The RISC OS user base has always been a bit schizophrenic. Even in the 90s - the period when the OS had its greatest mainstream success - RISC OS users were a diverse bunch. Most OSes had a clearly defined stereotypical user: Windows was for business-types, Macs for design gurus, Amigas/STs for gamers and Linux/UNIX for developers and academics. RISC OS never had a clear rationale. Many users came from the education sector, others from the scientific community, and a few were home computing enthusiasts.
 
Despite the fact the community has shrunk rapidly over the years, its seems to me that the divide is still with us. There are essentially two types of RISC OS user. The first is the thoroughly traditional non-technical type: perhaps a retired teacher, probably quite 'mature', with a decent knowledge of BASIC and the workings of the Desktop, but no advanced programming skills or interest in the more esoteric side of development. The second is the hobbyist developer. Maybe an ex-IT student, perhaps employed as a programmer in the computing mainstream, they will be comfortable with C/C++, build scripts, open source politics and all the other features of the current IT world.

Two Worlds

The problem for RISC OS is that the two groups have little (superficially) in common with each other, and their needs can seem to be pulling in very different directions. Take software distribution. Traditionalists are much more comfortable with the old model of paying for closed-source applications. This meets their needs - there's no need for potentially confusing downloads or manual building of libraries. In theory, commercial products ought to be fully stable. If not, then the paid-for model normally includes proper support, either by telephone or via a mailing list.
 
By contrast, newer users are much more comfortable with open source software. Being comfortable with downloading resources and making sure dependencies are in place, they can take advantage of the faster development cycle of such software. If something goes wrong, they're likely to be confident about reporting it correctly to the right person, or even fix it themselves. The 'polish' of the application is probably less important to them than the function it performs, and they're likely to have strong opinions on the kind of licence the application is distributed under.

The Open-source Shift

At a guess, I'd say that for most of RISC OS's history, the former group has been more influential. Notwithstanding the excellent PD software of the past, major RISC OS software has tended to be of the commercial, out-of-a-box type. Even PD software tended to be paid-for, in the sense that PD Libraries handled the shipping of catalogue disks. The lack of updates over the internet meant that free software needed to work in a similar way to commercial offerings - beta quality editions were rarer. And, of course, the OS itself has been closed-source until very recently.
 
But this is changing. There are very few commercial applications left. MW Software, David Snell, R-Comp are the only three companies/individuals still developing products that come readily to mind. Even RISC OS itself is moving from a commercial product to a freely-available one. ROOL, of course, distributes the sources and pre-built components for RISC OS 5, but ROL also sells the 'virtually free' edition of its closed-source branch.
 
So the future, to the extent RISC OS has one, looks like being open-source. The geeks have triumphed over the luddites, and in future we're likely to download our OS updates in a way similar to Linux distributions. Is this good news? Well, in the sense that it's the only conceivable way forward, yes it is. You just have to look at the progress made on the shared-source OS, NetSurf and Firefox to see the power of open development. Even some of the commercial software for RISC OS has an open source heart - such as Music Man from R-Comp.
 
But care needs to be taken here. I'm guessing again, but would be pretty sure that the bulk of remaining RISC OS users are a pretty traditional bunch. Things that seem obvious to a seasoned developer can be utterly baffling to a non-technical user. You just have to look at the frequent misunderstandings and frustrations when the two get together, such as on the comp.sys.acorn newsgroups and product mailing lists. Developers can often find the more traditional kind of user immensely frustrating, and the feeling is wholly reciprocated. At their worst, users expect the moon on a stick from free software, fail to follow simple installation instructions, and insist on using hopelessly out of date hardware and OS combinations. At their worst, developers can be intolerant of honestly-asked questions, too opaque with any documentation they provide, and dismissive of the real difficulties some users have with modern ways of distributing and maintain software.

A Happy Compromise

I think it's likely that both kinds of users will continue using and enjoying RISC OS, and it would be sad to see either group feeling that the platform was no longer for them. There are good examples of excellent software from both sides. MW Software's excellent commercial offerings will hopefully continue being produced and supported, and show how capable and popular closed-source applications can still be. On the other hand, premier open source applications like NetSurf have made real efforts to make installation and use as simple as possible for non-technical users. Their website is a model of clarity and professionalism.
 
So even though the future of RISC OS looks brighter for fans of the more collaborative, open-source model of computing, there's room for the OS to remain the broad church it's always been. Personally, I'd not wish RISC OS to go the way of Linux distros, where a certain degree of technical expertise is generally expected in order to make best use of the desktop. With any luck, open-source RISC OS will remain the rather strange hybrid it's always been, appealing to both seasoned hackers and retired geography teachers alike.