@Faz-
I'm not sure why you think the Cannons are too hard to destroy Faz-. I'm not really curious here either. I truly believe that makes no sense whatsoever.
Considering there is no reward for destroying cannons (2 exp or something, if even that) there is no impetus to destroy them, and it's difficult to. Because spawn die so ridiculously fast to even a single cannon, you are forced to basically tank the cannon until marines, at which point they rip through and don't really need your help anyway. Tanking a cannon makes you hugely more vulnerable to hero kills, and there's no reason for you to kill the cannon in the first place. Doing so just gives your opponent more opportunity to kill your spawn, because they don't get killed by enemy cannons.
This doesn't even take into account that by killing cannons at an outpost, you have no defense of that outpost if you capture it without rebuilding them, which isn't worth the money anyway. Any player with a brain simply takes down the pylon, which dies 3 or 4 times faster than a cannon anyway, captures, and then rebuilds it for 5 minerals.
If you take down cannons at a base, you have access to their Spawning Gates! Woohoo! Too bad killing those is pretty much an auto-loss. Not to mention that your spawns clump up to attack them (not that they do any damage) and basically feed the enemy.
Thof dozens of reasons that the metagame of Temple Siege is counter-intuitive and frankly just stupid.
Solutions:
1.) Respawn cannons when capturing an outpost.
2.) Give a significant reward for destroying cannons. (A civilian, for example, or 20 minerals for everyone on your team.)
3.) Make pylons not 3 times easier to kill than cannons.
4.) Improve spawn so that pushing is actually possible without having to tank the cannon.
5.) If your team kill a spawn gate, automatically give exp equal to killing what would have spawned to your team's heroes.
Either 1 or 2, not both. 3 and 4 I think I said were good ideas in my first TS post and never got a response. 5 Moose already agreed was a good idea three versions ago but somehow still hasn't incorporated it.
None.