There is very little now in RISC OS which is unique. Once we could claim to have the stability, even though RISC OS (through Select admittedly) has become far more stable than it ever has, it is still lacking as there is no real memory protection (other than page 0). The UI no longer holds the same level of ease as all three of the other main OSes now have this level.
Even the claim of OS size being held on ROM doesn't hold much water. Remove kudzu from a Linux box or have a correct install of 2000/XP and they will boot quicker than RISC OS to the desktop. Yes, it's a big advantage to have the OS in ROM (not going to be corrupted), but it's not a big enough advantage.
Other than Impression, there is very little to recommend survival and survival based on a handful of applications is not enough
There are probably two ways to get folks back. One would be a *massive* reduction in prices - I mean an Iyonix/Omega sub 700 UKP and a *massive* investment into writing software with *all* developers pulling together, irrespective of personal grudges.
The second is VARPC. AIUI though, only MD will be able to do this on a laptop which is frankly overpriced and not particulary well spec'd.
The third (which I've just thought of) would be to target those with OS 3.1 machines and offer them a large discount to change to OS 4 (say one of the ARM7500FE based machines for 250UKP + their old box). Unfortunately, that's been tried and failed.
 I'm excluding porting efforts here - this is down to the uniqueness being referred to. A port, while really good (don't get me wrong) is not going to draw people either back or in. The likes of Vantage is the sort of unique software I'm on about.