Yes Tom, this is pretty much what we want to try next: We were afraid of a complex confidence system at first and wanted to have a ruleset for it that's easy to remember but it turned out there is a number of situations that require special treatment.
For the exact implementation, we will need to decide what kind of behaviour we want to reward with it. Tremulous rewarded primarily individual skill and at the same time punished unskilled players hwo took the risk of leaving base. Our first implementation rewarded base destruction over anything else and while this is also the ultimate goal of every match, it made rounds shorter since there was no reason to struggle when you had the chance to spend your personal resources on destruction.
We would like to encourage teamplay but it's rather hard to find rules for that. Another option would be to reward taking risks. In theory, this can lead to interesting games as the weaker team gets a lot of chances to exploit the enemy's risk-taking while the stronger team doesn't really get to relax. A typical risk would be moving your base to a better location or leaving the main base undefended to build a forward.
I like the idea of giving rewards for killing players and buildables that are proportional to their resource cost. In order to account for the risk factor, there could be synergy effects between structures and players, similiar to what you describe.
Here's a draft that just came to my mind, mod being a fixed modifier to balance the value of structures against the value of players, all constants are examples:
[table="width: 800"]
[tr]
[td]Building a strucutre[/td]
[td]bp_cost[/td]
[/tr]
[tr]
[td]Destroying a structure[/td]
[td]bp_cost * min( 4, 1.2[SUP]close_enemies[/SUP] )
[/td]
[/tr]
[tr]
[td]Killing a player[/td]
[td]normalized_equip/class_cost * mod * min( 4, 1.1[SUP]close_enemy_structures[/SUP] ) * 1.1[SUP]-close_friendly_structures[/SUP][/td]
[/tr]
[/table]
The system would reward not being inside a friendly base. There are still a number of unhandled special cases that would require extra rules though and that's the part where decisions get difficult:
In order to encourage forwarding, it might make sense to give a greater reward for building structures outside of an existing base but then eggspam would yield more confidence than hunting down all those eggs.
Suicide attacks against bases are still efficient, there is no incentive to kill defenders first.