Let me clarify a few things here: A long WU in RNA World does definetely NOT mean that you are at risk of not completing it without errors except of course you abort it manually (you shouldn't have done that because if you hadn't you would have seen that they finish and give credits). I hope the time of error reports as we had in the past is over now. At least, the malfunctional WUs are COMPLETELY out of the queue and, again, the RAM requirements are more than moderate. You need to note, however, that with Windows you will loose your current calculation results if you restart the system due to lack of checkpointing with this OS.
As said above, they are not back in the system and the problematic ones have indeed been deleted from the work queue. Still, some WUs are long runners. But that really is a minority. You need to understand that RNA World unlike other projects where e.g. a fixed number of MD simulations are computed over and over again is a project with very heterogenous WUs. And this will remain. But, anyway, I think I gave enough of explanations concerning all this. It ultimately is up to you what you like to support and what not.
Michael.