Let’s say we have a game where one of the primary win conditions is being the last one standing. We have ways to build our capacity to achieve things up, which can snowball as we build things that let us build more things, and we have conflict in which we expend that capacity to tear down each other’s capacity to achieve things. Only the winner wins, there is no second place anyone cares about.
That’s all you need for problems to arise.
If A and B fight, they’ll be weaker afterwards and C can exploit the interval in which A and B are weakened post battle. This means one on one wars are a bad idea. Instead the game focuses on brinksmanship in which the goal is to be so strong that no one attacks for fear of being annihilated by someone else afterward.
If A and B and C are balanced, then A + B beats C. This means that if coordination is possible, the real game is convincing people not to coordinate against you. But there’s no means by which you can do so via skill because in a balanced state all pairings are equally viable.
If A is behind B and C, A’s only way to win is to hope B or C cripple each other in combat. A can’t snowball faster than a larger player without this, by definition, unless there’s a random factor that feels in satisfying anyway. But B and C are incentivized not to do this for the reasons discussed above.
But A, while behind and incapable of winning, can tip the balance between B and C by joining one of them. A won’t win, but A can choose who wins. This is generally called a Kingmaker issue.
There are more issues but I’m typing at work and can’t remember them all, so that’s enough for now.
There are entire games built around the above issues, and they’re well beloved, but usually hobbyist board gamers find them a little shallow after a while. Prominent examples include Munchkin, but there are so many games with these issues that it’s unbelievable. Even truly great games, like Agricola, have this issue. In Agricola you have limited actions. Your actions improve your score and your future capacity to take higher scoring actions. You have the choice of taking actions which are sub par for you, but prevent your opponent from taking their optimal actions. And that alone is enough to cause even a lauded, award winning, classic game for the ages to have these dynamics.
A lot of games try to work around the problem by disallowing certain acts by which the players can most easily pursue these incentives. But that’s a hack solution.
Settlers of Catan partially solves the problem. In Catan, the real action is in the last few rounds of the game, when everyone is on the brink of winning. The best way to achieve ANYTHING in Catan is via cooperation. And importantly, who’s ahead isn’t always easily apparent because “ahead” includes not only your points but also your capacity to get more points. The “game,” properly understood, is to conceal your lead so that weak players will cooperate with you, while declining to cooperate with your true competitors unless you can swindle them into a one sided deal. Even better, convince everyone that your toughest competitor is in the lead, and that they shouldn’t cooperate with him. Then once it’s too late for anyone to take back the help they’ve given you, you reveal your hidden strength and take the win.
So there are ways around these issues- but they involve building the game around them.
Hopefully that all makes sense. I’m writing this on my phone, etc, excuse, excuse.
PS- this site should do an Exursion on Terra Mystica. It’s a hardcore boardgame 4X. Or 2X, kind of. Exploring is replaced with terraforming, and you don’t exterminate. But if you like the expand and exploit aspects of a 4X, it’s exactly that in an excellent board game.
Hell, I’ll write one if you’re interested.

Troy- those feel anticlimactic because you have to sit and watch a counter tick down to a predictable finale, round after round. Worse, there’s often only one player viably in the running by the point they happen. So everyone just groans and throws everything they have at the guy who’s winning, knowing it won’t work because if there was a chance they could stop him he would have delayed and built up his military further, first. The real gameplay in those scenarios is usually
1. Survey your opponents and build military until you’re sure no one can possibly stop you no matter what.
2. Build your wonder or whatever.
3. Win a battle you know you can’t lose.
Catan has mechanisms to fix that. Cooperation is important to progress in a way that it is not in most 4x games. And a lot of info is public. If you’re in the lead and others know it, everyone will work together without you, accelerating their progress and making a tense, neck and neck finish more likely.
Further, there’s very little going backward in Catan. You almost never lose points, and everything you do creates points. This creates a certain inevitability to the final conflict that you don’t have if two players can cripple each other in a fight. Players are thrust into the climactic final race whether they like it or not.
This is tangential, but… A lot of games have mechanisms like this that adjust difficulty on the fly. Like… the best rogue likes usually have a push your luck element that incentivizes the player to create tension even where they don’t have to. This is what makes Dungeon of the Endless so brilliant. The best move is to stay on a floor as long as possible so your production can keep running, while building as little static defense as possible because those resources are left behind when you leave the floor. Even easy rounds become tense because the player will MAKE them tense by stockpiling production instead of building the turrets that would easily ensure victory.
This is what, in my opinion, diplomacy should do. Provide ways for weaker players to catch up, ensuring a climactic finish.

I can see the problem Cadfan, but I’m unsure what could be a solution. All strategy games share these characteristics to some extent. And you yourself acknowledge that so far we have only been capable of coming with half-measures.
And maybe this is because this is just an inevitable aspect to multiplayer strategy and a parallel to real-life diplomatic relationships. It’s just something you cannot avoid towards the endgame. The “game of power” ends up being played like this, because this is the way it has to be played.
The problem however is that MP is too narrow minded in terms of scope. The MP experience usually starts and ends with every game. One thing you pointed out is exactly the lack of any significance to the 2nd, 3rd, 4th, etc places. I would add the also almost complete lack of support for cooperative MP strategy.
Back when we used to play board games, we used to create elaborate house rules for scoring mechanisms that would allow us to play these games in a tournament fashion. The real winner was not found until after 5 or 10 games, with each game simply accumulating points. Certainly there were point bonuses for the relative positions of the players at the end of each game. But you would only win by an accumulation of points given for every action in the game, across several games.
We could extend this to team-based tournaments. And when we had country-wide tournaments, we would even extend it across different games for several weeks in an effort to determine the best board gamers in the country. Large scoring tables where created and affixed to the wall of the gaming clubs at the time.
What I am saying is that an MP strategy computer game does not need to limit or try to circumvent kingmaking and bandwagon. Within reason, of course. A long series of games being played by the same people in a tournament fashion is how you end up actually removing this problem in the most natural way, without affecting the game mechanics or its victory conditions. What strategy MP games need to offer (which they don’t) is well thought out scoring mechanisms with plenty of depth and that allow players to determine their relative abilities.
A player may win a game because of one or two kingmakers. That is fine. Or may win a game because he got the lucky draw on a bandwagon situation. That is fine too. He came out first, score wise and all. But if he scored 12,362 points and the second place scored 11,943 and the third place 9,211, we have here very interesting data. Much more interesting than ending an MP game with 1st place, 2nd place and 3rd place. Data that can be accumulated with further games and eventually eliminate any of the negative side effects of diplomatic power struggle.

I’ve thought a lot about this. And I don’t have any perfect solutions that leave the traditional 4X concept wholly intact.
A society system that periodically forces war would resolve some issues. At least, it would force the breakdown of deterrence based standoffs. Social systems in general can do a lot.
Unbreakable alliances, team victory, and restrictions on who can ally with who would do some work. For example, Imagine a Cold War based game, in which the two strongest nations are permanently at war, with team victory, where alliances can only be proposed by players in third place or below, and only with the alliance that is currently behind. And where that player can break alliances at will, and go neutral. This would reward a player who performs effectively as a client state so that his alliance wins, while declining to provoke the opposing superpower. But this dilutes the world conquest fantasy.
A strong espionage system could help. I mean strong REAL espionage system. Not tech stealing. Build the entire game around nuclear deterrence with enemies who might not even have nuclear weapons. Create things you can do that signal strength, but which can be bluffed. There’s a simple game that I’m thinking of right now, but I can’t remember the name… You have cards that let you do special actions. Each time it’s your turn you play one face down, and say what it is. Maybe you’re lying. Maybe not. If no one challenges you, it doesn’t matter- the card works. If someone does challenge you, you go off alone and show them the card. And maybe they decide to call you out as a liar in public, which costs you greatly. Or maybe you bribe them to tell everyone you were telling the truth all along. Imagine that, but with nukes that may or may not be in North Korea. “I demand new world colonies, or I unleash the mind control bees!” “Shut up, Ghandi!” “You challenge me, Ramses?? Feel the wrath of my bees!” The challenger expends a spying resource and either does, or does not, feel said wrath. And is under no obligation to tell everyone else what did or did not happen. “Oh God, The bees! The bees! I vote we concede to Ghandi’s demands! Now pardon me while I quietly count this briefcase of money I found… in a… bin.” Unfortunately, this probably works best with humans, but I’m not an AI programmer so what so I know.
I’ve also put some thought into a way to use “eras” to fix some of these issues. Imagine that in a real world history 4x, the game strongly rewards you with points for anything you conquer in the 1940s. But in the 1950s it strongly rewards you for signing peace treaties, and massively dials back the points you get for conquest. And in the 80s everyone’s colonies revolt, so you want to earn points by divesting yourself of them to the weakest players, the weaker the better. But you need to be careful, because you might just be strengthening them for when war rolls around. Cycling objectives prevents stasis and disrupts otherwise stable strategies by adding in an element of concern for very different future needs.
Or, if equal wars are the most enjoyable, have a new threat enter and attack everyone in proportion to their strength. This doesn’t wholly solve the issue but it does create a welcome diversion. Or even have the strongest players get hit the hardest, and let them obtain aid in exchange for resource per turn trades with weaker players, creating a catch up mechanism. Note that this also incentivizes differential military investment between large and small empires. You can tailor the new threat or non competitor threat to deal damage to whatever you think most needs kept in line.
One final option that may be anathema to some but fits my euro gaming background- just disallow total war. Make the primary struggle about resource scarcity and reputation. This would take some finesse because you can still have these issues in a non combat game, but it’s doable.