Testing Economic Balance Changes

[SgtArmyGuy]

As you say, we could introduce a bunch of more complex mechanics to try and address the situation, but, I wonder if just changing the heuristics behind AI voting might be sufficient. Right now, the AI only looks at pop totals when making it's decision about whether or not to support an allied player -- but, arguably, it should be checking other variables as well. For example, should the Phidi really have supported you for Chancellor in that game, even though their fleet was significantly bigger and more powerful than yours?

On the other side of things, maybe there are situations where your opponents should refuse to accept the results of the election, and form a competing alliance. (Arguably, this should have happened in your game as well -- given that the Gremak still had a huge fleet on the board.)

Neither behavior would require big changes to the current rules -- they're really just smaller AI tweaks that would add a bit of nuance and variety to the way council voting plays out. But I think both could help cut down on cases where you feel like the council victory is ending the game prematurely.

That's... really clever. I love it. Currently, there is no DRAWBACK on being declared the emperor - you just win instantly. It becomes a whole different ball game if the vote results in the galaxy being divided into your ALLIES and your ENEMIES. That way you need to make prep work before triggering the vote (forge alliances, make sure you have a fleet ready, etc.) and be damn sure you actually WANT to be the emperor at that stage of the game.

Delaying the vote also becomes a thing - as the vote would now FORCE the rest of the galaxy to unite against you, you want to undermine your enemies efforts to form a combined front beforehand. Divide and conquer etc.

Additionally, this sort of mechanic would make me FEEL like I'm actually forging a galactic empire, forcing my will on those who dare stand against me through fire and steel! The emperor should DO something to assert dominance, after all.

I love this already. It's elegant and simple, and really adds to the narrative!

Just make sure there is a mechanic to make the damn buggers declare unconditional surrender when it's obvious they've lost. The end-game stage should not drag on for as long as wars in general do in SiS (i.e. everybody tends to fight to the last planet). A proper surrender mechanic is something I've been wanting to see in SiS for a long time, but I've already written about that topic in this thread before.

[sven]

I love this already. It's elegant and simple, and really adds to the narrative!

I've been looking at this more carefully, I'm not certain it's actually a great idea. I mean, I've really put two 2 proposals on the table here. The first was to change the AI's voting behaviors so they don't always blindly vote in favor of any ally who has a larger population than they do. In particular, I suggested that AIs probably shouldn't ever vote for someone who's fleet is weaker than theirs. This is a simple change to make, and in most cases, it doesn't really seem to change the way council voting plays out. I've had one playthrough where I *almost* won on a map where winning just didn't make sense at that stage of the game, and this change to AI voting behaviors would have prevented that near-miss. So I think it's a sensible change, but, it's really a small tweak, rather than an earthshaking gameplay change.

The second change I proposed was to add a chance that AI's might, in some circumstances, refuse the outcome of a council vote, and instead unite against the player and their allies. But the question there becomes, when, if ever, should we actually be triggering this behavior?

I took another look at your playthrough (game 8691), and by the time you won, you had actually built up a large enough fleet that you had more combat power on the board than either the Gremak or the Phidi. The only empire that really could have stood up to you was the Orthin -- their super high-tech dreadnaughts probably could have taken out your lower-tech battleship swarms in a straight up fight. The problem is that the Orthin empire, at that stage of the game, was relatively small. Particularly if you'd annexed both your allies after the council vote, you would have had massively more territory than the Orthin, and while their tech advantage might have held you back for a while, I think an eventual victory against them would have been inevitable.

What would have made for a really interesting end game, in that playthrough, would have been for the Orthin and Gremak to unite against you. But, that would have strained the established narratives, as at the time of the council vote, the Orthin and Gremak were actually engaged in a bitter war against each other.

High level, the council vote plays an important role -- it ends the game and saves players the trouble of grinding out a forgone conclusion. And if the games I'm looking at are any guide, in almost all cases, if the player can win a council vote, we already are in a forgone conclusion sortof situation. I can certainly construct artificial scenarios where I think it would make sense for some of the AIs to refuse to accept a vote result, but, I'm not convinced we'd often see such cases in real playthroughs. And that becomes doubly true if I add a little extra logic so allied AI's are a bit less blindly willing to vote in support of any anyone with a sufficiently large population.

So I think the smart choice for now is probably just to tweak the AI voting rules slightly, keep an eye on how votes are playing out in real games, and not rush into coding up the refusal case unless/until we see some real evidence that it's needed.

[Dragar]

For what it's worth, those were my exact concerns when I read the above discussion. I think the eventual conclusion you've reached is the correct one.

[SgtArmyGuy]

So I think the smart choice for now is probably just to tweak the AI voting rules slightly, keep an eye on how votes are playing out in real games, and not rush into coding up the refusal case unless/until we see some real evidence that it's needed.

Well, damn. I was looking forward to that update. Though I must admit, one of my previously voiced-out concerns was the unnecessary drag of the end-game as well.

Lets revise a bit, then. In my game (and in many of the preceding games) the issue was the lack of end-game: the sudden rise of total population through conquest, bringing an abrupt end through Council victory. I never get to duke it out with the remaining AI's, who in the narrative still might seem pretty strong (in our example game, the Orthin and the Gremak). It often doesn't FEEL like they could be subdued through politics alone, or that they should outright surrender, either. Because of this dissonance, I don't FEEL like the Chancellor when I get elected. But it might be ALMOST obvious that I've won at that stage, so we don't want to drag the game for too long.

How about we try to target the preceding stages of the conflict instead?

I've hit the "we're worried about your rapid expansion, so we're making a stand against you" event a couple of times (exactly two times if I remember correctly) - what are the exact trigger conditions of that these days? Because in my mind, it would make sense for the remaining AI's to take action against the player - or anyone who's about to win by council votes, for that matter - to mark the beginning of late-game. At that stage, we already have stable alliances, clear-cut borders, and everything boils down to that one final showdown between the remaining factions. The underdogs might unite into an alliance (I've had this happen in one game, followed by the "we're worried about you rapid expansion" event, and that was WAY cool!) if their history and current relations allows.

If that one event was more common in late game (or, as I suggested, the TRIGGER for late game, even), the narrative would probably make more sense even if the game did end at the exact same time. There would be SOME sort of conclusion through the formation of alliances (that could then be crushed in the end game), vassal states (that could merge into bigger empires) and independent states (that could stay neutral through being intimidating enough on their own, but having too bad/neutral attitude towards everyone else to be partners in alliance). If the political alignments of the board are then known, the only additional victory condition needed in the Council vote would be "assert dominance". This could have rather simple triggers, such as:

1) Have x% of the galaxy's population in your faction (this one's already in place)
2) Show-of-force (destroy n% of the opposing factions population, where n could be very small, and dependant on the relative fleet strength of the factions involved - just take a system or two and that's it)
3) Political power (pay n Influence to grease your opponents into believing you'll make a spectacular Chancellor. This would be a general diplomatic event or treaty available for the player, and it would affect the eventuality of the "we're worried about your expansion" event.)

And like the political power trigger, even the show-of-force trigger could already be hit earlier in the game by winning a war against said faction.

In my example game, I already triggered conditions 1 and 2, actually. The one that SEEMED to be missing was 3, or more precisely, the lack of political narrative. I mean, I didn't UNDERSTAND why I had won. Remember those video messages in MoO2 when the random aliens would just call you to tell you how much they like you and how impressive they find you to be, and we'll bang, ok? Yeah, those. Your allies do that in SiS as well (offer technology and pop), but you don't get similiar info about your enemies. They don't gravel before you after successful campaigns, or send provoking messages to you when they're feeling strong (apart from issuing ultimatums, but that's different from what I'm talking about). If I'm to be elected Chancellor, I would want some sort of indication beforehand that I've somehow managed to subdue the rest of the galaxy, and not just hit some random percentage of pop in a game.

The obvious problem with my "assert dominance" victory triggers is that they could once again be just another redundant mechanism that just drags the game on for no reason, or adds arbitrary tasks for the player to accomplish before the eventual victory. When I think about it, we're not actually missing more complex rules, we're missing more coherent and visible NARRATIVE (and thus, back to the "I want to understand why I won" argument). I probably represent a rather small percentage of the playerbase, but I play SiS for the POLITICS! Even the alliances in SiS are so unstable and so interesting that they actually involve meaningful player choice, unlike in other 4X, in which they are pretty much no-brainers with no real drawbacks. I want to FEEL like the Emperor before being elected as such.

Do you think there are ways to do that which don't actually require the addition of new rules - maybe just new/tweaked diplomatic events or options?

[SgtArmyGuy]

"A near-fire incident! Tensions rise on the border!"
"Gremak raiders on the move - the Gremak government denies any alledged allegiance."
"Orthin ambassador apologizes for the loss of Human lives during the Sirta II campaign on stardate 1145."
"An official peace treaty was signed today between the Orthin and Human empires, following the cease-fire of 1189, now bringing an official end to the confilct."
"Civilian cargo vessel of the Colonials received aid from Imperials, alleviating previous tensions between the two superpowers."

etc.

[sven]

Lets revise a bit, then. In my game (and in many of the preceding games) the issue was the lack of end-game: the sudden rise of total population through conquest, bringing an abrupt end through Council victory. I never get to duke it out with the remaining AI's, who in the narrative still might seem pretty strong (in our example game, the Orthin and the Gremak). It often doesn't FEEL like they could be subdued through politics alone, or that they should outright surrender, either. Because of this dissonance, I don't FEEL like the Chancellor when I get elected. But it might be ALMOST obvious that I've won at that stage, so we don't want to drag the game for too long.

How about we try to target the preceding stages of the conflict instead?

Yeah, exactly. I'm starting to think that the "problem" with the way that game played out was not so much that you won a council vote when you did, but that the two empires that looked positioned to be big end game threats (the Gremak and Orthin), spent the end game fighting against each other, and entirely ignored you when you started conquering their territory.

Ideally, I think the Gremak and Orthin would have noticed your rise to power, remembered that you'd prosecuted an ugly war against one of them in the past, and maybe formed an alliance to resist. If they had, I don't think you would have ended up positioned to win by council vote, at least, not without finding a way to win at least one more fairly tense war.

I've hit the "we're worried about your rapid expansion, so we're making a stand against you" event a couple of times (exactly two times if I remember correctly) - what are the exact trigger conditions of that these days? Because in my mind, it would make sense for the remaining AI's to take action against the player - or anyone who's about to win by council votes, for that matter - to mark the beginning of late-game. At that stage, we already have stable alliances, clear-cut borders, and everything boils down to that one final showdown between the remaining factions. The underdogs might unite into an alliance (I've had this happen in one game, followed by the "we're worried about you rapid expansion" event, and that was WAY cool!) if their history and current relations allows.

I'm wary of making things too deterministic -- reliably triggering underdog alliances or wars against a rising power whenever the player hits a certain vote threshold would make the game world feel more artificial, I think. That caveat aside though, yes, I agree that it's great when this sort of thing ends up happening in the late game, and there are various ways we could make it more likely. One simple step might be to make AI's less likely to fight against each other, or squabble due to trade wars, etc, when the player looks threatening enough.

In my example game, I already triggered conditions 1 and 2, actually. The one that SEEMED to be missing was 3, or more precisely, the lack of political narrative. I mean, I didn't UNDERSTAND why I had won. Remember those video messages in MoO2 when the random aliens would just call you to tell you how much they like you and how impressive they find you to be, and we'll bang, ok? Yeah, those. Your allies do that in SiS as well (offer technology and pop), but you don't get similiar info about your enemies. They don't gravel before you after successful campaigns, or send provoking messages to you when they're feeling strong (apart from issuing ultimatums, but that's different from what I'm talking about). If I'm to be elected Chancellor, I would want some sort of indication beforehand that I've somehow managed to subdue the rest of the galaxy, and not just hit some random percentage of pop in a game.

Council wins can feel rather abrupt. I've thought about asking Arioch to write up some Starcon2-style post-game leader dialog -- so in the case that you win via council vote, you might get messages from other leaders with some sortof parting words that could help tie up the narrative a bit.

More generally, the fact that much of what you hear from other leaders seems to be the same 3 resource request events repeated over and over is not exactly ideal....

There's certainly room to add more texture here, and doing just that is one (of many) items that have been sitting on our TODO lists for a while.

[SgtArmyGuy]

I'm wary of making things too deterministic -- reliably triggering underdog alliances or wars against a rising power whenever the player hits a certain vote threshold would make the game world feel more artificial, I think. That caveat aside though, yes, I agree that it's great when this sort of thing ends up happening in the late game, and there are various ways we could make it more likely. One simple step might be to make AI's less likely to fight against each other, or squabble due to trade wars, etc, when the player looks threatening enough.

The dynamic randomizing element is already in place: foreign relations status (the metric that goes from hostile to neutral to friendly). One of my most epic SiS games was a clash of my alliance against a huge Colonial empire - which ended in a cease-fire, followed by me appeasing the Colonials by giving them back some of their old systems. And eventually allying with them. And starting a war against my old allies. And winning the game in the process!

The dynamic would look something like:

1) Friendly, very friendly: more likely to join you than oppose you
2) Neutral, no bad blood between you historically, not strong enough to challenge you: claim neutrality
3) Neutral, bad blood historically, strong enough to challenge you: claim independence, random chance to go to war alone if necessary
4) Hostile, not strong enough to challenge you alone: random chance to make an alliance with other hostile powers, go to war if capable

And like in my example above, pushing the metric towards friendliness during/after the ensuing conflicts would make it possible to appease the AI. I don't want the AI in general to be out to get me - that would just seem artificial and gamey. I want the BAD GUYS to be out to get me, and cease being the bad guys if the situation so dictates. The worst thing that can happen is that nobody does anything, and I just randomly win.

Council wins can feel rather abrupt. I've thought about asking Arioch to write up some Starcon2-style post-game leader dialog -- so in the case that you win via council vote, you might get messages from other leaders with some sortof parting words that could help tie up the narrative a bit.

This alone would've probably solved most of my issues with that previous game. Sounds great!

[SgtArmyGuy]

Every time the changelog mentions "AI behavior tweaks":



Glad to see new updates again!

[zolobolo]

Yep: the AI has come a long way
The fact that it is now using mercs is a huge step and it is cool looking back at its beginings and now

[sven]

Every time the changelog mentions "AI behavior tweaks":

r38720 includes a change set partially inspired by this thread. I'm hoping I haven't overdone it -- but, as part of my testing of the latest AI change set, I did replay your game 8691 from about mid game, and the endgame felt more substantive. Instead of wasting their time squabbling amongst themselves, the Orthin, Teros, and Gremak did try to form some kind of meaningful opposition block. That's not to say AIs won't fight each other once you're powerful enough -- in my test playthrough, the Gremak saw an opportunity to poach some Teros planets, and they took it. But big late game inter AI wars, like the Orthin/Gremak conflict that effectively won you game 8691, should be much less likely than they once were, especially if you're playing on 'hard' or 'brutal'.

[SgtArmyGuy]

r38720 includes a change set partially inspired by this thread. I'm hoping I haven't overdone it -- but, as part of my testing of the latest AI change set, I did replay your game 8691 from about mid game, and the endgame felt more substantive. Instead of wasting their time squabbling amongst themselves, the Orthin, Teros, and Gremak did try to form some kind of meaningful opposition block. That's not to say AIs won't fight each other once you're powerful enough -- in my test playthrough, the Gremak saw an opportunity to poach some Teros planets, and they took it. But big late game inter AI wars, like the Orthin/Gremak conflict that effectively won you game 8691, should be much less likely than they once were, especially if you're playing on 'hard' or 'brutal'.

Sounds very good! I should boot up a new game one of these days and see how it goes.

[SgtArmyGuy]

Do you still require test data on the latest build? Uploaded game_9707 for review. Humans on Hard.

The external politics part played out really well! I managed to appease the Phidi, annex the Imperials, and after finding out the Colonials had disputes with the Gremak, also made attempts at befriending them. And it worked! After the Colonials were attacked by the Gremak (as I expected based on their relations), the Colonials suggested a tactical alliance. Nice! It was pretty obvious that the Gremak were strong enough to face the rest of the galaxy, but all the major powers combined could resist them, and that's exactly how the AI played things out. A very satisfying and credible scenario!

The Phidi didn't join the war against the Gremak until way after when, however - what was up with that?

Also, the AI still can't do ring galaxies. We still need that algorithm that would only move ships system-by-system to allow better tactical positioning, and be able to recognize frontlines. I think ring galaxies are the best test bench for the AI tweaks on this matter. If you manage to get them to work, everything else should be easy.

Cloning should still be end-game tech. I keep winning too easily with it...

The refit option using metal and labor was awesome, I didn't know this one was implemented! Any particular reason it doesn't que the refits automatically, however?

That's pretty much all the feedback I can come up with. A good game once again got better!

[Serenitis]

I never seem to reach late-game (tech-wise) in any of my playthroughs. Am I winning too fast, or teching too slow?

Maybe some of both? Your playstyle is very aggressive, so I think you tend to either win fast or lose fast.

I'm the other side of this coin, as a fairly passive builder/developer player.
I feel like I'm flying through the tech tree even on 'marathon' settings, because I have a fairly limited amount of territory which has to be super-optimised for research in order to reach the colonisation/terraforming techs in any sensible timeframe.
And this tends to 'drag' everything else along with it, so you discover tech for your ships and before you've implemented that you've discovered something else that's superceded it.

I'd love a start game option to specify the global tech costs independantly of everything else so the player can control how the whole game behaves progression wise.

Hopefully the new terraforming mechanics will also remove some of this issue by making colonisation/development options available much earlier in the game, which should reduce the 'need' to over-optimise in this fashion in order to develop a limited territory.

[nweismuller]

I suspect I'm also pretty far from SgtArmyGuy's playstyle. Been a while since I've uploaded one of my games (partly because I've been waiting for the next major update before I play much) but I've made quite a few reports in my day, and I suspect Sven has seen how my games tend to turn out. I try to be peaceful, but politics in SiS is tricky enough that I've fought many wars I wasn't looking for, heh. That said, the feeling when I manage to finesse the politics to assemble a growing federation of races united under the Phidi Combine or bring every remaining empire to the table in a single alliance is great. And I've definitely seen some games where I've been able to finesse that.

... honestly, I suspect my playstyle might well be 'more markets than is actually optimal', because I'm gleefully trying to increase planetary morale and interstellar trade in my Glorious Seahorse Utopia, but that's my own quirk.