Different games have tried fixing this in various ways. The cleanest fix is to hard-limit the number of units you can have. Most roleplaying games have party size limits, for example. The new XCOM restricts you to 4-6 units in the field, down from one or two dozen in the original. This change drew a lot of criticism from veteran players, but I firmly believe it was a very good decision. With six units in the field, game turns stay quick and individual decisions meaningful.
This approach doesn't work in a game that's about building up, though: growth is part of the basic gameplay, and while you can try to restrict it, you can't just abolish it.
The major approaches are automating away rising complexity or penalizing aggressive growth, but I've seen neither of them really work. A bunch of games, such as Alpha Centauri, let you set an AI "governor" for individual cities, turning over control to the computer and only giving general guidance such as "concentrate on science". The complexity is still there, but now it's being (mis)managed by the computer, which means that you end up with the distinct feeling that the computer is just playing itself. Agency is a core part of most game experiences, and players do not relinquish it willingly.
Penalizing growth means maluses caused by having too many units under your control. Typical effects are reduced productivity (especially science, culture, money), reduced population growth, or increased policing costs. These partly work, but because they do nothing to change the fact that more equals better, the game now revolves around finding ways to avoid or compensate for growth penalties.
I'd like to propose two alternate mechanisms for dealing with rising complexity in strategy games: reducing unit complexity through improvements, or reducing it as a gameplay tradeoff. The approaches above focus on limiting the number of units to keep total complexity manageable, but we can just as well ramp down their complexity to compensate for increasing numbers.
In the improvements approach, throughout the game, new ways of doing things are unlocked that are both simpler and more powerful than the old ones. If the units are troops for example, the following new technologies could be unlocked:
- Weapons that do more damage and don't require ammo, which removes reloading as a mechanic.
- Sensor goggles that let troops see through walls, removing line of sight as a mechanic.
- Bioenhancement tech that gives all soldiers identical perfect stats. Stats no longer need to be displayed.
- Miniaturization turns all single-use items like grenades into infinite-use items.
- An auto-medkit that automatically cures status effects like poison and panic, removing them as mechanics.
- A powerful energy shield that can absorb any attack that doesn't outright kill the unit. Units no longer have hit points, they're just alive or dead.
You can see that with these kinds of changes, a game that lets you comfortably micromanage four units at the start can support dozens at the end with no increase in complexity. In a civ-like strategy game, advances would replace city buildings with civ-wide bonuses, remove unit mechanics and restrictions, and remove tradeoffs in production and tile development.
This is quite nice, but it has two massive problems: First, instead of designing for one set of mechanics, you are forced to consider a whole series of variations where some mechanics are there or not. Getting difficulty level or level design right is a lot harder if you have to make it work for all of them. Second, if the unit growth curve fails to match the simplification curve, you can end up with situations where your game is still too complex, or has become too simple. If the player has a strategy where they build dozens of cities right at the start while ignoring research, they run into the same complexity trap as before, and if they do the opposite, they may find that the game has turned into a tiny "lite" version of itself.
My most recent and favorite idea is embracing reduction in unit complexity as a gameplay tradeoff and working it into the theme of the game itself: When you reach a certain number of units, you have to give up some of the detail in exchange. This is applied either to all units, or just to the ones beyond the limit.
Let's say you are leading a rag-tag band of mercenaries. You start out with three of them, and you can see detailed info on their abilities, attitudes, and equipment. When you get your eighth mercenary, you have to put her under the command of one of the previous seven. You still see her name and specialty, and you can select what equipment spec you want for her (pistols, stealth...), but her exact equipment and emotional state are handled by her boss.
Of course, each of your lieutenants can directly manage seven people. You can still assign more mercs to them, but these additional ones no longer have any information beyond their basic job. You can reassign mercs, making them more or less prominent in this org chart, but reassignments damage morale, you can't promote the faceless mercs beyond the first 49, and you may not instantly get full access to your newly-promoted unit.
The necessity of reducing unit complexity gets worked into a delegation mechanic that can drive the gameplay. Which squad do you send on this mission? It can also drive a theme of the loneliness of command and the changes that happen when you stop being an underdog. In short, embrace the dehumanization of simplified unit stats as a theme!