Ok, so love it or hate it, think it ought to be 5v5 s3 t6 whatever. Chances are it's not changing so I feel the more relevant discussion is given the rules(3 characters per account, 3v3 and Item List). What does everyone see as The Winning Comp, Specs, Gear Mix etc. for the tourney.

Since Priest/Mage/Rogue has a massive lead on all other compositions, I expect that we will see mostly that team and some pure counter-comps (I believe someone made mention of Rogue/Rogue/Feral Druid as a possible one). You may see the occasional Warrior/Druid/X, but I think they will be a rarity, especially if the new patch goes in before the tournament.

I fully expect 2.4.1 (for the first week fixes) to be the patch for the tournament. That said I also expect a decent bit of gear to go into the gear vendor, simply to finish adding things (BS weapons/armor). Regardless of counter-comp teams how effective will a Rogue/R/Feral be versus a war discp/resto/Hpally? If you tailor a specific rock you will find many papers wrap you.

It is not going to be a standard build that people run, if you watched any of the last few Tournaments people often try to guess what the other team will run, then use some sort of wonky counter-comp (like R/R/FD) for an easy win. Though if they guess wrong it can be a disaster.

A part of me hopes the tournament makes more plain how ridiculous counter-comping is as a concept by just making it more public. In a perfect scenario they would actually respond to these and make extensive balancing changes for 3v3 if they intend to make it their new publicity bracket.

Some general thoughts on the gear:
PvE Gear: Rogues and holy paladins pretty much use PvE gear, so they probably take the biggest hit from the t5 part.
Melee Scaling: Melee has scaled incredibly well this season given armor penetration and better weapons, thus the huge rise of double melee teams. With S2/T5, those teams might not have as much success.

Depending on shaman changes (not assuming anything is final at this point), my ballpark guess is paladins and shaman will have the poorest showing at the end of the tournament, while PMR and drain teams are mostly the flavor of the tourney. Really, though, without seeing final 2.4 changes it is hard to speculate accurately.

Counter comping is arena pvp. If you don't like it than this part of the game isn't meant for you. Counter comping also does not mean an automatic win, it simply gives you a favorable match up advantage. It comes with a disadvantage as well, which is the possibility one might not be proficient in the class he picked just to counter comp as well as gambling that your make up would be a counter and losing the gamble.
Rogue mage priest doesn't have any sort of advantage over any popular comp at the current time (Patch 2.3) The group comp is based on heavy burst damage and control with counter spell/fear/blind. Stop their burst, outlast their cd's and you win. RMP is a labeled incorrectly as an overpowered comp because it capitalizes on mediocore players who cannot counter a burst if their lives depended on it. Yes RMP destroy's paladins. However just about any comp destroy's paladins in 3v3 so this is a moot point.

Tournament Test Server is up, and for the moment it seems players can 2v2 and 5v5 there as well. So far I've seen no word on whether they plan to disable those brackets on the actual Tournament server or not.

I'm having mixed feelings. I love the idea of the tournament server, but six weeks is too short for me to go there with the purpose of meeting other dedicated PvPers and eventually forming a team. For any who are interested, I left a feedback post suggesting that they extend the tournament server timeframe to a few months to give lone players a chance to meet other PvPers and form teams, even if the tournament itself is done faster. (WoW Forums -> [Feedback] Tournament Timeframe )

From a software developer's perspective I'd say that Blizzard needs a much better testing approach. In order to tune a class you either need to know all in and outs of the class yourself _or_ you need someone who has this kind of knowledge. Personally I doubt that Blizzard's developer have enough time to play high ranked competitive PvP. Therefore I doubt that their developers really do understand the class issues. But there are a lot of high ranked players out there. Why not invite at least 5 high ranked players of each class to the PTR? Then let them play against each other in various combinations and talk to the players afterwards what they think about the changes. Also it should be possible to monitor/record the matches for an analysis.

I am pretty sure it is very possible to improve class balancing. This, however, requires a lot of testing and a very solid communication between the pro gamers and the developers. The question is, however, does Blizzard really want to invest money into this process? Do they really want to improve/change things? If I look at all those last-minute hacks from Blizzard I feel that they are not as focused or concentrated on this game as they should be ...