Facebook Is Down On Web And Mobile [Update: It's Coming Back After 20 Minutes] | Hacker News Reader

Facebook Is Down On Web And Mobile [Update: It's Coming Back After 20 Minutes]

techcrunch.com

13 pointspeterkchen13 comments

This assumes that the traffic doesn't just spike right back up after it goes down.

eg. Facebook may have a few minutes of $0 rev, followed by a few minutes of $44K per minute rev.

Also, having a larger concentration of users online at a given time can lead to increased rev. because there are more new stories for people to look at.

Interesting thinking. Wonder if the money ends up evening out with the post-downtime spike.

Neither is likely, some users may obsessively be using the service, so delays in access will instead with other times of access that they would normally use. If you assume half fall obsessive users and half fall into the non-obsessive users for that min, even $11k a min is a very large impact.

If it was an ordering system, I would expect a higher rate of recovery simply because of a service/good that was not purchased. Passive ad display is primarily driven by views and I wouldn't expect views to time shift at a high rate.

I remember reading something about Amazon where they stated that they didn't actually lose any money when their site went down, customers just came back later to place their purchases. I'm sure something similar would be the case here. They would certainly be impacted in one way or another, but not as significantly as this simple math. It would vary drastically based on the time of day the outage occurs, regions effected, devices effected, and how many users came back once the site is back up.

The math is defiantly not that simple as taking the revenue / seconds but a big difference here is also consumer goods vs. advertising money spent.

I don't know if this logic still applies here because Amazon makes it's money primarily off of selling products where Facebook makes it's money from users clicking ads.

Pocket change, all things considered.

Did nobody notice how wrong that math is?

http://www.wolframalpha.com/input/?i=2.91+billion+%24+%2F+ye...

They did $2.91B in the quarter that just passed. And are likely doing more and more revenue by the week. So doing the math by the quarter instead of year gets you the same number.

> With earnings of $2.91 billion in Q2 2014, Facebook theoretically loses about $22,453 for every minute it’s down around the globe.

Let's say they earn $2.91b per quarter for a year. $11.64 billion.

31,536,000 seconds in a year.

11,640,000,000 / 31,536,000 = $369.10 per second.

$369.10 * 60 seconds = $22,146 lost per minute

I got pretty close to their number. How do you think it's wrong?

Well, which minute matters. :) Revenue at noon Pacific is much higher than 2am, because neither humans nor wealth are evenly distributed around the world. In fact, noontime Pacific is kind of a perfect storm: lunch in one populous wealthy timezone, after school in another (east coast), and after dinner in a third (europe). These three fuck-around-on-Facebook peaks combine to form one super-peak every day.

Sure. I was just trying to get the number from the article. Not arguing its validity.

Reply on news.ycombinator.com