If Canonical didn't spend so much time engaging in Not-Invented-Here fantasy 5D chess, and instead focused on advancing the emerging standards, then they would have had more momentum. By working against the community (e.g. Mir [especially the Wayland FUD]), they cost themselves time and enthusiasm.
I think this had a lot to do with the cancelling of Unity 8 as well.
Thinking back, I'm impressed with how far they got. If they had collaborated where it didn't help to compete, then I think they would've had a better chance of entering the market. I hope that this experience doesn't steel them against working on ambitious things entirely in the future.
What community? Is there really any Linux community that speaks with one voice?
'Community' seems to be a pretty nebulous ill defined term in this context used to prop up whatever argument the poster happens to be making.
If Ubuntu wants to invest in Mir how is that a bad thing, if Redhat invests in Systemd inspite of a multitude of other init options how is that a bad thing? This is usually how progress happens in the diverse and uncontrolled open source ecosystem and failures are part and parcel of this. Using words like 'NIH' seems misplaced here and sounds like some people want to 'control' what other people can do.
For instance who gets to decide what is NIH, given Ubuntu's Snappy is Fedora's Flatpak NIH? But there is no 'community' antagonism towards Flatpak. There appear to be some double standards at play here.
Snappy's Git history starts at basically the same time (literally one week apart) as Flatpak, and both of them only became viable recently. Snappy is written in three languages (Go, C, and Python) and two of those are not as popular for Linux low level infrastructure projects, Flatpak is written in one language. Flatpak was developed by an independent developer, Snappy was developed almost exclusively at Canonical it seems. Snappy is also GPLv3, where Flatpak is LGPL; while this is a minor issue for workstation and server users, on kiosk/embedded systems the anti-tivoization requirements of GPLv3 can be an issue.
Snappy has higher memory requirements, takes up more disk space. Flatpak is implemented in about half as much code (granted with slightly different feature sets) despite being written in C rather than Go.
And despite these signals, I don't see any broad community "antagonism" towards Snappy at the moment. You can find an angry blog post on almost any topic, doesn't mean the opinions are popular. I don't feel any particular way about it since it seems to do something at least a little different than the competition. For example, Flatpak only really tries to sandbox desktop applications, much like .app directories on OS X.
It is a little suspect that Canonical always shows up with their own version of something which everyone else (SUSE, Arch, Red Hat, [Oracle/Novell]) mysteriously can't even manage to build and package successfully. In the case of Snappy, I think it's more or less a coincidence that it started within seven days of Flatpak. It might turn out to be better (or serve a completely different purpose). Another thing in favour of Snappy is that at least snapd and snap-confine (but not snapcraft!) are widely packaged for distributions other than Ubuntu.
The thing is, the rest of the industry is running on independent contributors. These independents manage to widely distribute and standardize on the best. Meanwhile, Canonical consistently goes off on their own and builds something that ultimately nobody can even package (Mir, AppArmor, Unity), let alone adopt.
Canonical clearly wants to differentiate, and that's entirely their right, but their differentiation factors often seem to wither and die after a couple years of bragging and some small progress.
I got the impression that everyone who didn't work for Canonical thought Mir was a bad idea, so that was a rare case where there was pretty much one voice. IIRC, the best case for Mir (if it had shipped on time with no bugs) would have been basically at parity with Wayland, which was started years earlier. This justifies claims of NIH.
The Snappy/Flatpak situation is different because each system has its own advantages and disadvantages and Snappy didn't come years after Flatpak.
You whole assumption rests on your 'personal impression'. You simply can't know what other people in the open source world think or presume everyone agrees with you. To claim 'one voice' seems a rather large stretch. There are tons of Ubuntu users and others in the ecosystem who may think differently.
What makes you so sure Mir would not have some advantages? It seems impossible to say either way. For all we know it could offer benefits and choice to the ecosystem. Or die because users would reject it.
Open source has always been about diversity and choice. There is now an unfortunate trend of some commentators misusing words like 'community' to push their own agendas or trying to pass off personal opinion as a widely held fact. This basically challenges open source diversity.
The "tons of Ubuntu users and people who support diversity" are completely unimportant in both the Mir/Wayland and systemd cases. They don't contribute anything (except maybe a lot of drama), don't have the technical knowledge to decide anything and don't have to support anything. They're just spectators shouting from the tribunes.
systemd was resented by some users, but for developers it had so many advantages (especially stuff like logind) that a huge part of them just started happily using it.
investing in systemd or mir is one thing. what the post you replied to say is that they were buildind their alternatives from scratch, like Xservers used anywhere else but by them.
I think this had a lot to do with the cancelling of Unity 8 as well.
Thinking back, I'm impressed with how far they got. If they had collaborated where it didn't help to compete, then I think they would've had a better chance of entering the market. I hope that this experience doesn't steel them against working on ambitious things entirely in the future.