What I would do if I were an admin is add a counter to each dragon which increments each time it is leveled up but resets to zero when the dragon changes ownership and then weigh the counter heavily in the dominance calculation--i.e. reward dragons trained IF and weaken the relative impact of OOF/mercs/prehoarding.
Then, make being counted in dominance more of something to opt-in to rather than just exist as an active user. I.e, the way it is currently communicated is that all users who logged in the past two weeks are counted in the denominator for calculating dominance, but I would change it to only users who have exalted a level 2 or higher dragon during the week in question. Why level 2? Anyone who has observed newly hatched dragons during ID countdowns would see that casual players often immediately exalt their own hatchlings, but those players' motivation for doing so is likely to free up lair space, not to contribute to Dom. If the algorithm tried to only count players who appear to either be actively profit exalting or actively contributing to dom, it could help remove the "dead weight" feeling large flights have. Also, the self-trained counter would mitigate some problems that the opt-in system would otherwise have: even if a flight tries to game it by bottlenecking all of their dragons to a few users who opt-in, self-trained dragons are worth more so it may not help them win to do that.
I also approve of a lot of suggestions I've seen for a dom token trade-in shop which opens the week after dominance is calculated. You exalted a trained dragon this week? You get a token next Sunday! You exalt a lot more of them? You get more tokens! Tell the player how many tokens they'll get at the next dom tally could strike a medium between immediate feedback and making it feel more rewarding (generally good) but still hiding their exact numeric contribution to dom (which could be gamed if fully revealed). With this system or something else, if I were an admin, I'd be worried about how much dom coordination is done off the site on services like Google Sheets which is impossible to moderate and has user privacy concerns and do what I could to help players participate without feeling obligated to use outside apps.
Re: The problem with dominance
Then, make being counted in dominance more of something to opt-in to rather than just exist as an active user. I.e, the way it is currently communicated is that all users who logged in the past two weeks are counted in the denominator for calculating dominance, but I would change it to only users who have exalted a level 2 or higher dragon during the week in question. Why level 2? Anyone who has observed newly hatched dragons during ID countdowns would see that casual players often immediately exalt their own hatchlings, but those players' motivation for doing so is likely to free up lair space, not to contribute to Dom. If the algorithm tried to only count players who appear to either be actively profit exalting or actively contributing to dom, it could help remove the "dead weight" feeling large flights have. Also, the self-trained counter would mitigate some problems that the opt-in system would otherwise have: even if a flight tries to game it by bottlenecking all of their dragons to a few users who opt-in, self-trained dragons are worth more so it may not help them win to do that.
I also approve of a lot of suggestions I've seen for a dom token trade-in shop which opens the week after dominance is calculated. You exalted a trained dragon this week? You get a token next Sunday! You exalt a lot more of them? You get more tokens! Tell the player how many tokens they'll get at the next dom tally could strike a medium between immediate feedback and making it feel more rewarding (generally good) but still hiding their exact numeric contribution to dom (which could be gamed if fully revealed). With this system or something else, if I were an admin, I'd be worried about how much dom coordination is done off the site on services like Google Sheets which is impossible to moderate and has user privacy concerns and do what I could to help players participate without feeling obligated to use outside apps.