Here is my take:

1. Stability - I'd say very important, especially for a DE that is used by a lot of people as a daily driver.
2. Customization - IMO this isn't a simple yes/no question. It shouldn't be about the number of settings exposed to users but rather the number of "useful states" a user can achieve.

Let me give you an example with the multiline Icons-and-Text Task Manager. Of the main DEs I've tested, KDE is by far the most customizable and provides many related options, resulting in lots of different combinations (states). However, with none of those states I can achieve what I think would be much better (see my recent post here). My proposition is of a completely new alternative layout. I don't think that can be efficiently achieved by simply adding a bunch of "partial" settings (such as, for example, a checkbox for keeping the rows at fixed height), because going that way would result in:

• Many more potential combinations (different possible states). And an increased % of those would be unnatural (not very useful & confusing) to the end user.
• Increased complexity of the code that tries to handle all the possible combinations (hence, more potential for bugs).

So for the multiline task manager, I don't need most of the current settings, so I could say I don't need a lot of customization. But what I'd love to have is something that is currently not available, and that in itself could be interpreted as "a need for (another) customization".

Finding the "right" amount and type of customization is the key, but that's very hard. Largely because making changes gets increasingly harder at the later stages. And also, what is "right" is subjective, of course.