Sal, I think you're actually arguing two different points here without realizing it.
The first and most important thing you're trying to do is increase viewership by reducing the time it takes to watch a TF2 game. The second thing is that you're trying to speed up 6v6 gamesplay. What you've done is raised an interesting problem, and then followed it with a rather controversial solution.
So, to touch upon the first point: Yes, I think that more people would be interested in watching the game if it didn't take as long. However, I think there are simple ways to prevent this in post-production: as you said before, you can speed up the playback during boring parts of the game; in Starcraft, this is during the "build order" phase, but in TF2, this is when two teams are sitting on opposite points building uber and watching flank, before trying to make a play or grabbing a kill.
I think there's a better solution though: rather than always putting up entire games, why not pick out a high quality round or two, that showcase the better parts of playing TF2? We sort of have an advantage here, because in a game like DotA2, sure you could just fast forward through the farming phase of the game, but it doesn't really last long enough to make up for the other 40 minutes. What we can do is take out rounds that have 5 failed last pushes, and instead watch a round where a team has a successful mid, a successful push into second, and then takes the round with another successful push into last. There's always action, and there's still TF2 you can show off without having to drastically change the game.
Now, for your idea on 3-CP maps; could they be balanced to work with TF2? Sure they could, but I don't think they could be balanced for 6v6 play. There are a lot of inherent issues with it that are solved with the "in-between" points. In this game, there are a lot of stalemates for sure, but this is because a successful uber push typically means you gained ground and captured a point, while failing a push results in either a) losing the previous point or b) having to regroup and hold the previous point. In your example, the previous points in question could be the deciding points of the round, which would mean almost no teams would ever push out of last.
To combat this, you could take several routes; reduce spawn times, increase relative map size (ie. distance between points), increase the number of rounds needed to win. Say we reduce spawn times; now, there's much less incentive to push after one or even two player picks, because they will respawn in time to rejoin the current fight; thus, a player's life becomes much less valuable. Okay, let's also make the point distance a bit bigger so that teams have a chance to cover ground after winning/losing a fight; sure, that helps, but now the game still has the problem of there only being one chance to defend your ground (akin to DotA having one or two less lanes/towers), so a lot of turtling will happen. If you did manage to break the turtling problem, now your rounds go by much quickerl so much so, in fact, that it becomes difficult to justify winning a game in just 5 rounds (which is why CS games have like 11 or something); to solve that, we up the number of rounds it takes to win, and in doing so, bring the total time a match takes back up to relatively the same point.
It's a really hard subject to touch, and I'm sure one that Valve already looked at a lot before settling on 5-CP maps (notice how TC maps have fallen completely out of existence). Instead of changing the game, we need to change outside circumstances; players get up to 15 minutes after the scheduled match time to actually play the game, which leads to a lot of live viewers losing interest. Because of how buggy the game is, there are often several pauses during a match due to disconnects/rendering bugs/computer crashes/etc. The only other thing that reduces interest is simply not understanding why stalemates happen.