I'll throw my 2 cents in here...
AROS is absolutely usable, looks great, performs well, etc. It has for a long time now. That said I think all development needs to cease, and be re-focused on 3 vital projects:
Modern web browser (started, but needs more hands)
JAVA runtime environment (I see no mention of this anywhere)
Multi-IM client (at least Yahoo and G-Talk)
Take care of these projects and everything else will fall into place quickly. Give us web access and chat capabilities. Getting JRE working we'd instantly have a lot more applications to run.
This gives us BASIC usability using the old PCs most of us already have. With that, more programs will be written as more people will be interested in it, some of them programmers.
Look at BeOS, how much stuff was written for that before it tanked? Just because people were using it. QNX, looked promising till the morons who owned it closed it off. Linux's problem is it's too complex for the average user's desktop... Yet look how much development goes into programs for Linux, just because it CAN be used.
AROS gives us the ease of AmigaOS, with the power and cost effectiveness of x86. We don't need it on dead architectures, we need it on x86. We don't need backwards compatibility, we need to move forward. We don't need to run Amiga programs on it, we have UAE and real Amigas for that.
Give me my AROS!
