Hacker News Reader

Comment by ljoshua | Hacker News Reader

I'm curious to know what the metric is actually measuring. Is it one billion people that hit Facebook.com or used the mobile app? Or is it also counting people that might have triggered a hit just by browsing around sites with Like buttons on them while logged in, but didn't actually use Facebook?

To some extent it doesn't matter because it's just such a massive and impressive number, but I'd be curious to know.

View on HN

Back when I worked there, I "audited" the systems that count Facebook's active users. I needed to analyze code efficiency and feature growth over long periods of time, and so needed very accurate user counts in order to subtract user growth.

Without going into details / trade secrets: there is more than one system, they can cross-check each other, and if anything it's conservative. You have to perform legit, non-drive-by actions on either the site or the app before you're counted as an active user on a given day.

Does that include "Like buttons" and other widgets on third party websites or not?

Yes this metric counts those too.

They are not talking about active users. The metric quoted was "used Facebook in a single day". I know for a fact that they count any "Like" click or any Facebook comment on any web page on the internet when they quote "used Facebook". It does not mean logged into one of their apps.

I personally audited the entire pipeline, and code, from collection to analysis all the way to the little DAU ("daily active users") graph that informs this number. That experience, and years of working on related data sets inside Facebook, are enough to convince me that the 1B number is legit.

If that's not enough for you, well, maybe I can't help that. But I am curious what evidence leads you to believe something different. Serious question, genuinely curious.

Edit: to be clear, and IIRC, seeing a Like button is not counted. Actually Liking something, which posts to your news feed, does.

You both seem to be saying the same thing - actively clicking on a like button gets counted as a real genuine user of Facebook. That seems reasonable (one is logged in, and clicks something sending data to FB)

Neither of you have said just having FB button appear on a 3rd party site counts - that would get what 3bn users :-)

I think so, rereading. But many people assume that simply visiting a page with a widget counts.

Given that both actions require a Facebook account that you're logged in to, that seems pretty reasonable

"I know for a fact" and how do you know this? do you work at facebook? Someone who used to work there just gave a contradicting statement

That sounds pretty amazing. Can you offer any insight about users with multiple accounts? Back in the farmville days, people with 10 accounts seemed pretty common. An account just for tinder, or a professional account seems to be a common tactic as well.

I'm curious if multiple accounts are common, or just a few percent of users take it to that level.

Tricky question. I was working on perf, so I could avoid the tricky philosophical / data science questions like that. FB did say in 2012 that they believe 8.7% of accounts were dupes or fakes. I can't say what the number is now, nor comment on how that relates to DAU accounting, but that's a good rough upper bound. Further, deponent sayeth not.

My guess would be 1 billion unique accounts logged in today, via one means or another.

Absolutely mind blowing.

I wonder how many of those accounts are fake / sockpuppet / astroturf / paid likes

Again, totally guessing here, but I would guess a negligible minority

I had a GF that worked solely to create fake facebook accounts.

She alone had about 15 of them, she worked in a company that had about 10 employees that did the same thing as she did.

The company that she worked for, was one of the smallest ones, there are several others that are much, much bigger.

> I had a GF that worked solely to create fake facebook accounts.

Yes but did she connect to all of them on monday by a mean or another?

I think that's the fact it was on a single day that makes this number impressive.

Her fake accounts had as one of the objectives reach 5000 friends (when this happened you could convert it to a page with 5000 followers and sell it to a company).

To do this they created a program that searched the internet for pictures of someone in various situations, then prepared a easy-to-use library of photos to be uploaded every "x" period to look like a real person.

The account operator (or operators for some particularly prized accounts) would then log-in, post the photo scheduled, and write a post on the fly, the best operators could make the fake account look like a real person, fooling not only other users but bots, tracking software and whatnot.

The accounts that looked particularly real were very useful for PR efforts, some of the biggest clients were appliances manufacturers (example of one of them: GE), and one of the uses was for example one day try to distract the public after some bad news about washing machines causing accidents went viral, or try to create a counter-post that is viral too (in the washing machine case it worked, the viral post stopped being viral in 3 or 4 hours, and not even me remember what was the issue with the washing machines).

Assume the average such company has around 100 employees with 15 fake accounts each, now assume there are 1,000 such companies: that's 1.5 million fake accounts. At this scale, that is "a negligible minority" ;)

Presumably they only had 15 fake accounts because they put a lot of work into each account. If some company just optimized for quantity over quality, they could get that number up a lot. And if another company used bots instead of employees, they could easily have tens of millions of fake accounts.

So that is 150 in the company. Lets assume there were 100 companies like this, each one 10 times bigger than this smaller company. I'm guessing that is actually overstating the reality. That is 150,000 fake accounts. That would account for about one percent of one percent of these billion logins. So very negligible.

I would like to know more about this. Who is paying for fake activities, and to what ends?

Obvious guesses (with prior evidence!) is: Government astroturfing & information gathering (hot girl friends you + gov't gains access to your "private" content), and brand astroturfing.

http://www.newrepublic.com/article/121551/bot-bubble-click-f...

No one pays me and I had over 20 FB accounts just to play games. A lot of games give you in-game rewards for referring your friends and allow friends to gift each other in-game items.

Edit: And most of my friends have two or three real FB accounts, one for close friends, one for older relatives/their parents' friends, and one for work acquaintances/bosses. I am in the minority in only maintaining one real account.

Can you elaborate on the company's purpose? advertising?

Yes, it was a advertising company.

Beside creating false rumours for PR objectives, another use of the accounts was try to reach 5000 people in them (I dunno now, but when she was my GF facebook converted people accounts with 5000 friends into pages with 5000 followers, then ad companies could sell those to kickstart company pages).

Since we're all totally guessing here, I feel like I should point out that we're talking about FB scale here. Negligible minority for FB would be many products entire user count.

Right. 10 million fake accounts is still a negligible amount.

Well, it's only negligible within context. Nothing is just automatically negligible at scale but within the context of active users for a single day, I agree, it's negligible.

I would say you are wrong.

For a billion users, they need 1/6 of the worlds population. No way do they have that.

How many kids under 7? under 5? in the world. Don't have accounts. How many elderly over 80? 70? don't have accounts.

What is the % of non-"developed" countries? Not 1/6

Compared to people running multiple accounts (personal / professional/ seo accounts / spam accounts), now that I have heard of.

Regardless, they still have a hell of a lot of users, but 1b real, I doubt

Do you guys think this is just somehow not possible, or something? It's Facebook.

It's Facebook circa 2015 though.

Probably at least 20%. But is 800M any less significant? It's an insane number.

Who care? Even if it was 1 billion bots loging info FB in one day, it would still be massive.

I don't know but I'm not really inclined to be skeptical of this.

Also missed this information, but it's expected from such a vague post directed at non-tech users. My guess is that they counted the maximum of everything to blow up the number as much as possible - means it is separate user account hits for desktop, separate user account hits for mobile app, for messenger, API hits, etc, everything add up to one big number. Using this methodology it's not that surprising they hit 1 billion, since a single user could trigger up to 4 hits on average day and even non-users that have logged in cookie produces hits.

Also it's worth looking into the timing of this announcement. Logically, if Mark is as smart as I think he is, he would use these kind of announcements to boost share value and/or give publicity.

Facebook actually has an easier way to measure this than most: how many users logged in. I'm not sure if that's how they did this count, but that's how I initially read it.

I understand it as 1B unique users logged in, and not counting logs from same person on different devices as different users.

Leave it to a PR announcement to omit critical details.

Unless shown otherwise, I will assume this number includes anyone whom Facebook sent an email to or whom ended up on a page that loaded a script from a Facebook property.

Facebook doesn’t measure usage quite so haphazardly. This has been a topic of discussion earlier, and relevant to their SEC / financial filings. Viewing an embedded widget doesn’t count, but clicking it or signing in to an app with Facebook does.

Well, I don't see Mark claiming any of that in this post.

If your machine requests data from Facebook's machine, it's a use of Facebook. There's no way around that.

If Facebook sends personalized content in that request, then there really is no argument that this does not constitute a technical use of Facebook.

However, in most cases, users are not voluntarily making those requests to Facebook, but are more or less forced into making them by some third party (eg. any site using Facebook for a comments section).

Why does everyone always think Facebook is lying when it comes to these type of things? What good does it serve them to do so?

Seems to me it is mainly just due to the fact that most here haven't worked at anything anywhere close to FB scale and can't believe it can actually be true.

Who thinks they are lying? The same sentence can mean different things to different people depending on what definition is assigned to the words, among other things.

What they count as users is in their SEC filing: http://www.sec.gov/Archives/edgar/data/1326801/0001326801130...

A public officer of the company isn't going to be haphazard about redefining that without saying so.

You're talking about a different word.

Edit: You'll note that Mark didn't even use that word.

Reply on news.ycombinator.com