It would be nice if Google endeavoured to keep old Android working and security patched, rather than force a faster e-waste cycle.
By that logic we'd still be supporting the Space Shuttle, the NES, the Sega Genesis, the Apple II, the Altair, the 80286, the Laserdisc player, the Video CD, the Kodak cameras... We should never drop legacy, ever.
Plus, your Nexus S actually got 4.1.2, which means it's still supported even after this.
While it may be 0.2%, it'd be interesting to see what numbers that actually equates to.
Versions up to 8.0 constitute around 20% of the phones (as of June 2021¹), which creates a significant amount of e-waste, particularly embarrassing when considering phones like the Nexus 5x, which is a perfectly usable model.
[¹] https://www.statista.com/statistics/921152/mobile-android-ve...
But they can still access Google.
Edit; if there are 3 billion active android devices, this is putting 6 million into landfill?
Honestly, if having every other app than Google drop support for your device didn't motivate you to upgrade, Google dropping support probably won't either.
Importantly: they are repairable.
My Nexus S is a paperweight because Google kept it relatively closed. My BlackBerry Priv is even worse, since I can't root it.
Unfortunately, the reality with online devices is that:
a) They are sufficiently complex that they have an indeterminate (and effectively unending) supply of vulnerabilities
b) Effective vulnerability management requires vendor support
c) The market favors one-time ("sale") models, not ongoing ("subscription") models
As a result, vendors support devices for a fixed amount of time, often subject to arbitrary constraints (like going out of business or unexpectedly high costs of fixes).
From a consumer perspective, this usually manifests as computing devices getting worse over time--less performant, less secure, and surprising and arbitrary end-of-lifing.
You might argue that a more open device ecosystem allows open source contributors to fix vulnerabilities. This is true, to a degree, but only to a degree--just as (past some point) it isn't worth it for a vendor to keep patching a long-obsolete product with a tiny userbase, it's similarly true that an esoteric open source product won't attract a robust community that can meaningfully support it.
While free software advocates like to say "But at least it would be up to me to patch it," in practice, for most users (who aren't going to write their own patches, or hire a team of professionals to do so!), predictability, transparency, and fairness are the critical components in the products they use, not "openness" per se.
If phones had followed a similar open architecture as PCs then they might have enjoyed such support, as well.
i386 is kept as the arch name, but the baseline has went up.
Jessie needs at least a Pentium, and Stretch needs an i686.
Because it is already forced in multiple ways and most devices which can run only 2.3 are already on landfills.
Google+ userIDs were for a long time effectively new Gmail and Android registrations. Those peaked at about 3 billion before the automatic assignment of a G+ account was broken in 2015. Conservatively, there are likely well over 6 billion such registrations, which would be 12 million Android 2.3 devices.
Small fractions at Google scale are large numbers.
Realise too that Android shows up in numerous devices, including some embedded tools, as well as tablets, e-readers, and other systems a casual reader might not be aware of.