A damage spike at the end of the ten minutes is more valuable because your opponent either doesn't have or has less time to react unless they predicted the damage spike. In order to predict a damage spike late in the tiebreaker I think you would probably have to be watching the clock in relation to your opponent's behavior. However, a damage spike of the same size at the start of the tiebreaker gives a lot more time to break even again. If the tiebreaker is divided into rounds in the way that Kichard described, then a single damage spike won't win the tiebreaker on its own, since each successful spike can only earn a max of one tiebreaker point. I think damage spikes tend to cost more resources in the short term then playing normally would. Therefore mages will probably be less likely to spend their ten minutes building up to an anticlimactic finisher that normally wouldn't win the game, and will try to get ahead the way the normally would if the game weren't about to be cut off.
Now that I think about it, there could be exceptions:
Say player 1 deals 3 damage in both the first tiebreaker round and second, while player 2 deals 0 in both. Then on the third round, player 2 deals 7 damage.
And now I think perhaps they should use both the total damage dealt during the tiebreaker and the 2/3 pts system. Technically mode and range represent this better for an individual game, but regardless, different stats are sometimes more reliable to use than others depending on the circumstance. I would say that both should be used, then figure out which tiebreaker measurement has the most approximate reliability based on the damage potential of objects that are in play, and focus on that one.
For example:
Player 1 deals 3 damage in both the first tiebreaker round and second, while player 2 deals 0 in both. Then on the third round, Player 2 deals 7 damage. Player 1 has taken 7 during the tiebreaker and player 2 has taken 6. However, player 2 is walled in with 2 iron golems, and player 1 has destroyed the wall of thorns he was just pushed through and put on a hidden nullify. There is no object on the field that can get player 2 out of that situation (like teleport-Thoughtspore or Huginn or divine intervention).
In this situation, the mode (Kichard's original plan) is more reliable.
The idea is that the arena configuration--the present state of the game at the single moment in time where the tiebreaker ends, provides great insight into who has the advantage/disadvantage, and how stable that advantage/disadvantage is. You can approximate that advantage/disadvantage by looking at the total combined mage-damage potentials of objects in the arena controlled by each player. (This takes position into account, since if the mage isn't in range of an object's attack then that object's mage-damage potential is 0.) for cards like Poison Gas cloud, their mage-damage potential is only applied to a mage that is in range of its damage effect, regardless of who controls it. In the case of poison gas cloud, that range is 0-0.
You can approximate the stability of the advantage/disadvantage by looking at the objects most likely to be able to change the situation and make their own controller's total mage-damage potential greater than their opponent's, in a shorter period of time and using fewer resources. The more different cards and actions that a strategy to reverse the state of the game is, the less likely it is to happen before the player who already has the advantage can pull out a win.
It's just a theory of course, but I think it could have a lot of merit. I'm not sure if the amount of focus it has on current total mage-damage potential and possible ways it can change for future total mage damage potential (that stability I described) underestimates the influence complexities of the social interactions and mental processes of the players can have on the state of the game. But if they were even enough to get that far in the tiebreaker process in the first place that their "mode" and "range" disagreed, then it's possible that their strategic intellects in regards to Mage Wars are relatively close to each other too. How relative that is, however is up for debate.
So that's what I think we should do when the "mode" and "range" disagree, in order to evaluate which one is more likely to predict who would win if the game continued to the end. Even though it can't perfectly predict who would win, technically speaking no tiebreaker can do that in this game because of dice rolls. However, it looks like we can get EXTREMELY close.
So to recap.
When deciding which tiebreaker more accurately evaluates the outcome of a game, ask these two questions:
1. Which mage is in more immediate danger of receiving the most damage?
2. What are the easiest/most likely/least resource-consuming ways the answer to the question above can be reversed?
What do you think?