I agree it might be nice to have a system in place for refusing duplicate nodes, but I don't think this is the way to do it. Right now, moderators can submit a node "for consideration", meaning duplicates that need deleting, etc. Very few people make this mistake twice; maybe all that's needed is to improve the message text in a few places to make it clear what's happening when you post a node.

The drawback, of course, is that I can't type up a comment, submit it, go back and type up a new/2nd comment, because the "ID" in the form will already be added to the system due to my previous comment. I'd have to refresh the page that has the form so as to get a new ID.

I do the same kind of check in one of my scripts: I check if the ID has been already submitted, and in that case whether the form content is the same as what was submitted with the same ID. In this case, and in this case only, I consider the post to be a duplicate.

One of the major problems with checking for "same post from same user" is that you have not taken into account anonymous monks. I wrote* a bulletin board (in Perl, naturally) and this is how I determine whether a post is a duplicate: if the previous post has the same title, and is from the same IP, it is marked as a duplicate and the person is encouraged to change the title if, indeed, they want to post it and it is not a duplicate.

This approach may not be ideal here, in a place where new posts can appear very quickly, but it's one way to do it. For this site, I think that using a combination of lastnode_id, IP, and title would be best: if all three match, consider it a duplicate post, no matter how long it has been. Having the lastnode_id addresses Ozymandius' concern, as long as you reply to the person, and not to the top node (which is what you should do). Having the IP addresses the issue of 2 different anonymous monks being able to post replies to the same post, with the same title. As a final perk, you could have a simple "yes, I really want to post this" checkbox along with duplicate post message.

In summary:

$IP = $ENV{'REMOTE_ADDR'};  ## Yes, there are caveats here...
$lastpost = $lastpost{$IP}; ## Stored in a database
## ($lastpost is a hashref)
if (
  $user           eq $lastpost->{"USER"}     and
  $Q::lastnode_id == $lastpost->{"LASTNODE"} and 
  $Q::node        eq $lastpost->{"NODE"}     and 
  !$Q::PostAnywayCheckbox
) {
  &DoubledPost;
}
[download]

*Wrote or "am writing"
Perl scripts: artwork that never
Assents to be done.

That won't work well for a lot of corporate users. If someone uses a load balancer for outgoing posts then every post will come from an essentially random ID in some range.

Instead lookup the last post on the user name. That should work much better for accidents (if not for spams).

True, but most duplicate posts happen within a few seconds of each other, so it should work even for corporate users. More importantly, a lookup by user will not stop Anonymous Monks from double posting - hopefully, once you become a user, you are more aware of what you are doing and tend to make less mistakes. In reality, however, it seems as though registered users make double posts more often than AMs. Go figure.