Windows 10 turned five years old this week, and after going through each feature update for the OS, I got to thinking about how Windows as a service has evolved over the years. When the OS first shipped, we didn't even know what the next update would be called; some speculated Windows 10.1. It was a couple of years before Microsoft decided on its biannual spring/fall schedule, and that was so it would align with Office 365 ProPlus.
There have been 10 versions of Windows 10, and a number of them have been problematic, some more than others. However, version 1809 was the biggest disaster of them all, and it inspired some significant changes to the process. Starting with version 1903, users were no longer forced to take automatic feature updates, and then version 1909 arrived in the form of a cumulative update with no notable features.
That extremely condensed version of history brings us to today. Windows 10 20H2 is set to arrive in the same way as 19H2 did, as a cumulative update. And while we expected 21H1 to be a full feature update, it turns out that that might not be the case.
To be clear, Microsoft has never committed to a major feature update in the spring and a minor one in the fall. It's just what happened in 2019. What I was originally told was that Windows 10X was set to RTM this fall, so that's why regular Windows 10 is getting a minor update again. But now, Windows 10X is delayed until next spring, leaving 21H1 in question.
One thing that I've absolutely heard is that 21H1 will certainly not be the major feature update that we were expecting. I don't know if we'll still get a minor 19H2-style update, but ZDNet's Mary Jo Foley is hearing that 21H1 won't happen at all, as Microsoft is shifting to an annual release schedule.
I'm here to say that Microsoft shouldn't do it, and that it should stick to feature updates twice a year. Give us one major update in the spring and one in the fall. It's fine, really.
I understand the argument. The idea is that Microsoft has fumbled so many of its 10 versions of Windows 10 that it just can't handle pushing out two feature updates a year. I don't think that's the case, and here's why.
The second update is the good update
First of all, between the two updates a year that we get, Microsoft is killing off the good one. Let's be clear about that. 19H2 can barely be considered a feature update, but the fact is that it was a good feature update. In fact, if you went directly from 1809 to 1909, and were planning on going from 1909 to 20H2, you're in pretty good shape in terms of stability.
Think about it. If Microsoft pushes out a bad update in the spring – something that's going to happen from time-to-time – the fall update fixes it. That spring update gets serviced for six months, Microsoft throws in a couple of minor goodies, and that becomes the fall update.
Bad feature updates are going to happen from time to time, and keep in mind that for consumers, these updates are supported for only 18 months. That means that you're not going to be able to skip feature updates like you can now. In fact, if Microsoft kept things the way they are now, where you have to opt into feature updates and they're only installed when your version is nearing the end of support, these annual updates would end up coming through automatically anyway.
To be clear, this is only if the current system stayed in place, and the only thing that changed is that Microsoft only made one feature update a year. I have no reason to believe that support lifecycles wouldn't be extended. The facts that remain, however, is that bad updates will happen from time to time, and that Microsoft would be killing off the update that fixes them.
You can't guarantee a good spring update, as demonstrated by the May 2020 Update
Windows 10 version 2004 is not a good update. When it launched, it had a bunch of known issues, and even now, over two months later, a whole bunch of PCs are blocked from getting it. This wouldn't be so crazy, if it wasn't for how hard Microsoft tried to make it good.
Microsoft began testing Windows 10 20H1 with Insiders on February 14, 2019, 15 months before it was released. No other update has been publicly tested for that long. That's not all though, because the RTM build, 19041, was released to Insiders on December 10, 2019. After that, it was serviced for another five months before being publicly released.
No one could have guessed that there would be so many issues. Servicing 20H1 for five months seemed like a good idea. Many updates are rough on day one until they've been serviced for a few months. So, why not just service them for a few months before shipping it? It didn't work. On day one, you'd be hard-pressed to find a PC that even got offered the update.
Between when Insiders started testing 20H1 and when it was released, two Windows 10 feature updates were released: versions 1903 and 1909. But 1909 was barely a feature update. Off the top of my head, I can't think of a single new feature that was in it, and there was certainly nothing major in it.
Version 1909 is a pretty simple idea. It's a refined version of version 1903 but with a different lifecycle, and a longer lifecycle for business customers. I also opined that version 1909 was something that users should opt into. With the way things work now, Microsoft automatically updates users to the latest update only if your version is nearing the end of its 18-month life, so you'd end up on an annual cycle anyway. If you opted into version 1909, you should be on the more stable H2 update path every year.
But given how simple and minor version 1909 was, are we to believe that had version 1909 not happened at all, that version 2004 would somehow be better? I really don't think that it would.
The point that I'm getting at is that that second feature update of the year, the more stable one, does a lot more good than harm, and Microsoft should keep it around.