The Guardian publishes stats on the size of their commenting community

 by Martin Belam, 27 December 2012

I don’t want to unnecessarily poke the hornet’s nest that is user reaction to the Guardian’s introduction of nested comments, but Chris Elliott’s recent column about it contains one fascinating stat, which I don’t think has been made public before:

“The Guardian website publishes around 600,000 comments a month, with 2,600 people posting more than 40 comments a month.”

I’m not aware of another newspaper website breaking down their community usage figure in that way, and identifying the size and activity of the heaviest users. It would be fascinating, for example, to compare that 2,600 figure with the people exhibiting the same sorts of behaviour on The Times’ website, behind their paywall.

Something else struck me about the numbers.

2,600 people posting at least 40 comments each adds up 104,000 comments, a significant chunk of the overall total of 600,000. That leaves a maximum of 496,000 comments being left by everybody else. Add the 2,600 prolific commenters, and you therefore have a maximum possible number of monthly commenters being 498,600.

How does that figure stack up against the Guardian’s overall audience, and the typical 90:9:1 rule about participation that people experience on the web?

Well, the Guardian’s audited audience figure for November 2012 was 70,566,108 unique users. 498,600 users leaving comments out of a total audience of 70.5million is 0.7% — not a million miles from the 1%.

The figures are fuzzy here though. It is unlikely that the 2,600 people mentioned all left just 40 comments each — it took me about five minutes to find four users who alone have left over 1,500 comments between them so far in December[1, 2, 3, 4], a number nearly ten times higher than a model of them making 40 comments each would predict. It is also unlikely that the remaining comments were all left by drive-by users making one comment each. Both of those trends would suggest 498,600 users is an over-estimation.

Equally, “unique users”, whilst being an agreed industry standard, does not track users across multiple devices or access points, and so will also over-estimate audience.

So, without the exact figures, people can argue themselves blue in the face about whether that 0.7% figure gets closer or further away from 1%.

But one thing is clear from the numbers in the article.

At least 20% of the comments left on the Guardian website each month come from only 2,600 user accounts, who together make up just 0.0037% of the Guardian’s declared monthly audience.

You might also be interested in Guardian comment system changes: The perils of designing for all users, not just the vocal ones


Re the awful nesting: So, why did it take you so much typing, Martin, to say the words "Piss Off current commenters as we only want twitter-type crap posted"?

Have a nice life.

Giselle, thanks for the comment, but just to be clear, I left the Guardian in June, I didn’t design the changes and I didn’t implement the changes. I’ve had nothing to do with the nesting other than also express an opinion about it as a regular commenter on the site.

I think you'll find that the nesting debacle will have repercussions well beyond the change itself.

With the shallowing of commentary, there's been a distinct shift to the right politically, and hence the number of drive by right wing posters has rocketed.

At the same time, the number of left of centre posters had visibly dwindled, partly because of the new shit format, and also because right wing trolls aren't fussy about hitting the 'report' button should they not like what they read.

I've just been banned from posting on the Graun site after many years on there, and repeated requests from them to explain why this has happened have been ignored. They've made it clear to me and others like me that - like Clegg and his dislike of the left wing in his party of frauds - left wing views aren't welcome at the Graun, so I've taken the hint and absconded permanently to

As for the Graun? Well, they can kiss my arse. I've been a loyal reader of the Graun and Absurder for 32 years, but with their current attitude, they'll be losing more readers like me. My £10 a week to them is chicken feed I'm sure, but for every vocal protest voice like mine, there are bound to be lots more who vote with their feet, for the same reasons.

Remember that ratio - 90:9:1; I'm the one gobby enough to say it.

Well that’s all very well and good Blackops. But why did The Guardian not just give users the option of whether to view the site in Threaded or Chronology order?

i think it was you

Ooof. Well that's my new year resolution sorted then... (>_

I also think that was you.......!!!

Shit, Martin, they're on to us!

The figures are kind of confusing. You cannot certainly predict how many users leave comments per month – it is certainly an over-estimation of the audience.

I agree with you. If we take into account that only 20% of the comments are left by the 2,600 registered users, the numbers left behind are quite amazing.

Nested comments may have brought about a larger number of comments; nevertheless, it is an over estimation of the newspaper's audience.

Having a small number of regular commenting contributors vs. the total number of readers is a norm actually, but these are your most valuable commenters. At the community where I'm a top commenter, there are about 1,000 regular commenters out of an audience of about 250,000 readers, but that small number carries a lot of weight and value to the 249,000 others.

I don't think they should look at the small sample of high-frequency contributors as non-significant. These 2,600 are probably responsible for at half of the 600,000 other comments because they are the instigators and pot stirrers of other discussions.

Here from Andrew Ducker's link:


I used to quite enjoy reading comments below the line on the Guardian, but lately the majority of comments seem to be from a community of combative, somewhat hostile long-term commenters who love fighting and hate what they just read. Maybe it's always been like that, but the addition of nested comments (which usually makes a discussion nicer to read) seems to have made it even more negative.

Occasionally I consider joining, but mostly it looks like a complete bear-pit, best avoided.

Shows what a desperate roll of the dice it was. To knowingly introduce a design that would by (by its introducer's own admission) dramatically worsen the experience for this core group of users, in the hope of attracting a new and larger crop. I expect the curve gets even more steep - e.g. 10% of comments from a few 100 accounts?

Much as I dislike the new format, I hope it helps turn things around financially. But comment quantity not quality wouldn't have been my choice of strategy to keep people coming back.

Another interpretation (suggested by cbarr) was that it could have been a deliberate clique-busting move to disrupt the current community and as such reduce the perceived barrier to entry to potential new commenters. It would be interesting to know to what extent The Guardian uses deliberate 'community shaping' techniques, I always assumed that MoveAnyMountain was one, and a pretty effective one at that.

Ok, 'dramatically' is overdramatising it.

But in the article justifying the change (which now I can't find... ), Julian F said something very much like: 'the new system is better for people that read once or are only interested in a particular thread' which I think most people would view as a nice way of saying 'the system will be worse for people that follow a whole discussion and make repeat visits to it (to make further comments and see how the discussion has developed)'.

I think it's reasonable to assume that people that make high to very high volumes of comments do follow this repeat-viewing behaviour quite a lot of the time, i.e. their high volumes of comments are not just from one drive-by comment on lots of articles, but several comments on each, you can see this by looking at the profile pages of the more prolific commentors. Whether or not they also watch the discussion afterwards I don't have a way of measuring, but I'd be pretty confident they do.

Julian went on to say that only a very small number of users appeared to exhibit this repeat viewing behaviour, with the strong implication that the change to nesting was justified on the basis that it would only adversely affect this small minority. He then went on to describe how the changes were intended to get more people commenting.

So while there was no literal admission, I think the message was pretty clear that repeat viewers would be an unfortunate casualty in the quest to grow (or retain) a wide audience.

Aha, found it:
"We realize that there are a smaller set of comment readers who repeatedly return to a page and want to read every comment that is posted and I think it is these readers who will be most irritated by threading."

So either:
1. They didn't realise that this 'smaller set of comment readers' is likely to heavily overlap with their tiny set of high volume comment generators - your 0.0037%. So didn't think the change would have a negative effect on 'higher value' users.
2. They did realise and but thought that the tiny set would just put up with it as they are addicted to commenting (a fair guess).
3. They did realise but reasoned that on balance the drop from the high commentors would be compensated for by an increase in comments from new people, and a 'better' experience for read-once viewers.
4. They want to discourage this kind of repeat-viewing, repeat commenting per article behaviour as it isn't as valuable as single viewing, single commenting (assuming you make the same number of views / comments).
5. They actively wanted to disrupt the current set of high volume commentors - clique busting.
(3 & 4 are pretty similar)

I'd like to think it was 3, as the paper does need to earn a crust.

In any case, it is a mystery why they didn't choose to develop behaviour like cifFix, which better supports both the single and repeat viewing use cases.

Keep up to date on my new blog