Identity" shows two bars for 61% and 51% with the same length.

Based on the info here in the comments, it seems that pseudonyms are any names other than FB logins.

Perhaps this could be as much an analysis of the quality of a comment left by your typical FaceBook user. I know that as the Shepard of my own digital identity I'd much rather register www.myname.com and have a blog than use FB on a regular basis. Perhaps those of us that have been posting online for longer still live in the days of 'doing it ourselves.'

It wouldn't be surprising if FB adoption is higher among web novices than web pros simply because FB wasn't there when web pros started. This is then reflected in web pros leaving more useful comments for a variety of reasons.

In addition, as other commenters point out that pseudonyms are not just another form of anonymous posting, often with more personality than the poster's actual name. Being forced to give a real name requires a lot more commitment from many posters.

I completely agree with the conclusion of the infographics, but just from the point of view of the approach... wouldn't it be a possibility that people aren't necessarily avoiding real names but rather facebook profile access?

my personal thoughts here is that this question represents tunnel vision of the tech community. Anecdotally my techie friends avoid it, but the vast majority of my non techie friends use it all the time.

The vast majority of my techie friends use it all the time. Maybe I just need better techie friends.

Content aside, I dislike the giant-image-as-page-layout style. I remember seeing stuff like that in the 90s, but given the rich abilities of browsers to style content, no one should do this anymore.

Giant images have more power to do striking layouts than CSS does. The advantage of CSS is that it will probably look good in every browser, while a giant image is a particular size and doesn't scale.

Images can have greater.... imagery? Can't think of a good word for it right now.

Bonus gripe: the pie chart is totally misleading, for two related reasons:

- Some segments are arbitrarily taller than other segments (makes them look bigger)

- Some segments are deeper than other segments due to where they are in the bubble.

"Real identity" is listed as 4% of the graph, but it's not clear if they intended for it to be 4% of the volume, top-down area, or horizontal distance on the graph.

It's the worst kind of chartjunk: not just unnecessary, but actively misleading.

This doesn't surprise me. Pseudonyms are the middle ground between anonymous and real. This article seems to point to the fact that pseudonyms are the best of both worlds, ie more likely to comment (from anonymity), while the quality remains high (from actual identity).

Whether or not this report is comprehensive, and I believe it to be pretty airy, this is still extremely fascinating.

This has been intuitively known since the Internet emerged. Hell, more people know me by a pseudonym that I've used since 1992 (not this one) than my real name.

The only people pushing "real names" are Facebook and Google, and they are doing so because it helps them build a better profile of you and make more money. Building community is a secondary concern.

The real identity thing also helps build the fuel that keeps the social engines running: gossip.

1) I am not surprised by the findings considering that it came from disqus 2) Posting more doesn't mean better. As a reader, it's often more helpful to have better comments than to have more comments. A good example would be the drop in trolling comments after TC switched to the facebook commenting systems.

I agree 100%. Couldn't have said it better myself. So true. Up vote this comment.

My issue with Disqus' pseudonyms is that they don't seem to be personally owned.

I selected a pseudonym, put in an email and a password, made a few comments... then I realized that some girl's photo is next to my comments! Apparently I hijacked her account.

If you registered an account with your email address you wouldn't have been able to hijack someone else's account -- all registered accounts are tied to unique email addresses. Can you reach out to us at http://disqus.com/support with some more specific details? We'd be happy to help get everything clarified.

I love that we're able to publish research like this.

This is of course true. look at 4chan. they have enormous traffic and is real active, but totally annonymous!

http://www.readwriteweb.com/archives/4chans_chris_poole_face...

http://www.ted.com/talks/christopher_m00t_poole_the_case_for...

Reply on news.ycombinator.com