[dredmorbius]

Google have been spewing gratuitously specious numbers for some years now. A piece in VentureBeat from 2012:

http://venturebeat.com/2012/01/20/google-skews-google-plus-s...

"This post on Google+ statistics is a billion* times better than any other post"

But what concerns me most is that Google is touting these meaningless statistics in the hopes that journalists will misunderstand them and report that Google+ is seeing rapid growth. The bottom line is, those 60 percents, 80 percents and 90 million registered users are just there to mask the fact that Google doesn’t want to tell us how many people are actually using Google+.

(Disclaimer: I helped lift the lid on actual G+ usage numbers.)

[jacquesm]

what's the trend on G+ usage? Any chance that in the near future we'll be able to search for anything with a '+' sign in it and get a relevant answer rather than some G+ page?

[dredmorbius]

I haven't run a re-assessment, and I'm not aware that Stone Temple Consulting (who'd greatly expanded sampling based on my methods) have, though Eric Enge's mentioned doing so.

There are a few sources which track online social network usage, and my experience has been that G+ trends are down, sufficiently enough that they've fallen off the survey's entirely.

If you wan to trac total posting activity and trends per year, a Google search on "this site:plus.google.com" restricted by year should give a rough idea of how the relative quantity of English-lanugage posts are trending. I'm suggesting this as a nonsemantic, high-frequency, English word not likely to be biased strongly by annual trends. (Alternatively "upon" and "thus" are suggested by another person who's done similar research.)

My informal sense is that people are tending to defect from the system. I'm very hesitant to apply personal impressions though due to the enormous sampling bias implied.

[jacquesm]

Thank you.

[dredmorbius]

A quick check on "this" and "upon thus" searches suggests that Google's search results depth may not be sufficient to measure the size. I was seeing ~400-500 results for each going back to 2012, on an Oct 26 - Oct 25 year (just to get most current data).