Linux is free. OS/2 was strictly more expensive than Windows, and required you to buy a copy of Windows to run the Windows apps.
Like it or not, Linux was effectively second to the party. We faced a chicken and egg problem, and had no first-mover advantage. We simply weren't strong enough to cajole users into defaulting with us because we could hold all their favorite apps hostage if they didn't. We had no apps, legacy or otherwise.
We tried to get people to want to migrate badly enough by excelling in other areas (like being free). But our biggest problem was that they couldn't come even when they wanted to: 80% never heard of us, 80% of those left had hardware issues, 80% of those who remained had an application with no native equivalent. 20% of 20% of 20% is 0.8%, which incidentally is the exact market share we had.
But we're solving those issues now. People are becoming more aware of the viability of the platform. Distributions with commercial backing and their hardware partners have been solving the hardware issues. But there's still that application barrier for migration. Wine is a huge part of the solution to that.
There were lots of DOS apps. A few windows apps. And no OS/2 apps.
Windows always had compatibility AND reasonable system demands. On a windows 3.0 supporting machine you could not run OS/2. On a windows 3.1 machine you could -badly- run OS/2, but not OS/2 2.0. And on a windows 95 supporting machine you could maybe run OS/2 2.1.
OS/2 is an example of why design by committee loses out to a technical person really controlling the development of an application. OS/2's code, I'm betting, is far more flexible than the windows code, and has support for all sorts of weird features that no-one but academics care about, and massively increase demands on hardware. They also constantly require rewrites to actually make use of these features.
OS/2's failure had nothing to do with application support. If anything that was one of OS/2's redeeming features.
I like to think that the first thing you should do to make something succeed is to make it possible for said thing to succeed. OS/2 did not make it possible for itself to succeed, because nobody could run it. Somebody should have stopped the software architects and forced them to change until that happened. Nobody did.
The academic community was really proud. They even pushed it into the huge IT environments, like banks. No-one ran OS/2 for fun, because that was a $5k proposition. Running a windows 95 was a $500 proposition, that also got you access to duke nukem 3d and loads of other games.
Both ran well enough to run wordperfect on a 286 with 1 meg of ram. OS/2 wouldn't even boot under 4 megabytes of ram, or with a 286 processor.
So yes, vastly more expensive hardware requirements is certainly a valid criticism. In 1992 nobody had 386 machines. Some even say IBM purposefully designed it like that, to sell more machines. But if they did indeed intend that, why not release the OS for free ?
In 1992? I think this was after for example the infamous AMD 386DX-40 was released. Granted there probably still were a lot of 286s out there back then of course, which is one of the reasons why I never suggested abandoning DOS/Win3.x immediately. Of course, as mentioned before this would cause some pain for application developers, but this would last only a few years, and even PX00307 mentioned "Porthole" (aka WLO).
By 1992, 386 machines were common, and new machines for serious work were generally 486s.
I've been a fan on Wine since I tried it, but I think this article on OS/2 from Arstechnica argues very well, on a historical basis, for why Wine unfortunately won't be helping getting people to use Linux on a large scale.
http://arstechnica.com/business/2013/11/half-an-operating-sy...
"OS/2 ran Windows apps really well out of the box, so they could just write a Windows app and both platforms would be able to run that app. On the other hand, writing a native OS/2 application was a lot of work for Windows developers."
"The second lesson of OS/2—to not be too compatible out of the box with rival operating systems—is a lesson that today’s phone and tablet makers should take seriously. Blackberry once touted that you could easily run Android apps on its BB10 operating system, but that ended up not helping the company at all. Alternative phone operating system vendors should think very carefully before building in Android app compatibility, lest they suffer the same fate as OS/2."