The default design (by Matt Grantham, great work by the way) is only supposed to be "good enough", not so good that you never want to change it.
And I hope FOR THE LOVE OF FSM that we've created a default design better than the bottom of the barrel that is phpBB or vBulletin, otherwise, geez.. ouch.
The job here is to judge a thread. If a thread's good, people will post on it, and it will bump to the top. That's it. The only other things you need to know are title and size (number of posts).
Even the avatars are unnecessary here: either the community is small and it'll be all the usual suspects on every thread, or the community is huge and most of the avatars will be meaningless.
phpBB's deep problem is it overloads human conversation with metrics and geegaws that are there only because they were easy to code. They distract from the discourse, and I fear they'll detract from Discourse.
Edit: You can change both so neither are permanent.
Edit 2: Not disagreeing with you. Only mentioned that username and email can both be changed to say there's no reason not to use username for the MD5
To make it quick to add new icons I have a bunch of bookmarklets for finding the favicon for a site, doing a flood fill of the alpha channel from the corners and resize it to 16x16 if it is a newer larger favicon. So adding a icon generally takes less the 30 seconds.
I always seem to miss who's replying. In a sense it's great since we are treated as equals, but I sometimes miss a great reply by pg and others.
Take a look at the Pivory project - that's a great take on design that doesn't stick out, but doesn't hinder the usage either. Adopt something similar as your default theme, kill the badges and you will have a discussion software that is dramatically better than vBulletin, across the board.
This is a sight for sore eyes for someone who has to maintain a phpBB forum and its modifications (all of which, by design, require patching the source code with some bizarre XML-diff format).
Even with no formal plugin architecture, the decision to choose Rails is massively beneficial.
It's also beneficial from a security point of view.
Even if Ruby had a few security exploits recently, there shall never be as many issues as there have been with PHP.
I'm a huge forum users and they're regularly brought down (2+2, one of the biggest forum on the planet, was taken offline for weeks due to a PHP exploit).
This is a PHP bug: http://www.networkworld.com/news/2011/010511-php-floating-po...
Everything else? Bad code.
A larger than average proportion of bad code is a fault of the language, the design choices/compromises made for the language structure.
I think it's both bad code and PHP.. specifically bad code written on old versions of php on crap shared hosts where that's the only server-side language available. It should be harder to write insecure code with PHP but it isn't. A language whose entire purpose is to process requests and return content should have security baked into its core. I shouldn't have to wake up one day and find out that "oh, anyone who adds an ?-s flag to their queries can maybe read your source code or whatnot." Now granted, it wasn't that common, but it should never have existed. PHP may not be the great evil some claim it is but there are parts of it that jump out of the dark and bite you, no doubt.