Excellent editorial Steve. I've spent 3 years playing with all sorts of data technologies and have learned some hard lessons.
1. There are some fundamental disciplines that must not be discarded. They transcend a platform. Don't mistake them for just an RDBMS or SQL server thing. For example data modelling is more important in schema less DBs, not less
2. When did Oracle, SQL Server, Postgres, MySQL etc become reliable and solid platforms? My theory is that the complexity of DB platforms means that version 8 or 9 is when they become mature. A lot of the new stuff is in the lower half of the single digit range.
3. Understand what the new platform is for and, more importantly, what it is NOT for. NOSQL solutions tend to focus on a specific capability. If that is what you need then (allowing for#2) they're fine.
4. Where are the sources of expertise? Chances are you are going to need support and training. This instantly kills the 'Open-Source is free' myth. You are also going to be breaking new ground.
5. Organisationally are you good at sharing information? You really need to be if you are adopting a technology with limited pools of external expertise
6. Integration. OMG! Tooling. I hate VIM
7. Security. See #2. It's probably going to be infrastructure security to support pretty rudimentary facilities
Experimentation is fine. Going further must be done with your eyes wide open