We can argue until the cows come home about ARM based machine vs commodity hardware and most of the comments above have merit however the licensing of the OS is, I believe, still the major obstacle.
The current situation has left us all victims of chance and the whim of a few buisnessmen, some of whom knew very little about RISC OS itself. I would relect that since Acorn's demise I've spent more on things that looked hopeful for RISC OS but which turned out to be nothing of the sort than on products that still exist. How many other people bought Psion netbooks because of the rumours of RON? How many other people bought a copy of that graphics package that simulates the effects of the inks used in particular printers (the name escapes me now but it seemed to have great potential at the time only to be unusable now due to activation requirements not to mention the bugs! - I can well understand the annonyance many people are showing towards VA's similar choice). Even issues of the number of developers are related to licensing. Why write/update an application for a platform which could so easily disappear? Contrast this with the fact that we can be very comfortable that one thing related to RISC OS, support for the ADFS format, will survive for the forseable future - it's part of the kernel.org linux distribution.
I don't mean to be pessimistic - comments over the last few weeks have given me the first glimmer of hope that more future-proof licensing is possible.