Given how utterly cool plan9 is, I have some difficulty understanding how they so completely failed to find a userbase. I sometimes have wishful thinking that maybe google or someone like that is secretly running on plan 9 and that in time it will permeate out. But I think not. Anyway - not too important - the OS layer is becoming irrelevant (and yet when my parents asked what on earth I could want for my 30th I asked for a book on ncurses :) ).
Given how utterly cool plan9 is, I have some difficulty understanding how they so completely failed to find a userbase.
I think the reason it never became popular is that the existing popular operating systems have been (to this point) good enough. There just hasn't been a compelling enough reason to invest the time and energy in switching. I think the only way you'll see a large switch is if a "killer app" comes out which takes advantage of Plan 9's features which wouldn't be possible on other OSs.
I don't think things win market share by how cool or powerful they are - e.g. Betamax vs. VHS OR SPARC vs. x86 OR Haskell vs. C. Why something becomes popular has to do a lot with timing, price and creating something that's "good enough".
Other than this, plan9 is/was really cool and has a lot of interesting concepts (like a distributed architecture).
Minor nit: Betamax is a standard, but flawed example. The original betamax cassettes only worked for 1 hour, but movies last 1.5-2.5 hours. They eventually fixed it (at the cost of lowering quality), but too late.
And it's another example of a win not being necessarily what yo hoped for. Ok, Sony didn't get a Beta machine into every home as they had hoped, but they got at least one into every TV station.
Legend has it that after AT&T stomped all over their fun little research OS by turning it into a popular product and all the tedious concerns like "backward compatibility" that entails, the creators of Unix got together and said, 'right, we'll make a new OS and give it the least marketable name we can imagine so that it will never be taken away from us and ossified'.
Although I've never used Plan 9 myself (I don't think I've ever owned hardware it would boot on), I am impressed by the features I've read about, and I wish it had become more popular... but on the other hand, I'm quite happy with the chording-free and keyboard-friendly UIs I have available.
So why exactly is the OS layer becoming irrelevant? Is this that idea that everything in technology will henceforth occur in a Web Browser ... perhaps in some HTML4 compatibility mode?
I think the modern idea of a "cloud" is the Plan 9 philosophy coming full circle. Many breakthroughs in Plan 9 (distributed architecture, everything as a file, unified user experience) are being recreated on the web (cloud storage/photos/music/email, open APIs so all your webapps can intercommunicate--when I squint really hard while looking at screenshots of Plan 9, it almost reminds me of google's suite).
The OS will become extinct once everything happens in the web browser, and the OS becomes not much more than a web browser itself. And then they'll have pretty much reimplemented Plan 9.
I think that this couldn't and shouldn't happen.
It couldn't happen because of the services are provided by third parties.
First open apis vary greatly from one service provider to another - and this differs _vastly_ from the plan9 idea of unificating everything as a file. And do you really expect google apps to communicate with their microsoft counterparts ?
Second, I could hardly describe the web as having a unified user experience.
Third, and this is the most important thing, is that you're describing a closed system. What if I want to make something my service provider didn't think of ?
Plan 9 does have a userbase, albeit small it is >50
We have a GSoC entry http://gsoc.cat-v.org/ this year. Plan 9 is also used in some Lucent / Alcatel cell phone masts in RT mode. I know of three commercial companies with Plan 9 / Inferno based products.
Plan 9 was a for sale product when free Unices became the rage. When it became a free download, Plan 9 landed without a modern web browser in the days of the Web and we still don't have one. Until recently, with the addition of an AC97 driver, the only way to get sound was an 8 bit Soundblaster or difficult to find specific models of USB audio device. There's no video playback (unless you don't care if the sound syncs). So you end up needing two machines and only one of them expects to be just another node in the network and the other thinks it's the sentinel.
I'm surprised that acceptance of the browser is so global, but particularly in a community like yours. HTML is dirty, HTML+CSS+javascript is dirty. Plan 9 folks know about tool simplicity, and about nice documents. I've always half expected something to come from your team that says "no - this is all wrong. The browsable internet should be like this" that gives the middle finger to the HTML world, and which can easily be replicated as a firefox plugin and then go on to take over the world.
I guess I'm saying that I think you took the wrong path by attempting to follow the browser - you should have challenged it, but in a way that didn't rely on plan9-specific user interface innovations [1].
I've had several attempts at it but don't hang around in circles where I get to talk about the problems regularly face-to-face with other people who get it, and that's what a problem like that requires.
The site you've pointed me to is a real improvement over the online docs that were around two years ago. It answers the sort of questions a newcomer asks.
--
[1] I think sam would have gone further if the console tool had been the core with the gui as a separate app that leveraged the core. Console sam should be to ed as vim is to vi, but it's trying to do two things at once, and the gui editor - which I'm sure is very good for people who know it well - hampered the progress of the console app.
We're always saying to each other, HTML + CSS is wrong.
We like PDF & PS & troff and plumbing on http://example.com/cod.pdf will open it in the pdf viewer.
fmthtml turns web pages into plain text.
But ignoring the majority of web content is really not an option. HTML can be pretty lightweight, we do have some renderers that do HTML3 and a subset of CSS really well and there's also Abaco that does a good job with HTML4, we've even got a canvas widget!
We could probably port WebKit if only we had a native C++ compiler but the incentive to do that is pretty low, we've got other cool beans to play with and you can pick up a dedicated Web browsing machine is less than the price of a round of drinks.
They got rid of the TTY, huzzah!