Offering some points that may spark some thinking. Long post, so I will offer headers and break things up.
What does TF2chan have that other sites don't? How can we make the most out of that?
One way to look into it is thinking about what the chan has that Tumblr and dA and those other sites don't. If TF2chan can't compete with them in the areas they do best, what things can it boast that the other sites don't have?
For example, Tumblr is a network of private blogs run by individual people, and everything is pretty much in the "now." Here on TF2chan you have centralized boards people can check back on over time. It's a more organized place for discussion, whereas if you're trying to get opinions on Tumblr, you have to depend on spread and then you have to dig through notes upon notes to find actual comments. If you want to send out an update, you can't reach all the exact same people to tell them something new. If you have few followers, good luck trying to get people to see what you put out there.
And say if someone wants blind praise and to be their own celebrity, well, sites like dA would probably be a better fit for them. They wouldn't like it here too much anyway if they can't stand for anything other than "good job"s and faves and sticking only to their closest buds.
I think what TF2chan has going for it is its community basis, and that's what should be cultivated. You can be a nobody (Anon) and still be a part of the community.
Get people's opinions on what TF2chan does well and not so well? Or stop trying to get everyone's opinions and settle with the fact we can't please everyone? Who do we want to please, and how can we try to better please them?
Another way to look into it is investigating what people like(d) and don't (didn't) like about TF2chan.
The thing is...that's already been done quite a bit, so I won't bother discussing it. I know there have been a lot of whiny people complaining about a variety of different things, and trying to cater to everyone is impossible and really isn't always for the better good in the end.
There was something I was taught in a web class involving what are called "personas." They're sort of fictional stereotypical example people designed to typify the target (primary) audience, secondary audience, and non-target (negative) audience (who are specifically not being focused on or designed for). This forces developers to think about who exactly they are designing their website/software/product for and to help them stick to it. It helps reduce the chaos of trying to cater to everyone, and actually leads to a stronger and more focused mission, which usually is more attractive to people in the end anyway. (Think of it as choosing a PBJ sandwich OR a grilled cheese sandwich, not some horrible combination of peanut butter, jelly, cheese, and who knows what else.)
So, thinking in this way, maybe it would help to think what kind of people we want to cater to and which people we're not going to bother to try to accommodate. These people we let go of might include overly-sensitive people who are hard to please. And so, by not getting overly strict and walking on eggshells to avoid offending the most easily-offended, it might be easier for other people to have fun. By taking the middle path, it might not be satisfying to either group of people. Now, I'm not saying the middle path is the wrong path, but sometimes it's the case that certain compromises end up leaving no one satisfied. It's worth taking the time to think if that may be happening here, and weighing working to please the widest variety of types of people vs. trying to make the experience better for a somewhat narrower set of personalities. Not sure how that will translate to numbers, though, honestly.
Closing & Miscellaneous (Only so much we can do, my opinions & experiences)
Lastly, of course, there's only so much that can be done, as Ashe said about the age and progression of the fandom, the chan, and other sites. And, as Cat said, it would probably be best at this point to just make the most of what we have while we can. At least the remaining users can be happy.
Replying to some of Cat's suggestions in the original post, I agree that generally chilling out would probably be a good thing. That would be number 3 and number 5...even if they seem at odds with each other at times. Chilling out is good. People should do it more. Whining on Tumblr is annoying enough. (Hey, I think as much as Tumblr might be a safe haven for those easily-offended types, TF2chan can be a safe haven for those of us who are sick of hearing the whiny SJtastic everything-is-offensive-and-discriminatory junk on Tumblr. That's another thing we got going for us - put that down as another answer to the questions I posed earlier. Focusing on how not to be Tumblr and rocking it.)
I think it's really sad the chan's become a ghost town, even if I personally haven't contributed a lot here for the almost 2 years I've been in the TF2 fandom/lurking TF2chan. For me personally, I was here on the chan within the first few weeks of discovering TF2 (It all happened so fast, I have specific dates, darn it), enjoying all the wonderful
porn fanart and fanfics, and later for cosplay tips and talk. I was really pleased with the quality of contributions on this site. I hadn't ever read such good fanfiction, and it made me want to read fanfics more. Hanging around TF2chan also introduced me to a lot of names of people in the fandom, so it kind of helped me quickly discover a variety of cool people across the fandom. (And flip out when seeing a bunch in one place at a convention.)
And a side effect of TF2chan's tough love rules is that it made me start talking more maturely on the Internet almost instantly. As I told Cat before, I never even considered how overuse of emoticons and such can look really childish, and I'm actually thankful for the chan pointing this out to me. I'm sure this and the chan's attitude about critiques have affected other people too in similar ways, so it's not all a bad thing even if it scares some people off.
Dear god, I wrote a lot. It was an accident, I swear.