I suspect the no-mod inefficient mesh heads, bodies and clothes are here to stay, whatever anyone thinks about them. These products tap into fairly toxic RL beauty ideals, which isn't a surprise given how lucrative that RL market is. Creators who mine those insecurities in SL will sell stuff, however laggy it is.
I think intentionally trying to mine insecurities is a stretch, and from what I have seen, a significant portion of people with mesh avatars out there are
far out of the realm of any sort of typical beauty ideal. People like mesh heads and bodies because they can provide higher levels of details, and of course the compatibility with higher detail skins/makeup/clothes/etc., whatever their intended look may be.
Personally, I just like something I can relate to, or that feels like me, in my case it happens to fit what is considered attractive. Mesh helps achieve the end result I'm after significantly better than sticking with Linden Lab's system body can ever do. This really goes for any look somebody might want.
The real problem here is the creators of these attachments simply don't care about Second Life. They sell the products for the money with no thought to efficiency or the harm their content is doing. They are amature modelers whose motivation for learning to create anything at all was driven solely by creating content for Second Life. The problem with this approach is they entirely skipped the lowest level basics that are absolutely required in any sort of remotely professional environment. Very, very basic things, I cannot emphasize that enough.
What appears to happen then is that they develop egos because they become popular, and become wilfully ignorant, outright refusing to change their provenly bad habits because of some weird stubbornness, and complete inability to understand why what they're doing harmful in the first place, no matter how well and kindly it is explained to them. The up their own backsides ego wins out nearly every single time. I think here is where insecurities might come into play. They were nobodies in RL, and became "big" in SL, and now think they're too important to listen to anyone.
Unless LL intervenes and limits complexity at the creator upload stage, including the texturing, I think we've got a way to go before we reach peak "beauty" in SL. To think, we once complained about bling, invisiprims, full bright and facelights
It has been talked about with Linden Lab and is still on the agenda, but they are weirdly reluctant to change and don't want to upset anyone. When asked directly about implementing better ways to moderate content that significantly impacts performance, Oz's response was saying they are trying to find a solution that doesn't impact existing content
. That... simply cannot be done, and is the whole crux of the issue. I do not expect anything to come of it.
A partly working solution that attempts to address one of the issues was presented to Firestorm, but it was rejected because they appearently did not understand what texture thrashing was. I later saw the dev who responded to that in-world and they were wearing a 400MB+ avatar themselves and I was not the least bit surprised. Penny went on about the same thing with a Catznips dev on the official forum, which ended equally retarded. As abrasive as Niran can seem in threads like this, he is the only one in the public viewer market that actually truly understands the issue and is trying to do something about it.
I really recommend de-rendering avis over a certain complexity and just let them lag each other instead. I set my viewers at 20k and all I see is their names. I'm hoping it helps my PC last a bit longer if it's not trying to render that content all the time.
I set my viewer to derender by triangle count, textures, and complexity all seperately, and it's made a huge difference. It is rare that complexity will derender somebody before the texture and triangle count kick in, especially since complexity itself doesn't properly penalize the true performance killers. It is still a good idea have some sort of complexity limit though since it is the only thing most people can currently do.