Feature request: Search for identical or very similar decks

Identical decks

Hi,

Deckbox.org is easily the most awesome site for viewing decklists, so I copy&paste a lot of decks
from the Internet just to view them. Tournament decks, preconstructed decks, any random decklist
that looks interesting - just a lot of decks.

The problem is - I forget to put sources in their descriptions, and right now it's pretty much impossible
to figure out where these decks come from.

I'm sure other people do the same thing, but hopefully at least some describe their decks' sources better.

So if deckbox could find identical decks (or extremely similar, but that might be harder) and link them,
it would be easier to give proper attribution to decklists. I suspect a lot of people copy&paste the same
preconstructed and tournament winning decks, and at least some tag them properly.

Optionally having a central list of precontructed decks would help.

For example right now I'm left wondering if my deck labelled "Ears" is "Duels of the Planeswalkers: Ears of the Elves"
straight out of the box, or if I customized it somehow already.

And for tournament decks I have no idea which was made by whom, and I feel slightly bad about it.

Searching identical decks should be simple to implement and fast (good old hashing).

Part 2

More general search for similar decks would be even more awesome,
but what's proper similarity metric is not really obvious.

% mainboard card same + % total cards same sort of works,
it would probably overrate the difference between
24 Plains white weenie deck vs 16 Plains 4 Marsh Flats 4 Swamp
white weenie deck splashing for Doom Blade instead of Pacifisms.
(12 cards differ, but it's virtually the same deck),
while considering a wide variety of combo elf decks pretty much the same deck.

% same cards ignoring multiples + % same cards mainboard + % same cards total
might or might not work better.

Anyway, I'm just unnecessarily complicating things.
It will probably turn out to be not such a big deal, there's no unique similarity function,
and one can always click Next a few more times.

This might be harder to implement efficiently (you don't want O(n^2) comparisons
and with most similarity functions are pretty hard to get any better), but that would
be some great new functionality.

Part 3

I see how I could do that for my own use in maybe 1 hour of effort by downloading
all decks (http://deckbox.org/decks/mtg), and using a few lines of Ruby, but this
kind of mass download is usually frowned upon (I sometimes do it anyway, I needed
all cards' scans some time ago so I just took them from another website),
and I'd like others to be able to use this feature as well.

If you need some help with minor Ruby/Perl/Python scripting the website,
I have some free time occasionally, and I'd say I'm quite good at it. ;-)
Just email me.