I think the vast majority of users prefer a stable application as opposed to the latest and greatest(highly debatable) code...

This is my understanding and these are my expectations as well. I would much prefer a stable system with less variable economic incentives rather than a system constantly in flux with a vision that seemingly changes on a Ned whim. It's hard to ask for investment when an investor's money can be significantly affected by multiple variables...and multiple changing variables from fork to fork.

I'm not saying that we shouldn't seek additional/better ways to increase SP utility - we absolutely need to do this, but we should do it in a manner that makes potential changes in returns more easily predicted/calculated and in a manner that ensures a higher degree of behavioral stability among users.

Nobody likes unpredictability. And they certainly don't like it when it comes to their investments. Since Steem is primarily an economic and stake-based system, we should be doing as much as we can to treat it as an economic/investment platform. Minimize risk as much as possible and maximize returns for those assuming the risks...and keep these risks and returns as stable as possible with each protocol change. If major changes are needed, then we better be damn sure that these changes actually have widespread stakeholder consensus and that they are implemented as smoothly as humanly possible.

Flying by the seat of our pants and testing buggy code on the production chain is horrifically irresponsible for all parties involved.