13 comments

[ 2.3 ms ] story [ 30.9 ms ] thread
I don't know, like a UNIT? Check out todays response time squiggle versus yesterdays squiggle !
Answering your own question seems at best tacky, and at worst karma-whoring.
Didn't realize it was a big deal, but I always lean towards tacky. I had to make my squiggle line joke.
Apparently the error and response time graphs are expressed in dimensionless units of relative suckiness.
If you are an app developer, you know what typical latency you get. This page serves its purpose: to let you know if there is something unusual. What's the big deal?
It's a big deal because it's showing averages for all responses. I might, on average, get crappy responses for my app, but everyone else might be getting good ones on the whole. Without a scale on the graph, I have no idea what I'm getting is "normal" or not, and may overlook something obvious, like my connection to my server, causing the issue.
The left side has measurements, I'd guess, of 0.5s.
Well then, I'd guess, their graphs are measured by aardvarks.
Why are we jumping on Facebook for this? Google does this all the time with e.g. their search volume graphs. Perhaps Facebook is simply trying to show relative.

Admittedly, the unit on these graphs is quite important while the search volume unit isn't (as) important.

Adding a Y axis would immediately draw up hundreds of blog posts and academic papers citing it as a "benchmark", making various claims and hypotheticals about Facebook's infrastructure. Since we don't have access to the exact metric or method of measurement, the exact magnitude is arguably meaningless.

Also, since this has potential for FUD and Facebook-bashing, it's in their interest to not publish raw numbers.

Or they could just leave out the graphs altogether. As it is, they’re pretty much useless.
No it's not useless since it also includes yesterday's measurement for that same timeframe.
Oh. my. god! There's no "Like" button on the page.....