Google is announcing all sorts of wonders for Android M (Macadamia Nut fruitcake, or whatever it is) at Google I/O.

At the moment, Lollipop (Android 5.x) is on 9.7% of devices – 9.0% for 5.0 and 0.7% for 5.1, according to Google’s developer dashboard.

So when you get the announcement of “permissions for apps” (or indeed anything that is M-only), you have to ask: how long will it be before that is actually widespread? And by “widespread”, let’s define it as “on 50% of devices”. That 50% is useful because 50% of the billion of so Google Android devices in use is roughly comparable with the total number of Apple’s iOS devices in use. The thing is, the overwhelming number of active iOS devices get updated to the latest version within three months of the release of a new version; since iOS 6 in 2012 it’s taken just one month for the number running it to pass 50%. (At the time of writing, in May 2015, the current figure is 82% of devices on iOS 8, 16% on iOS 7, and 2% on something earlier. Revisiting it in March 2016 to add grammatical edits, the figure is 79% on iOS 9, which was released in September 2015.)

Rapid adoption is a key element of Apple’s iOS due to its direct update mechanism.

How do we figure this out? We let history be our guide. I’ve been collating the data about versions on the Android developer dashboard for a while, so we can look back to the past. In each case I’ve taken the beginning point as when a version first showed up on the dashboard, not when it was “released”.

The release of a new version of Android doesn’t necessarily mean it reaches a large proportion of users quickly.

If we take it that “modern” Android starts with version 4 onwards – given that 2.3 (Gingerbread) was super-old, while 3.0 (Honeycomb) was tablet-only, we get this data:

Android 4.0: 16 months for it, or a later version, to be on 50% of devices according to the dashboard (January 2012 to April 2013).

Android 4.1: 13 months (October 2012 to November 2013).

Android 4.2: 22 months (December 2012 to September 2014 – the date of Lollipop’s release, officially).

Android 4.3: 18 months (October 2013 to March 2015)

Android 4.4: 19 months (forecast). Launched in December 2013, at the start of May 2015 it was at 49.5% for 4.4 and successors. It’s a safe bet that in June 2015 the total for Android 4.4 and successors will pass 50%. (Update: this was exactly right: the figure for May 2015 showed 49.5% on 4.4 or later; for June 2015, it was 51.6%, including 5.x; Android 4.4 itself peaked at 41.4% in April 2015.)

Android 5.x: 18 months (forecast). Presently, it’s 9.7% after 4 months, and its uptake pattern is more like 4.3 than 4.4. My forecast for its 50% point: July 2016.

On that basis, I’d say it’s a pretty safe bet that if Android M is released in December 2015 (which is likely) that it will take until June 2017 before it’s on 50% of Android devices according to Google’s measurements. Of course, that will vary regionally – there are still 6% of devices running 2.3 or earlier, which translates into about 60 million devices still in use. Some will get more rapid takeup, some will get less.

To put it into perspective (thanks Mark Blank-Settle on Twitter):
• presently, Apple’s offering iOS 8. It will show off iOS 9 in June.
• by the time Android M is released on devices later this, iOS 9 will already be on 60%+ of iOS devices
• by the time Android M – shown off in early 2015 – is on 50% of devices, Apple will have shown off iOS 10.

So if you’re depending on something such as, oh, the permissions model or Android Pay being introduced in Android M reaching the majority of your users any time soon, might be best to hold your breath. The revolution looks more like an evolution.