While this diverges even further from the main topic of this thread, I think it's an interesting question. Beyond that though - you need to value how useful the item will be to you or your guild. To justify where your guild should spend time raiding (farming versus progression, etc) which is only very important as you transition from one raid instance to another (prior to that you have no choice but to farm the bosses to reach your progression bosses) requires knowledge of the drop rate, and some measure of the benefit to the raid.

While it's a bit too fuzzy to ever actually put a number on "raid benefit" and such a number depends on the encounter(s) you are attempting, it'd be fun to try to justify raid time in a more systematic manner. This kind of reasoning is the same one I would use to justify not farming Gruul for [Dragonspine Trophy] if you could instead run a couple of groups through ZA. Time V relative benefit. It doesn't matter if an item is best in game, it only matters how good it is compared to other things you could get with that time investment.

Granted, anyone with enough playtime will attempt to exhaust all possible sources of benefit to their character even at extremely inefficient levels.

Another element of this is, of course, buying a raid ID from someone else. Granted buying an 8/9 BT ID isn't likely, but buying a 5/6 SSC or 4/5 MH ID isn't unheard of. This substantially increases the benefit/time ratio.

Perhaps I'm just remembering a dream, but I could swear there have been times pre-TBC where a raid boss has respawned and been killed twice by the same people during the same reset. This was obviously before GUIDs were visible, but did they drop the same loot both kills? Should something like that happen again, it would potentially afford a nice chance to test.

In TBC, I believe the first Vashj kill also had her respawn immediately upon death, which would have been another chance to test, if the raid had been in any mood to kill her again.

Also if a server ever gets a rollback like it did a couple of times pre-TBC due to item duping, and it happens to rollback a boss you killed, that again would afford a chance to test (although given that the instance would probably soft-reset during the time the server was down, that just introduces another possible place loot could be regenerated).

We've bugged Aran a few times before (~3), when he died to dots just as the poly went off. His door stayed locked and after a soft-reset we had to re-kill him. Each time we got different loot than the original kill.

I dont mean to rain on the parade but honestly do you think blizzard would be stupid enough to encrypt a mobs loot table into its GUID? Something that anyone can sample? I understand that trying to link an item via Atlas or another addon parses the server db and is therefore a more viable means to determining what loot is on what mob but still. To me it seems highly doubtful that anything pertaining to loot has any attachment to client side functionality. I would think that the information would be stored on the server and on the server alone because there is no need for it anywhere else, no one needs to know what that mob has in its belly until the mob dies. The only thing a client needs is information from the server telling them the item ID to show in their loot window. Like someone mentioned in an earlier post the most probable way to find out what Illidan is going to drop is to spawn him and then crack blizzards db and retrieve the loot list. All the GUID would do is point you towards what mobs loot list you want to steal. Its just a way of identifying your raid's Illidan from the other 100's of spawned Illidans.