All I know is that out of large numbers of followers, little happened even in the most popular topics like TV & Movies collection which had 450,000 subscribers. This could be due to many factors, but I found the numbers off by a factor of 100, that is for 16,000, only 160 at most interacted.

at one time, it did not feel trivial to set up a G+ profile. It seems one had to jump through a hoop or two - but I have long since forgotten what sort of hoops they were. Google let go the reins at some point and let billions of accounts be created. It is like shooting a hole in the hull of your boat and blaming the ocean for pouring in and sinking you. This whole experience (coupled with anecdotes about their record of killing products) is convincing me that Google is managed by some folks with serious tunnel vision - which, in business, is very often a fatal condition.

@Cade Johnson I think and it was widely thought that on the contrary, they forced profile creation, no?

I recall that at least for a significant length of time, anyone who had an account at any Google-operated service also automatically had a Google+ account. That policy/practice effectively ended some years ago though.

Too lazy r/n to google the precise minute-by-minute play-by-play.

To the point where people complained you couldn't not have one! They did fix that, you can delete your profile. In fact, I think I will after tomorrow's Diaspora conference.

I guess I came in through the back door. I had to have a referral for a gmail account, and then someone had to invite me to join G+. It was supposedly essential that I use my real name - though "Cade" is just a nickname. At the time, if you did not have a personal photo for your profile, nobody would talk to you.

@Cade Johnson would that have been in the days of closed beta? Sure, then it was hard, but IIRC, shortly afterward they auto-created a profile for everyone, or was that just when you clicked on a link to Google+, it passed thru create. I don't recall now.

if I had a nickle for everything I have forgotten . . . well, I guess I'd have a nickle wouldn't I? ;) I don't know when I joined but I think it was after beta testing.

OK, some further numbers on both the 2015 active profile sample and communities.

At over 2,600 profiles in, the 404 rate among the 2015 active profiles is over 13%. That is, nearly ten times the rate of the 404s amongst the general 2017-03-01 population.

I'm running a second Communities sample of 12,000. At just over 1,000 of those run (rate's about 1/s), we're looking at a mean again of 22.9, a max now of 925, about a 15% private rate, and ... nearly equal open vs. closed ('Join' vs. 'Ask to join') options. Those are going to plunk away for a while.

The communities problem is a hard one. Think of it like this: you're going to randomly drop non-mobile probes over Earth's surface. What are the odds of landing in, say, New York City or Beijing? Odds of landing in any urban space are roughly 1% (human urban land use).

On the other hand, if you randomly sample among human beings you're far more likely to find someone who lives in a larger city than a remote area (urbanisation is well over 50% of the global population). The question becomes: how do you find large communities on G+ without a priori knowledge? I'm kicking that one around.

@Ferdi Zebua The automatic-assignment element was canned ~2015 as I recall, as the G+ - YT #Anshluss was being unwound.

God damn that was a clusterfuck.

Oh, I'd goofed on thousands separators -- found a few groups > 1,000 members, largest so far, 26,000 or so.

And some more on the communities, with 4,460+ polled:

Max is now 217,000 members, private are holding at about 9%, open vs. closed membership is 55% / 45%, mean membership is 127, and the median remains 2. 95%ile is 121.

Half the observed membership is in the top 2-3 communities observed. (This excludes private communities, but for reasons discussed over the years, I doubt these are significantly large in any way that would disrupt that statistic.)

Continuing to poll for data, still less than 50% of the way through.

Based on the 2015 active profiles poll, now complete, 13.24% show HTTP 404 status. Those are now dead accounts.

@Di Cleverly @David Thiery y'all might be interested.

OK, starting to get a handle on communities.

Most are small. To a rough approximation, the size of a G+ Community is 1 person.

There are exceptions. Though outliers, the vast majority of all Community membership resides in a minuscule number of enormous communities.

How minuscule? They're not appearing in the sample of 5,500+ that I've polled so far, with one exception. "La Palabra de Dios tiene Poder" has 217,297 members, larger by a (nearly) factor of ten over the next largest.

There are likely about 69,000 Google+ communities with over 1,000 members. That's actually a fairly sizeable number of fair sized communities, and indicates possible targets for migration.

How'd I get that number, you ask? With 4,879 public Communities showing membership (private communities don't reveal this), which are about 91% of all communities, there are 47 with > 1,000 members. With 7.9 million communities, 91% public, dividing by 4879 gives 1,473.4577, so that we can roughly estimate that each of the communities on my ranked listing stands in for about 1,473 actual communities. This puts the number of public communities with > 1,000 members as:

47 * ((0.91 * 7.9 million) / 4879)

Or: 69,252.511

Keep in mind, these are estimates, they're based on a random sample, and I'm continuing to poll more from my sample of 12,000 communities (we're now at 5766, just under half way). So values may slide around a bit, though I expect they'll stick fairly close to this (5700 is a pretty large sample). The data are Poisson / Power distributed, though, so outliers may be significant.

For grins, I've taken my ~5k results and sliced them into buckets of 500 (yay, shell tools). The median remains nearly constant -- either 2 or 1 for each subsample. The mean is all over the place, with a low of 25.35 and a high of 466.76. The 95%ile is reasonably constant: a min of 84.5, a max of 156, mean of 119.75, median of 115.75.

(Percentiles tend to be fairly insensitive to outliers, yay!)

Updating: I'm getting a count of 58,186 communities > 1,000 members. Let's call it 50 - 70k for now.

Gah, I goofed on the community URL above, it should be:

https://plus.google.com/communities/118362296060634412141

Also, revisiting my earlier 300 community sample there were 3 communities with > 1,000 members:

3805 members
2530 members
1118 members

That bumps stats to:

n: 256, sum: 13248, min: 1, max: 3805, mean: 51.750000, median: 2, sd: 300.865066
%-ile: 5: 1, 10: 1, 15: 1, 20: 1, 25: 1, 30: 1, 35: 1, 40: 1, 45: 1, 55: 2, 60: 3, 65: 4, 70: 6, 75: 8, 80: 16.5, 85: 24, 90: 38, 95: 238

That is great work!

@Christian Buggedei Thanks. Makes up a bit for ignoring the rest of Darcy over the past few days ;-)

@Dr.Edward Morbius quite! Also, those numbers are quite promising, even if we assume that only a fraction of those community members are still active. Got time for another chat this weekend?

@Christian Buggedei Yeah, getting a sense of activity in those communities would be useful. I can at least kick out a sampling of a few of the larger ones out there -- and that list is pretty small if you catch my drift, though it should give some useful inferences.

Good for a chat, ping by email.

Good work. This looks like a classic power law. So expect a very long tail. he interesting bit is going to be "The Fat Middle". That's the ~250,000 (wild guess) of communities with >250 people.

I assume you're getting member counts by wget-ing the community home page and scraping? Shame there's no easy way to get the total posts but are you looking at "date of most recent non-pinned post". The profile ID of the owners and moderators is out there but deep in javascript.

@Julian Bond Yes, this is straight-up web-scraping, a shell command (which I just tried to look for in ps output but cannot find), something close to:

cat sample | while read url;
    do w3m -dump $url |
        sed <bits to exclude>;
    done | tee communities-stats

Where sample is a file with 12,000 carefully curated (which is to say: randomly selected) G+ community URLs.

The output looks something like:

$ head -20 communities-stats 

>>> https://plus.google.com/communities/114856415048023462838 <<<
1 member
- Public
Bokep
Join

>>> https://plus.google.com/communities/117431085764762046475 <<<
4 members
- Public
YouTuber fans
Join

>>> https://plus.google.com/communities/109391514345612176114 <<<
10 members
- Public
Moçambique
Join

>>> https://plus.google.com/communities/114109821252528263148 <<<

(There are some bits that are messier, mostly from private communities). "Members" is any line matching ^[0-9][,0-9]* members. Public and private stats are read from that. There's a sequence which is followed for most of the interesting bits I've written an awk script to parse and dump.

There's also a discussion on G+, mostly here:

https://plus.google.com/104092656004159577193/posts/jgadFPDNvgE

(Also mostly with @Julian Bond, though that's OK.)

Preliminary complete results. That is: the data collection is complete, but my reports don't fully agree with each other (probably messy data and/or bad code).

Current sample (of 12,000):        12000
Total public:                      10830 (90.25%)
Open membership ('Join'):          6590 (54.91%)
Closed membership ('Ask to join'): 7745 (64.54%)

Member distribution:
n: 10358, sum: 1331992, min: 1, max: 217297, mean: 128.595482, median: 2, sd: 3301.906970
%-ile:  5: 1, 10: 1, 15: 1, 20: 1, 25: 1, 30: 1, 35: 1, 40: 1, 45: 1, 55: 2, 60: 3, 65: 4, 70: 5, 75: 8, 80: 13, 85: 22, 90: 43, 95: 118.5

Communities report
Total communities: 11992
Total public: 10821 (90.24%)
Total private: 1171 ( 9.76%)
Total open membership: 6578 (54.85%)
Total closed membership: 5414 (45.15%)
Total membership (public only): 1331992
Mean membership (public only):   128.60


Top 40 sampled communities
     1  217297 members
     2  166220 members
     3  126123 members
     4  124416 members
     5  33881 members
     6  27039 members
     7  26389 members
     8  26209 members
     9  23412 members
    10  22953 members
    11  19967 members
    12  17584 members
    13  16902 members
    14  12763 members
    15  11490 members
    16  10404 members
    17  10321 members
    18  9362 members
    19  8782 members
    20  7420 members
    21  7406 members
    22  6819 members
    23  6704 members
    24  6095 members
    25  5817 members
    26  5344 members
    27  5333 members
    28  5332 members
    29  5148 members
    30  5123 members
    31  4972 members
    32  4624 members
    33  4590 members
    34  4396 members
    35  4234 members
    36  4142 members
    37  3518 members
    38  3440 members
    39  3368 members
    40  3306 members

Each sampled Community stands for about 600 actual communities.

So there are likely:

• ~600 communities of 217,297+ members
• ~2,400 communities of 124,416+ members
• ~10,200 communities of 10,321+ members
• ~57,000 communities of 1,008+ members
• ~99,000 communities of 500+ members
• ~155,400 communities of 250+ members
• ~348,600 communities of 100+ members
• ~557,400 communities of 50+ members

I've given breakpoints as close to round values as possible given sampled data: 100k, 10k, 1k, 500, 250, 100, 50.

I'll give a post with links to the top 100 or so Communities when I get that reported out shortly.

https://youtu.be/1D9vt3XOT5s - actual video of Edward Morbius investigating G+

@Cade Johnson Cranial probe beats the other end any day of the week.

Discrepencies sorted:

• 8 communities were 404, the two reports read those differently.
• Public/Private communities also indicated differently in data.

No particularly material changes to results.

Updated:

Current sample (of 12,000):        12000
Total public:                      10822 (90.18%)
Open membership ('Join'):          6590 (54.91%)
Closed membership ('Ask to join'): 7745 (64.54%)

Member distribution:
n: 10358, sum: 1331992, min: 1, max: 217297, mean: 128.595482, median: 2, sd: 3301.906970
%-ile:  5: 1, 10: 1, 15: 1, 20: 1, 25: 1, 30: 1, 35: 1, 40: 1, 45: 1, 55: 2, 60: 3, 65: 4, 70: 5, 75: 8, 80: 13, 85: 22, 90: 43, 95: 118.5

Communities report
Total communities: 12000
Null (missing) communities: 8
Total public: 10814 (90.12%)
Total private: 1170 ( 9.75%)
Total open membership: 6586 (54.88%)
Total closed membership: 5414 (45.12%)
Total membership (public only): 1331812
Mean membership (public only):   128.67

I've also got an HTML-ized report (or Markdown source) but probably need to break that out, though interested parties can hit me up by email (dredmorbius at ye fine olde protonmail dot com).

@Christian Buggedei Ping.

I also wonder if there are so many G+ accounts because of Google's G+ comments integration with blogger. That still exists and no one seems to be talking about what may happen to the blogs using that integration.

Nobody from Google is talking much about anything to do with G+. There's lots of issues but a huge one is the large number of sites with some G+ buttons, integration, widgets, etc. And yes, comments on Blogger, but also links to G+ Profiles on Blogger.

It's a mess. Reminds me of Brexit. :(

@Julian Bond The solution to Blogger issues would be for Google to shut Blogger down. Problem solved! Another corpse for the cemetery.

@PresGas (OSR) Aspect Blogger is almost certainly a negligible contribution, and would be even if it accounted for tens of millions of users (unlikely). I suspect the vast majority of G+ profiles result from Android device registrations -- those require a Gmail account, which triggered a G+ profile from ~2012 - 2017 or so. That's when the service grew from a few tens of millions to 3.3 billion. It's added very few profiles since then, if the sitemap data are accurate.

@Dr.Edward Morbius, I wondered about that too. I do recall some of my phones having G+ by default. However, I also know that some apps, even if you have set up your android, will not actually create something like that (a G+ account) without actually opening the app.

I just was recently conscious of the G+ integration with blogger comments; just never really was thought about until Google said they were shutting G+ down.

@PresGas (OSR) Aspect Google's policies regarding automated creation of a G+ account have ... varied considerably over the past few years.

But the growth in userbase is almost certainly as I've described: Android device registrations. This is obvious from looking at Android sales data.

E.g., Gartner: Device Shipments Break 2.4B Units In 2014, Tablets To Overtake PC Sales In 2015, 2.2 billion G+ profiles noted in January, 2015 (my research, previously noted).

As for the discrepency between shipments and registrations, the TechCrunch article itself notes:

Note on terminology: “shipments” is Gartner’s classification for how many devices are sent retailers, carriers and others for sale to end users. You can think of them as closely correlated (but not exactly the same as) sales.

So: of 2.4 billion shipped, ~2.2 billion registered, or 92% activation ratio.

1Q16 vs. 1Q15 comparision in units shows 293m vs. 264m, or roughly a billion a year.

Google began decoupling G+ from other services in July of 2015, with the major break occurring in January 2016.

So, from 2014 - 2015 (inclusive), G+ saw roughly 1 billion+ users/year, which accounts for virtually all the 3.3 billion profiles noted as of 2017. Obligate growth stopped in January, 2016. Paradoxically, the profiles created after January 2016 presumably actually wanted to use the service.

@Julian Bond Agree a right Brexit type google balls up. What a mess. No loyalty to their users. I had thought maybe upload to Blogger (with the G+ exporter tool I saw on G+) many posts of a specific Collection topic, for instance, but if it may die too, then is that NOT a good idea.

Given the numbers in the federated web space prior to sunset of 1.8 million, the public reach of any federated web space user was half what it was on G+. However, given the growth of the federated web space to 2.5 million in the past three months, the reach is rapidly moving towards the pre-sunset reach of 3.5 million - even if many of those users are different.

As a G+ user, I am one of the remaining third still reaching to the G+ user base while also preparing to reach more into the federated web space as transition completes. My reach into the MeWe user base has been mostly ineffective likely because it is not public. My reach via Twitter will likely be far more effective than MeWe for sure. Friendica integration with various blog/CMS and Twitter is a plus.

These are observational notes based on my own experimental usage. I am considering a replacement for blogging, image, and video shares that are federated as they develop more. I suspect I will be looking closer at Hubzilla (for mini blogging), PixelFed (for images), and PeerTube (for videos). Respective alternatives now are: CMS/PHP blog, image, and video hosting. I do foresee multiple website replacements with federated solutions.

@Shelenn Ayres Taking note of your #tip that MeWe isn't public (I signed up but it looked a bit FB-y).

I was attracted by a note elsewhere that #pluspora links well into Twitter and Wordpress: which are my other main places; having started on Twitter 6 months ago (thankfully) without knowing G+ sunsetting.

I'm feeling #googleplusmigration #gplusexodus will even be fun: Collections, Communities to be sorted.

Taking note of the other places you will use. I'm starting a new website soon, and will integrate with all these new solutions, as I work it out.

That's my plan too. To replace all traditional web presence with effective positive technology federated solutions @ameliajhoskins

As for circles, collections, and communities: Friendica has those features already. Groups can be used like circles. Advanced account types can be used in place of collections and communities. Feature wise, Friendica surpasses both G+ and FB in what it offers. I originally chose Friendica over Pluspora for a few reasons: I can edit posts and comments I make, I have an integrated event calendar, and I can +1 aka like comments. My plan is to make a theme that improves usability along with implementing what someone else is doing to make it usable with Open Simulator communities.

@Shelenn Ayres Taking note on** Friendica**, especially if it includes editing, which I see this does not. An integrated Calendar (hopefully unlike Google as I don't need everyone knowing what I am doing! - maybe I could use this calendar as a personal tool to keep tabs on posts in the making, and recording frequency).

I don't have your platform knowledge but I know someone who does.

@ameliajhoskins There are around 24 Friendica nodes recommended so far by developers that take public registrations. I am on social.isurf.ca which has most features enabled. Some of the other nodes are actually testing a new release candidate right now so those could be buggy. One new node that shows in the public list that is still in development is opensim.fun so that is not a place for a home yet even though I have accounts there as well to help test it and establish those accounts used for Open Simulator activities. On the calendar, that is a big plus for me but know there are capabilities for private posting and messaging as well as public calendar features for things you want known.