Windows Insider Program = HILARIOUS!
OK Microsoft. April 1st is long past. You can stop now.
I'm talking about the fast ring Windows Insider builds of Windows 10 of course. I think the slow ring has largely done it's job and never really had any issues. The bar for a release going to the slow ring is fairly high and even on a good streak you're only getting an update once or twice a month. You still get early access, but you also get fairly stable builds. I think users can sign up and within a few builds know what to expect and won't be disappointed if they decide to stay.
The fast ring on the other hand... makes no sense at all. I've long argued that they need another ring.
In the past, they would promise faster builds, and for the next 2 builds you'd see one every week or two. And then progress would drop off a cliff and some times you wouldn't see a build until it was the next one the slow ring got. And today, the fast ring seems to get 3+ builds a week. THAT is ludicrous.
I use my PC a few hours a day. It spends more time installing fast ring releases than actually running. And I know that once the Anniversary Update is out, it'll drop back off the face of the earth.
The current fast ring doesn't serve anyone's purposes ideally.
Broken record time; Microsoft needs to step back and define what the true purpose or each ring is. And in doing so, they will probably discover that they don't have enough of them.
I still think Microsoft needs a minimum of 3 rings. As stated above, the slow ring today is fine. Even without a formal definition, the slow ring is consistent enough that users can quickly and easily draw their own conclusions.
The fast ring is amorphous and meaningless today. Near a new release, build comes fast and furiously. After a new release you twiddle your thumbs for weeks. To those actually interested in helping, neither scenario is all that useful.
If they're spaced too far apart, features get developed without feedback and over time there are diminishing returns on the amount of bug and other feedback data available. If they go too fast, then new builds are pushed out before the old are suitably tested. In fact, this last week it feels like I turn on my computer, it downloads a new build. It takes so long to install that I shut it off when done, and just do it all over again the next time I turn it on.
Sure, there is value for rapid fire releases as well. But most of the data is going to be about low level hardware issues. The users simply don't get enough time to test the rest of the system.
I think the fast ring should be defined as a ring aimed at users who don't want a high risk of a broken build, but are willing to gamble on it a little to get a build in time to have their feedback potentially incorporated. To that end, I think a few things need to happen. I think they need to lower the bar for fast ring builds, but at the same time put soft min and max limits on the timeframes for new builds. For instance, target of every 2 weeks with a min of 1 week and a max of 3 weeks. Sure, if there isn't a single build in that timeframe which meets the new lower standards, it should be pushed out. But even the average tester can't give a build a good run through in less than a week.
Lastly, a third ring should be added for the "ludicrous" speed. Which is for those with the patience and means to handle rapid fire releases. The bar here should be extremely low. As I said before, potentially even requiring some sort of additional registration to prove the user has an extra device which they can use to help them restore a dead machine and proves that they know how to do so or forces them to either validate that they have a valid restore disc/key or create one. Even here, some min/max might be worthwhile, like a build every 1-3 days with a target of 2.
This group would potentially see things the soonest and thus have the most timely chance for feedback. But they'd largely be helping vet the stability of the new builds on a broad array of configurations. More than 1 build a day though and you're just destroying bandwidth and probably not getting a ton of information.
I'm talking about the fast ring Windows Insider builds of Windows 10 of course. I think the slow ring has largely done it's job and never really had any issues. The bar for a release going to the slow ring is fairly high and even on a good streak you're only getting an update once or twice a month. You still get early access, but you also get fairly stable builds. I think users can sign up and within a few builds know what to expect and won't be disappointed if they decide to stay.
The fast ring on the other hand... makes no sense at all. I've long argued that they need another ring.
In the past, they would promise faster builds, and for the next 2 builds you'd see one every week or two. And then progress would drop off a cliff and some times you wouldn't see a build until it was the next one the slow ring got. And today, the fast ring seems to get 3+ builds a week. THAT is ludicrous.
I use my PC a few hours a day. It spends more time installing fast ring releases than actually running. And I know that once the Anniversary Update is out, it'll drop back off the face of the earth.
The current fast ring doesn't serve anyone's purposes ideally.
Broken record time; Microsoft needs to step back and define what the true purpose or each ring is. And in doing so, they will probably discover that they don't have enough of them.
I still think Microsoft needs a minimum of 3 rings. As stated above, the slow ring today is fine. Even without a formal definition, the slow ring is consistent enough that users can quickly and easily draw their own conclusions.
The fast ring is amorphous and meaningless today. Near a new release, build comes fast and furiously. After a new release you twiddle your thumbs for weeks. To those actually interested in helping, neither scenario is all that useful.
If they're spaced too far apart, features get developed without feedback and over time there are diminishing returns on the amount of bug and other feedback data available. If they go too fast, then new builds are pushed out before the old are suitably tested. In fact, this last week it feels like I turn on my computer, it downloads a new build. It takes so long to install that I shut it off when done, and just do it all over again the next time I turn it on.
Sure, there is value for rapid fire releases as well. But most of the data is going to be about low level hardware issues. The users simply don't get enough time to test the rest of the system.
I think the fast ring should be defined as a ring aimed at users who don't want a high risk of a broken build, but are willing to gamble on it a little to get a build in time to have their feedback potentially incorporated. To that end, I think a few things need to happen. I think they need to lower the bar for fast ring builds, but at the same time put soft min and max limits on the timeframes for new builds. For instance, target of every 2 weeks with a min of 1 week and a max of 3 weeks. Sure, if there isn't a single build in that timeframe which meets the new lower standards, it should be pushed out. But even the average tester can't give a build a good run through in less than a week.
Lastly, a third ring should be added for the "ludicrous" speed. Which is for those with the patience and means to handle rapid fire releases. The bar here should be extremely low. As I said before, potentially even requiring some sort of additional registration to prove the user has an extra device which they can use to help them restore a dead machine and proves that they know how to do so or forces them to either validate that they have a valid restore disc/key or create one. Even here, some min/max might be worthwhile, like a build every 1-3 days with a target of 2.
This group would potentially see things the soonest and thus have the most timely chance for feedback. But they'd largely be helping vet the stability of the new builds on a broad array of configurations. More than 1 build a day though and you're just destroying bandwidth and probably not getting a ton of information.
Comments
Post a Comment