It's basically game theory's dominant strategy. There are a few different factors into whether something is a DS.
The DS must always be the best choice either by maximizing benefit or minimizing harm. The DS must always be the best choice regardless of the choice the opponent makes - that is, no matter what choice the opponent makes, the DS is in a better position.
Basically, the DS eliminates choice. You're always better off using the DS and the opponent can, at best, only limit how far behind they are while never being able to catch up.
This makes it a two part problem. If the counter to DS is too rare, too limited, too costly, in some fashion, that's bad. For instance, if I have to tech up a very specific tree to defeat something, that's a poor counter (more so if that path doesn't really help me much). However, if the DS is the best choice because the other options hurt me more or help me less, that's bad too.
In context, you can't only look at what they're doing. You have to look at what the other players are doing and what options are available for both players. What are the costs to making the choices. If I did something different, how does that comparatively harm and hurt me. Is it possible that another choice may not give me something as good but also doesn't hurt as much.
So in the end... a strategy can be powerful, even be a massive advantage. However it must come with enough opportunity costs as well as enough alternative options that it is not the -only- possible and advantageous option. If making a melee Sovereign allows me to crush early opposition but a magical Sov can match me and also both of us lose out on something the other has and we have weaknesses that can be utilized readily - and we're also still evenly matched against all the other types of Sovs, then it doesn't matter how powerful the choice is because there are always better (and different) choices in some situations.
In addition, one also must be wary of mixing up the wants of different types of players (something game theory does not cover). For some players, games are puzzles. They figure out systems and see what solves the puzzle (what wins). These types of players are less likely to see something as overpowered compared to other types of players who may approach the game as a competition, who may approach the game as fantasy (fun doing something out of the ordinary), or who may want or have other desires from their game. For instance, a player who likes competition may prefer a little unpredictability and human randomness; they want that sense of jockeying back and forth. However, for the player who sees the game as a puzzle this is -bad-. To them, this would be like the puzzle changing rules and pieces on them. This is mostly a matter of player appeal though, but it is important to understand as different types of players will see things in different fashions. This does not make them -wrong-, however.
