I suppose what I'm suggesting is that a system based on the actual groups demographic will always work better than any theoretical academic construct or a personal preference/attraction or emotional distaste for a particular method.
Admittedly there are around 12000 groups and counting I haven't looked through but a sample size of 7000 seems large enough to be helpful in identifying the popular categories that actually exist, here, now.
I believe that basing a system on the realities of what is there will work. I think insurance companies use actuarial data that way. What actually happens defines thier rates.
Getting a category for your group is not pigeonholing you. It's a big up that you're an important/popular topic and it creates a way for less popular groups to be found more easily.
If the suggested categories were implemented it would make it easier to manually inspect what remains (other) to see if there are sufficient numbers of groups that share an interest to suggest additional categories that actually exist as popular group types.
Frankly I'd never even heard of incredimail but it spawned a lot of groups so deserves a category. It's not about what I like (mostly) it's about what's there.
Hopefully a good chunk of the unviewed and new 12000 and counting groups would self identify with one of the currently observed categories and not need to be manually scanned. Thus making it easier to look through a smaller uncategorised "Other" section to identity possible additional categories. It seems simple and obvious to me. "Self evident" I think is the American phrase, Bill of rights or constitution, I forget. Wish we had them though.
I'm just trying to solve the problem I encountered in a rational way that works and is as simple as possible. Tags may be a newish trend and I'm sure they have their place but they didn't help me here. No one has told me how they could have.
It doesn't matter how attractive a theory seems if it doesn't actually work in practice.
I understand that a search function and bottom up tagging have a low administrative burden. I also understand that's a genuinely important consideration. But I'm not suggesting anything with a high burden. I've done the slog to observe and identify the initial actual categories plus some I added in out of preference. I'd be happy to identity which witch is which.
That's the one thing that hadn't been done in the previous posts I read suggesting this system. I believe that's enough to make it work without creating a high administrative burden.
If implemented I may do more.