Very succinctly put! Nothing wrong with O silently updating bug or security fixes, but updating to a new version of O with new features potentially add new problems.
I don't want to be a guinea pig to test the latest version of O. O22 updated to O23 without my knowledge or permission, and that resulted in taskbar problems(see 'Opera always in foreground' thread).
Certainly it can be argued that this is a way to improve security by forcefully and universally removing software holes and flaws as they are revealed, patched, and updates pushed. But that comes at a great 'instability' cost, particularly to technically-informed users who stay abreast of security issues, run "tight" systems, employ layered security, and strive to keep their systems "stable" above all else. The answer is to give users control over if/when software on a system updates itself.
The reality is that every single piece of software has bugs, including every update. A user sometimes has to thread through a minefield of software incompatibilities and hiccups to get a system fully stabilized and smoothly running. An update invariably introduces different software behavior at some point or another that may or may not be evident as a conflict or problem. When a user performs a manual update, has a known scheduled update, or receives a clear and prominent alert that an update has occurred, he at least has a definitive starting point to troubleshoot any performance hiccups that may be the consequence of the update. With silent, forced updates, the user has no starting point for figuring out why things broke on his system - and the more software that pushes auto-updates, the greater the likelihood both that something will be broken by an update and that it will be unclear what its cause is.