Evidently there’s a script available for vBulletin called LatexRender. Maybe we’ve got it already, maybe not. If we have it should handle this
[tex]H_{DR}^n(X)=\Omega^n(X)/d\Omega^{n-1}(X)[/tex]
If not, maybe install it for math/physics threads?
They won’t even fix the longstanding character entity bug.* Why would they install a whole new script?
*(Typing é should be as easy as é. Everything else is an ugly hack.)
Typing é is as easy as option-e, e for me 
é (<alt Gr> and e)
So I have to learn another unportable ugly hack every time I sit down at a new computer? Sorry, no, that’s stupid.
Not to mention that Mangetout’s directions don’t work for Macs. Macs have keystroke sequences to render unicode, and the web supports unicode. The only reason it’s not more widely used is that Windows insists on holding to its own method of producing accented characters.
I’d be all for the convenience, but there would probably be some downsides. How much processor time does it take to generate the images? How much disk space to store them? What bandwidth would it use to transmit them?
Then there’re the questions of mis-use. What happens if a person who doesn’t know TeX sticks the tags around some random (not valid TeX) input? What if a person encloses an entire post in TeX tags?
And then there are the support issues. Will installing this third-party script impact the support we can get from Jelsoft? Will the script follow proper vBulletin rules for tag-matching? If it does do something bad, do we have any recourse for remedy?
Now, I don’t know the answers to those questions, and it may be that all of the answers are good. In that case, go for it: It would, indeed, make a lot of the more technical threads a lot easier.