An observation: I guess it started around the time when customers started upgrading to the 2009 version of Epic. Few customers (that I've been aware of) are implementing any enhancements. They want to, and it's on their priority list, but they are just too busy fixing things to work on enhancements. The 2010 version seems to be no different. I remember the 2007 and 2008 versions--there were few persistent problems. Most of the issues seemed to be enhancement related. "We have a doctor who wants to be able to do X. Is that possible?" At the end of my term of service, my customers had just upgraded to '09 and the issues were of the variety: "None of our doctors can send prescriptions. Was that functionality removed? We'd like it back."
Maybe it's just an artifact of later IUs vs early IUs. The first few years I was at Epic, my customers were on late IUs where the bugs had been worked out, and then upgraded to early IUs that were buggy. If that's the case, it's acceptable to a degree. Every new version of software will have some bugs. Testing will catch a lot of them. Testing can't catch everything; it's impractical to test every single esoteric combination of keystrokes/mouseclicks that a user will enter. But, some of the issues cropping up just look like no one tested the workflows at all. A discussion in the Epic Users One Stop Shop group on LinkedIn mentions constant crashes, in all kinds of workflows. That should have been caught earlier. If there were beta testing, it would prevent a lot of these issues.
To prevent these issues, every workflow needs to be tested. With 200+ customers, all of them with different combinations of applications, in different states with different regulations, using different workflows--there's no way Epic could staff that many QAers, and no way the QAers with no healthcare experience could mimic the actions of a user in the real world. The best solution then is to allow bleeding-edge customers the opportunity to be the first customer on a new application/new version for several months, with the promise of increased technical support in the form of more TS and on-site Epic developers. This will allow for faster end user feedback, faster fixes, and "test" scenarios that "mimic" actual, live environments.
During my tenure, Judy Faulkner constantly brought up KLAS ratings regarding needed functionality. "The needed functionality is there! You just need to get our customers to take the enhancements!" It's great to say that, but if the basic functionality is too broken to use, no one is going to want to install new, untested modules.