Contrast this with Engadget, who can enforce discussion that aligns with whatever site policies they choose to adapt. Or you can take this even further, like on my blog, where openness means nothing and I can just delete anything that's not a question or an anecdote that adds to what I've written.
In an open discussion forum, your biggest problem will be always be open discussion.
"nice article I wrote one like it at http://buyviagra-cheap.example.org
and "HOW TO INSTALL THIS???!!!!!?!!! IT DOES NOT WORK!??"
Invite readers in, tempt them with an interesting article and then leave them in some internet shady backstreet area. It doesn't need free speech, delete the tripe, edit the rest mercilessly (but note that they have been 'edited').
> It doesn't need free speech, delete the tripe,
> edit the rest mercilessly (but note that they
> have been 'edited').
Of course, I always end up missing some, so I tend to go through every few weeks and prune those that I missed. What we really need is gmail-esque spam filtering in disqus.
Comments close after a period of time, so it shouldn't scale temporally, comments for a large site with lots of authors, well, they have lots of authors.
If you want commentors to be able to suggest corrections indefinitely, then either a wiki is a more suitable medium, or an "email me" link. Depending on what you want to achieve - a communally edited document or a personal reply.
As far as newspaper sites go, they go beyond 'comments'; hundreds of comments is in the realm of a forum and deserves some kind of moderators and so on.
You want to publish to the world and have the world able to reply to you while you aren't paying attention? That's not going to work. If you want to publish to the world and have the world reply to each other, well that's what people have their own blogs for, surely; why would you want to host that indefinitely?
And if you are going to have an abandoned corner of the internet where nobody 'in authority' checks what happens, then you need to rely on the passerby, hopefully give them some tools to step in and help, e.g. several downvotes over several days from different sources = autodelete. I doubt that's a solution; if you could automatically tell who to trust in the first place you wouldn't let the spammy comments be submitted.
4chan also implemented Robot 9000 on one of their boards: http://boards.4chan.org/r9k/
Bit tangential, but has anyone (e.g., start-up) solved the problem of massive discussions? Anything over a hundred or two hundred comments becomes unmanageable in a hurry. Some of the larger sites (e.g., Huff Post) routinely have 2000-comment threads that would take days to plow through.
Is there a solution that combines karma, upvotes as a means of identifying the 'cream' as well as grouping of sub-discussions and points-counterpoints? I haven't seen any solution take a serious shot at a way of grouping sub-arguments and discussions.
The level of comments on the typical news article here in Australia is embarrassingly bad. Any discussion quickly becomes a confusing read without threading and a complete waste of time without persistent identities or moderation. Disqus has taken a crack at the basics, but is it much better for discussions over, say, 500 posts?
Here are her (very cogent) thoughts on how to moderate massive discussions:
Two useful links on the subject are http://nielsenhayden.com/makinglight/archives/006036.html and http://www.meatballwiki.org/wiki/
pg is already more than capricious enough already, I shudder to think of how ridiculous things would get around here if he started taking lessons from the dark master.
A choice summary quote: "i would never never ever hire tnh as a mod ever -- an inquisitor, maybe, a chief of secret police, certainly, but a mod, no!"
They take great pains to avoid banning people (excepting spammers, who get their signup fee refunded!), though sometimes they dole out timeouts. The community policing means that the shitty users tend to disable their own accounts.
pg takes the opposite approach, applying a single assholish technical approach to all problems: hellbanning for everyone!
But I agree that Metafilter handles moderation better than HN. The way voting is here, it seems like it mostly depends on the time of day as to whether a comment of mine gets -4 or +30. (Usually this works out for me pretty well... but I can't help feeling dirty when I write something that gets upmodded that high.)
But if you want a like-minded small community it's a tradeoff that's necessary.
Why do you feel dirty though?
"Benign dictatorship" works. Anything short of it is asking to be rules-lawyered to death by trolls and griefers.
A site with high-quality comments can increase the value of the existing content exponentially, but because there is a more removed connection to revenue generation it seems like the majority of comment sections (HN excluded, of course) are thrown in as an afterthought.