How Facebook likes are calculated | Hacker News Reader

How Facebook likes are calculated

almightylink.ksablan.com

88 pointsblazer748633 comments

The topic makes FB sound deceitful, but they don't hide this information, it says it right here: https://developers.facebook.com/docs/reference/plugins/like/

The number shown is the sum of:

-The number of likes of this URL

-The number of shares of this URL (this includes copy/pasting a link back to Facebook)

-The number of likes and comments on stories on Facebook about this URL

-The number of inbox messages containing this URL as an attachment.

You're right. I intentionally steered away from calling the number a "lie" because FB clearly states what the number includes.

> The number of inbox messages containing this URL as an attachment.

I LOVE that they do this.

It's fascinating! It would be great to have this information from just normal email -- the data's been there for many years -- but does it seem like privacy invasion then? There doesn't seem to be any difference...

My mail feels private, Facebook doesn't. A lot of people complained loudly when Google started scanning the text of emails to display "relevant" ads. My guess is, there will not be very many people complaining about this.

I suspect it only feels this way to you and me because we know better than to treat anything on Facebook as private, whereas email still feels like it at least should be.

But for my older relatives, and also for a lot of my friends on Facebook, I suspect there is no difference in their minds.

The mob only makes noise when shit hits the fan, the majority is either willfully ignorant or overly paranoid. Even among techs.

What if I'm mailing a friend because I don't like the item?

More Orwellian than deceitful, because they're using "liked" to represent the aggregate when "likes" is already its own metric, which happens to be lower than the aggregate.

It's kind of funny really - 'did you mean "liked" or "likes"?'

I wonder what happens when I like and then comment on an item. Is that counted twice?

I would imagine that they find it easier to compute the aggregate and display it rather than maintain different values and display it to people (probably helps them at the scale at which they operate? -- lesser things they show, lesser damage it does?).

To get the share of likes of any URL just

https://api.facebook.com/method/links.getStats?urls=http://n... (for this thread) or https://api.facebook.com/method/links.getStats?urls=http://n... for HN home

(View sourcecode if your browser doesn't format the XML to you as Firefox does)

We noticed this when starting out company (which measures these stats for your blog), and it never struck me as odd. Sure it seems a little disingenuous at first, but the Like button has replaced the Share button as the preferred implementation. Moreover, every like, share, or comment gets posted to your wall for your friends to see (privacy settings depending) so you are in fact "recommending" it.

The difficulty with that is that some comments actually contest the information in the article. For example, if someone commented on my blog post and said "This is bogus, the like number is accurate" ... I would hardly call that a recommendation.

In Czech localization, the button actually says translation of "recommended", as do the message that gets on you wall by using it. On the other hand, comments, wallposts pages and what not use wording with equivalent of "like".

I've too noticed that they aggregate 'button clicked likes', comments and shares as overall 'likes'. However, I just assumed that their logic is that if someone posts a comment or shares the link, they're 'liking' it.

I don't think that logic holds; I can comment on something, or provide a link to it, because I hate it.

True, but if you take "like" to mean "validate," then that logic does hold -- even if you hate an article you're sharing, you're still validating it.

Like does not mean validate though.

You may as well say if friend means totally random person you've only met once. Oh, wait...

The reality of what "like" represents indicates otherwise.

If they use the wording "recommend this" and not "liked this" then I expect sharing and liking (but perhaps not commenting on) an article to constitute as recommending it.

I think it's largely impossible for him to sift through this data accurately. Who is to say that diff people didn't share something vs liking it? Also he makes it seem as if the news sites are falsely reoresenting their content popularity when all they are doing is presenting a fb social widget.

I don't see too big of a problem with it regardless. If you spend your time liking, sharing, or commenting on something there is obviously an 'emotional' response to it. As long as FB doesn't start saying 'your friend x' likes this, when in fact they posted a comment disagreeing there isn't a huge problem. Even then I think the fact that it created a response of some sort is a 'like', atleast in the sense people I see use the like button..

For most purposes, yes. But the author called out news reporting that quotes these "stats". Its important for journalists to understand the differences.

I agree that number is still a good number to track, but it just isn't what it appears to be. Sure, Facebook spells it out in their docs, but most people won't ever see that documentation. The fact that the number appears next to the word "like" implies that it is the number of "likes."

I created a tool to calculate this over a year ago (http://www.allfacebook.com/facebook-like-count) ... this actually benefits the publisher in that the more likes, the more likely a user will like it. Not completely transparent, but it's a well known fact.

Interesting...First of @Ksablan - you did use the word Lie...and secondly while at first i thought it was a huge deal...its really not. The point of the metric is social sharing...and Comments, likes and shares are all social sharing you know? No need to get more granular than that i think

Unless the post has been edited since, I don't see anywhere in the post where says the word lie.

As to your second point, it depends on how the number is being presented. If it's being shown as the number of recommendations, then including the number of comments is clearly inaccurate. A comment disagreeing with an article does say something as a useful metric, but it certainly isn't a recommendation.

Not a particularly big deal in the long run, but I still think the article has a point.

Are there privacy options that hide the details of the likes from his tool (so he can't count them) that Facebook can count and display in the widgets? Might account for some of the difference.

Actually, all of the likes, shares and comments were accounted for (the parts added up to the sum) for each of the articles I examined. But there were a few sites for which my button simply didn't work. I'm not sure how those sites were able to do that.

They round off after 1000 Likes/Shares/Comments anyway (the widget reads 1K for more than 1000, etc.), so many people won't expect it to be a perfect statistic.

My world is upside down. I hope I don't find out today that LinkedIn recommendations aren't accurate.

haha amazing. What's next? Amazon recommendations are all a creation of Jeff Bezos imagination.

Reply on news.ycombinator.com