The best way to "fix" tower rushes

Objective: Make a change that reduces the incentive to tower rush, such that it is a situational strategy, rather than one that is viable on most open map games, regardless of civs. Ideally changes should try to keep the game close to as designed.

Ideally, I think the best way to reduce the impact of offensive towers, would be to have the cost of the building be proportional to the distance from the nearest TC or castle, similarly to how trade gold is calculated, with a slight reduction in the base cost, maybe to 20 wood and 100 stone, with the wood cost per tower increasing every x tiles from the TC. This gives the defender an advantage over the attacker and forces the attacker to go all or nothing with the trush. Only issue I could forsee with this change (if possible to implement in engine) is an additional complication for tower defense CS maps. The main disadvantage of this is that it's a major change in the way towers work, some might argue too big of a change.

Other ideas:

Reduce HP of towers in feudal similar to houses and walls ect. - This would make offensive towers easier to remove but they can still easily be walled in. This would also make them less effective defensively, which is how they were designed to be used in the first place.

Give villagers an additional bonus vs towers: A better implication of the reduced hp idea, but still easy to wall with palisades or houses, and towers are still disruptive to eco.

Slow tower build time: Allows more reaction time for defenders to put up counter towers, and slows the rush aspect of a trush, however, makes it harder to rush down defensive towers when defending vs archers.

Increase stone cost: change the cost of a tower by 20, or 25 stone ect, meaning players have to mine twice as much stone before they can drop the second tower. This affects both defenders and attackers equally, so probably not a good fix, might backfire since defender is further behind if they start collecting stone as soon as they see the first tower.