One thing to keep in mind about MLG, is that every match from the beginning to end is a bo3. So, we can think about player vs. player history not as # of matches won, but rather bo3 series won. Instead of arguing how many overall set wins a player has against his opponent, instead approach the problem as how many bo3's the player has won against his opponent. And I'll make the claim that an "Extended Series" can be thought of as 2 bo3's with tie-breaking games to determine the winner. I'll try to show that this method of thinking makes the subtleties of the extended series much clearer.
For now, assume regular double elimination without the extended series rule. In the loser's bracket, if two players meet up again, there's a chance that the winner the first time will lose the second time. That puts the players at a 1-1 tie for bo3 matches won. But this isn't completely fair either - it puts an emphasis on the order in which the bo3's were won (player who wins the second bo3 advances). The fairest way to break this tie is by making the players play more games. Extended Series is a good way of achieving this goal. Not only does it achieve this goal, but it does so in a way that handles all possibilities in a similar fashion while limiting the max number of games played(making it easier to explain the rules/manage/schedule the tournament).
Why Extended Series is a good way of achieving this goal in the loser's bracket:
+ Show Spoiler +
*Any player that wins both bo3's (first to 2 wins) will also win the extended series' bo7 (first to 4 wins)* - Extended Series internalizes the results of 2 bo3's. If a player were to win 2 BO3 series, that player would also have won the Extended Series
If a player wins the first BO3 but loses the second bo3, the players are tied 1-1 in bo3 series won. What extended series does, in essence, is let the players play more games in order to break this tie.
Possible outcomes of a 1-1 tie for bo3 matches won
2-0 first bo3
0-2 second bo3
2-2 overall
Playing Extended Series (first to 4 wins) would in essence break the 1-1 bo3 tie by giving: a third bo3 to determine winner
2-1 first bo3
1-2 second bo3
3-3 overall
Playing extended Series would break the 1-1 BO3 tie by giving : 1 game sudden death
2-0/0-2 first bo3
1-2/2-1 second bo3
3-2/2-3 overall
Playing extended Series would break the 1-1 bo3 tie by giving: 1 additional set win needed for leader, 2 for the player that is behind
Properties of Extended Series:
-The player who has the most overall wins always advances.
-All these seem like acceptable scenarios for breaking a 1-1 bo3 tie to determine who advances.
-The extended series rule covers all of these possibilities by itself (one blanket rule)
-Caps the # of games played (max 5 games in the loser's bracket) to make the tournament easier to schedule.
You can think of an Extended Series not as giving an advantage to one player over another, but *only* as a way to fairly break ties if the players split the 2 bo3's that they play. I therefore cannot say the use of Extended Series in the loser's bracket is "unfair".
But in the finals, if two players meet up again, it seems fair (and logical) to treat their matches as a "bo3-bo3" (first player who wins 2 bo3's). In this case, the player who lost the first bo3 earlier on in the tournament has to win 2 bo3's in the finals, whereas the earlier winner only has to win 1 bo3 in the finals.
But notice that the format above is THE SAME as a straight up double elimination tournament. In a double-elimination tournament, the player coming from the loser's has to win 2 bo3's, while the player from the winner's has to win 1 bo3.
Note that the winner of a bo3-bo3 will always have either more total wins or an equal amount of wins in the series than his opponent.
+ Show Spoiler +
0-2 first bo3
2-1 second bo3
2-1 third bo3
This outcome maximizes the # of wins by the loser of the first bo3 and minimizes the wins by the winner.
4-4 overall W-L, tie goes to the player who has won the most bo3's.
Any other combination of wins and loss, and the winning player of the 3 bo3's will have more overall wins than overall losses.
But what MLG does with extended series finals is that they throw double elimination out the window, and the winner of that one bo7 is the winner of the tournament.
The usefulness of the extended series comes when there is a tie that needs to be broken. In the loser's bracket, such a system is needed. But in the finals, such ties never need to be broken in such a way. There is no reason to implement it. Whoever wins the final under normal double elimination rules will not have lost to his opponent more times than he has beaten him *under any circumstance*. Standard Double Elimination Finals are fair no matter how you look at them - there is no need for the implementation of an extended series during the finals.
Conclusion: The extended series seems to be fair when implemented in the loser's bracket, but is completely unnecessary when implemented in the finals.