Quoting awuffleablehedgie,
you could have it based on flag count. Eg, if you own ALL the flags, damage-reflect is reduced to 0. If you have ZERO flags, you take full damage-reflect. If you're halfway, you take half-damage.
The 10% damage modifier makes spirit-rushing barely viable, but you don't want to punish Oak teams that are having "Citadel zerg rush". Eg, one 'winning' team has most of the map + enemy portals, then the 'losing' team only captured the winning team's portal and they are trying to kill the citadel faster than the other. In this context, the damage reflect is a serious drawback and hindering the "winning" team from zerging the citadel.
Yeah, like I said, this would not be difficult to implement, and I think it has some definite gameplay advantages.
Balancing the amount of damage reduction and/or return would be difficult, however, given the nature of the spirit rush problem. Just balancing against creeps and/or a combined demigod rush isn't too difficult, but because the spirit thing can be done with virtually zero risk, ~50% reduction and ~300 return (which would severely disincentivize a direct premature citadel attack by demigods) isn't enough to stop it from happening entirely. I guess one could still leave Spirit armor permanently in effect at 50%, so that with 50% DR, spirits only do 25%. Given how many spirits a late-game Oak can have, and how much damage they're capable of, a 50% damage penalty hardly punishing a 'winning' all-Oak team with 20-30 spirits, who can easily take down the citadel in seconds regardless of upgrades.
Another compatible solution would be to give citadels some substantial base regeneration - around 100-150hp/s, plus whatever they get from structure upgrades (which isn't very much). This is absolutely nothing against a late-game rush by a winning (or even tied) team who can be capable of doing upwards of 3000-4000dps with creeps from a captured portal, but it would definitely put a dent in the damage a mid-game spirit rush with -50% base damage (nevermind additional, flag-based DR) and damage return.
I still think both WR and flag count need to be taken into account when determining the level of citadel protection, so that a few poorly-timed deaths followed by flag captures by the opposing team can't drop all of your cit protection when they're only at WR5 or so. Additionally, flag consideration should only take into account the number of flags possessed at the beginning of the game - that is, when you have all of your pre-owned flags, your citadel has full protection. Losing flags in the center shouldn't affect that, as that does not indicate 'winning' or 'losing' until they have been held for a long time (at which point war rank comes into play). I'm not exactly sure how to formulate protection level from enemy WR + team flag count, but it's certainly doable with some consideration.
