I see your point, but few apps can just inform the user of an error and be shut down every time a service flakes out. It's not like the stand-alone app that relies only on itself and maybe a file server. It's more like apps that use a file server and a remote COM object or two. You've now got several machines involved, a network, maybe the internet, maybe an ISP. The application now needs to maintain as much functionality as possible given that certain services aren't available. What is or isn't possbile now? That question can cascade through the application...

In some cases, queueing will be required to hold processes until the services become available again and messaging is needed to alert administrators that a problem exists. Some functions will simply be unavailable for an indeterminant time. How often do you check to see if they are back?

I'm certainly not against distributed computing. In fact, I'm looking forward to a LOT of it's benefits, especially application deployment. I know however that those apps will be a lot more vulnerable and in some cases, slower than stand-alone .EXEs. I relish having the choice, but I don't see a distributed as all good stuff with no bad stuff.