Let's Play Dominus Galaxia

[RefSteel]

So, in case this is useful and not just annoying, more thoughts on randomness:

Yes, the progression to less randomness in large encounters in MoO 1 is absolutely and unequivocally a good thing from a design perspective. No, it does not reduce strategic depth. It would be cool if DG matched or exceeded MoO's strategic depth with more-random combat, and this could well be possible, but added randomness does not intrinsically add strategic depth: It actually reduces the importance of any strategic plan because as variance increases, the odds of success don't get as close to 0% with a lousy plan, and don't get as close to 100% with a great one, so anything you do only changes your odds within a narrower range of probabilities. Go (the board game) has incredible strategic depth. Chess has a bunch too. Neither features any random factors. War (the card game) and Craps (the dice game) have zero strategic depth. This is because they feature nothing but random factors. It is possible that there's some sweet spot between "no randomness" and "all randomness" that would enable even greater strategic depth than is possible with pure determinism, but do you need more strategic depth than Go in a computer game?

The best reason to include random factors in a strategy game like this - in the map, in AI decision-making even beyond what's necessary for good Game Theory play, in the tech tree, etc. - is for replayability: So that every game you play will be different and new in memorable ways. Another reason is to render the details of certain decisions less important intentionally, to allow more varied approaches to a given situation to be more nearly equal in terms of likelihood-of-winning. This is more difficult to balance while still maintaining strategic depth, but it can certainly have value too. And of course for combat in particular, there's the emotional charge and tension that comes with uncertainty: Weighting the dice as best you can with good strategy but then holding your breath as you watch them fall because they still can come up snake-eyes and ruin everything.

So ... that was way too long, and now I have more I want to say, which I think I wouldn't include if I hadn't actually written it first. So I'll put it in spoilers, especially as it isn't directly about DG anyway. It's just that a couple of the things you mentioned about MoO's balance ... aren't actually true.

1) RB vets certainly can and do get into situations where our fleets can't hurt enemy ships. We play crazy variants on Impossible difficulty on random maps! Sometimes you lose the game. Sometimes this happens because runaway Psilons have teched up to Shield VI and your missile bases have nothing better than Scatter Packs and there were no good early beams in your tree.

And sometimes that happens, and you find a way out of the problem anyway somehow, through diplomacy or espionage or invading another neighbor who has weaker shields but better weapons, or if you have a big enough production base, crash-researching Weapons tech on the spot. (Note also, Shield 4 does not hard-counter everything you can build even if you do no research! Heavy Lasers can penetrate them naturally, and though semi-worthless against Shield 6, they can still at least "scratch the paint" against them - and that's tens of thousands of RP into the Forcefields tree!) There are many strategic options; there is strategic depth. Pushing Weapons tech for the hope of the right tech showing up in the next tier and completing in time, or raiding an ill-protected enemy colony with ground troops, or pushing all the production you can into espionage for a desperately-needed weapon, are all "Hail Mary" plays that are likely to fail but have a chance to steal victory from the jaws of defeat - but they're plays with real strategic depth because you have to make real strategic choices based on your rivals' situation and your own, and follow up on them effectively.

All that said, I don't actually know what most people are going to hate or enjoy about the game. If lots of players can be expected to build a billion laser ships, fail to use them until they're completely obsolete, and hate the game because instead of usually doing no damage to much-more advanced enemy fleets, their lasers never did any damage, I can kind of understand that, and I can certainly understand taking it into account as a game developer. If I were making a game, I would want everyone to love it, whether they've never played 4X before or whether they're curmudgeonly old MoO fans like me.

2) Battle computers are not a "no-brainer" on well-designed ships in MoO. It depends on the ship type and situation. Though they're very (very) often useful - if they weren't, why would you ever commit the resources to research them on purpose? - it is also often worthwhile to use an out-of-date battle computer or none at all in order to squeeze a little more raw firepower onto a ship, or even just to reduce a design's cost - and thereby get more of them into space - depending on the game situation. "Depending on the game situation" is the watchword of MoO, and that's the reason folks like us can enjoy it more than two decades after its release: Every game is a special case, calling for subtly or extremely different tactics and strategy.

[Disclaimer: Of course that only applies when we agree not to exploit certain bugs in the game and flaws in the AI design. For instance, if you're playing the base game without kyrub's unofficial patch, and ignoring our exploit rules, a Huge ship with auto-repair, defensive loadout, and a stream projector changes the wording on the "WAIT" button to "WIN." That's not a problem with the design though; it's a problem with the buggy execution of the programming, just as the feeble AI is an artifact of early-nineties computers' capabilities.]

[Reformations]

No clue if you guys are still taking ideas, but I've thought MoO was missing early ground combat against non-racial enemies.

Here's a possible way to introduce ground combat. May make sense as a mod.

-----Non-Sentient Wildlife------
Many planets will have local wildlife which will need to be cleared by transported colonists to make room for expanding the planet. NSW do not prevent anyone from colonizing the planet but 40% to 80% of the planet's population capacity would be taken up by nasty bugs that need killin.
Example: Arid planet, 50 pop max. Due to 35 NSW, effective max is 15. Planet grows naturally up to 10 by itself. Then a transport containing 15 colonists arrive. 5 of the transports go immediately towards capping out the planet at 15 population. The remaining 10 are sent out into the wild to kill the bugs.
This is identical to MoO1 ground combat--it just happens to be in the early stages of the game and not against any race in particular (think space monster but on the ground).
The 10 transports take out 15 of the 35 bugs. This means the new planetary population max is 30. The planet may continue to grow naturally, or more transports can be sent. Any transports arriving OVER the current max will be sent into ground combat against the bugs. More bugs killed, higher population cap.
Bugs have 3 types: +10 combat, +20 combat, +30 combat. You must kill all the +10's before fighting +20's etc.
Different races could respond to NSW in different ways. The mechanic is a buff to bulrathi as it is the best way for them to out-expand other races without modifying population growth, propulsion, or planetology tech.
Sillocoid could slowly kill off the NSW through pollution. Sending troops would be faster but perhaps not necessary as the NSW would die off naturally on silicoid planets.

[Jeff Graw]

It seems like a really cool game!  It doesn't look like a MoO clone, but a) MoO exists, and b) Remnants of the Precursors is on the way to update the same basic gameplay.  We don't need another clone.

A funny thing is the more people are familiar with MoO 1, the less they see it as a clone. But at a distance we're still very similar to MoO 1.

It sounds like what you would want to aim for is a case where tactical combat remains important throughout the game, strongly affected but not completely determined by strategic decisions.  A game like that could be really cool!  (See e.g. X-COM - the original or either new version.)  Tactical situations would have to be highly varied - exactly the sort of thing you're talking about with more complex flanking rules, terrain features, range effects, and other specials.  I'm skeptical about the tactical game holding up across a full game when it's normal to have around a hundred stars (like Alex in his video) but it can definitely be done, and it sounds like you guys have the team to do that if it's a priority!

Yeah, that's the goal, but it's an audacious one. And it's entirely possible we won't be able to do it justice, and decide to settle on something simpler.

On randomness favoring the AI:  Assuming equally good combat AI is available to both sides, if there is any situation in which a player can gain an advantage by controlling a tactical battle manually instead of leaving it to auto-play, and it is possible for the player to correctly identify such a situation, then the player is better at tactical combat than the AI by definition.  (If this is not the case, then that player's best move is to auto-resolve every time.)

I both agree and disagree with that. There's a ring of truth, but we're adding randomness to a very specific subset of tactical combat that a well programmed AI has the advantage in. Say for example we took MoO 1 combat, taught the AI how to play it more or less perfectly (it's debatable, but MoO 1 combat is simple enough that I don't think it would be difficult for us to do it), and then took out the entire random element. What you end up with is a math based problem, and I have yet to meet a human being that can beat a computer when it comes to math. And I'm not just talking about about when you get into the tactical layer. The AI would be able to calculate the results of theoretical confrontations before they happen to never let itself get into a battle it will do poorly in, similar to what you would see in Minority Report (assuming there's no way for fleets to sneak up). In this scenario, randomness benefits the human player humongously (regardless of whether or not he has the ability to auto-resolve.)

And I'm aware that the scenario I described might actually sound really good to you

So, in case this is useful and not just annoying, more thoughts on randomness:

Oh no, I'll talk game design until I get blue in the face! It's just that very specific feedback on our incomplete combat system that isn't based on actual play testing has limited usefulness

Yes, the progression to less randomness in large encounters in MoO 1 is absolutely and unequivocally a good thing from a design perspective.  No, it does not reduce strategic depth.  It would be cool if DG matched or exceeded MoO's strategic depth with more-random combat, and this could well be possible, but added randomness does not intrinsically add strategic depth:  It actually reduces the importance of any strategic plan because as variance increases, the odds of success don't get as close to 0% with a lousy plan, and don't get as close to 100% with a great one, so anything you do only changes your odds within a narrower range of probabilities.  Go (the board game) has incredible strategic depth.  Chess has a bunch too.  Neither features any random factors.  War (the card game) and Craps (the dice game) have zero strategic depth.  This is because they feature nothing but random factors.  It is possible that there's some sweet spot between "no randomness" and "all randomness" that would enable even greater strategic depth than is possible with pure determinism, but do you need more strategic depth than Go in a computer game?

This is hard to respond to because the first part of your response is a bit of a non-sequitur. You're making an argument for less randomness, but the question isn't about the ideal amount of randomness, it's about the inclusion of a gradient as opposed to something more static.

The second half is also difficult to respond to, but for different reasons. Lots of different assertions and questions that can be approached in different ways. Yes, more strategic depth is always a good thing (although there may be outside trade offs to be considered. There's a reason Go isn't a 4X game!). No, it's unlikely that any 4X will be able to approach Chess or Go. Are games like Chess fundamentally strategic since they are solvable problems? Strategic thought surely comes into play, but that's based on our inability to come anywhere close to solving the game. Since I'm more familiar with Chess than I am with Go (but by no means am I a gifted Chess player), I'll use that as a paragon of deterministic strategy, and look at how randomness might improve the game:

As I can tell, the amount of strategic thought that goes through a Chess player's head resembles something like a roller coaster as his proficiency with the game increases:

1. At the beginning, strategic thought is modest. He is largely looking for pieces he can take, and trying not to miss any of his pieces that are in immediate danger. Thinking ahead more of less amounts to: "If I capture this piece, will that leave me vulnerable to a capture myself? Is that a good trade off?"
   
2. As the player progresses, so does the amount of strategic thought he puts into the game. He becomes better at immediately recognising vulnerable pieces on both sides. More and more thought is directed towards figuring out longer term consequences of actions.
   
3. But then he starts delving into the realm of openings, historical scenarios, tried and tested strategies. There's an emphasis on memorisation that is not at all strategic in nature.
   
4. The last stage, where two masters are playing with intimate knowledge of historical precedent, causes thought to become more strategic once again. Those memorised positions become a tool that is more ingrained (similar to the beginner gaining the ability to immediately see at risk pieces instead of spending a huge amount of time searching for them) rather than the main instrument.
   

(It's possible I'm off on a few minor details, since I'm admittedly not a good Chess player, but I think the overall progression makes sense)

Given that, a very easy way to increase the practical strategic depth of Chess (compared to the fundamental strategic depth, which is non-existent for any solvable problem) is to simply randomise the starting order of pieces, so that you can have pawns in the back row, bishops in the front, etc. I think this is actually a somewhat common modification of the game, actually.

But aside from all of that, I'm not sure if there's much point in comparing games like Chess and Go to a 4X strategy game. Chess and Go can succeed the way they do because they are very abstract. 4X games, even the more abstract ones like MoO, are still victim to numerous limitations from being modelled after concrete situation like running a galactic space empire. While Chess and Go, without any randomness, can appear to be strategic through being very difficult to solve, I'm very skeptical that the more concrete 4X genre can accomplish the same thing. (Design for a purposefully abstract as possible 4X game would be an interesting segue). Worse yet, typically in a 4X game you only need to solve a small subset of the problems in the game to win it. Take GalCiv's deterministic and readily solvable diplomacy system as an example: You can play non-optimally in many ways, and still exploit that one aspect for a nearly guaranteed victory.

1) RB vets certainly can and do get into situations where our fleets can't hurt enemy ships.  We play crazy variants on Impossible difficulty on random maps!

To be fair I was talking about DG, not MoO 1, but I suppose that's my fault for not being very specific. And even then this can actually happen too easily in DG as well (or could, if it weren't for crits) since having an AI that plays the game properly lets it snowball so much harder when it gets an exceptionally good starting condition. But the goal for DG is to do a better job creating balanced starts, and balancing the races in way where they're still unique but none are obviously OP. So in that sense, even without crits, I doubt an RB regular will get into that compromising position against an AI without production bonuses when we turn our attention back to the galaxy generator and balance starts better.

We're also probably going to have a better mechanisms in place to make sure no one is totally screwed over by random tech picks, while still allowing a good amount of variance. It can be tough to find the right balance.

2) Battle computers are not a "no-brainer" on well-designed ships in MoO.  It depends on the ship type and situation.  Though they're very (very) often useful - if they weren't, why would you ever commit the resources to research them on purpose? - it is also often worthwhile to use an out-of-date battle computer or none at all in order to squeeze a little more raw firepower onto a ship, or even just to reduce a design's cost - and thereby get more of them into space - depending on the game situation.  "Depending on the game situation" is the watchword of MoO, and that's the reason folks like us can enjoy it more than two decades after its release:  Every game is a special case, calling for subtly or extremely different tactics and strategy.

It might be that I just have a slightly more hyperbolic definition of "no-brainer" in mind. The DPS increase from a one point increase to attack can be really significant, and isn't likely to be under ~15%. But yes, sometimes you might use your second best battle computer, and some exceptions don't necessarily even need much attack. But the general point is that attack/def is very, very significant in MoO.

No clue if you guys are still taking ideas, but I've thought MoO was missing early ground combat against non-racial enemies.

I actually like this idea a lot! I don't know if we'll include it (a lot would be down to dev time and good presentation), but it's a fabulous idea that I'll keep in mind

[RefSteel]

Oh, absolutely: DG's connection to MoO is obvious and close. It isn't meant to be a "faithful" clone of the original game though (like RotP) right? My sense is that it's meant to go further, using Orion as the base from which it's working.

The scenario you describe in which randomness favors the AI humongously is the one in which the AI is so much better than the human player at the entire game that the chance of the human winning is humongously worse than 1/N, where N is the number of players (AI and human combined) in the game. The only way for the AI to ensure that it never fights a battle it can't win is to win the strategic game so thoroughly that it can simultaneously be prepared to repel any attack the human can muster at any time and be developing its economy fast enough to retain this advantage. If the AI has that kind of advantage over me, any strategic decision I make will at best change my odds of winning from ~0% to ~15% in a galaxy with five AI enemies if luck plays an enormous role in the game. For me at least, this doesn't qualify as favoring my empire at all, because if my chances of winning the game only exist because I might get some lucky die rolls, why am I playing a strategy game?

Your description of Chess and the way players randomize it is perfect, by the way, except that skill at Chess tactics and memorization of the openings are independent variables, except insofar as studying them helps a player to understand the game's mechanics better. A low-skill player can use (or blindly memorize) a book of openings and still lose every time because of failing at the mid-game. But to the point about randomness: All those known openings exist because the game is played on the same "map" every time, with the same pieces in the same positions. The player who has memorized them is gaining the benefit of millions of playthroughs from those same starting positions. The solution, swapping pieces around so the defacto layout of the map is different, is analogous to rolling a different random map every time you play MoO or DG. Randomizing starts (including tech tree and opposition) is a great way to increase replayability and reduce or remove the relevance of rote openings. But increasing the randomness of combat would be like playing a version of chess where moving your piece onto a square occupied by your opponent's forces a die roll: On a 1-5 you take the piece as usual, but on a 6, that piece takes yours instead! This would not add to the strategic depth of the game.

Why ever use random variables like to-hit chances in combat then? There are a few reasons, which I think we've both touched on at different times, but the main thing is, 4X isn't Chess. There is more to the game - as narrative, as simulation of space-opera-style interstellar conflict, which must have random factors, etc. - than a highly-complex strategic logic puzzle. And look, if the balance of these factors makes DG more fun as a game when combat is more random, great! Bring on the dice!

(Just one other thing though: I'm not sure how clear this was, but finding myself with a gaping hole in my tech tree that makes it an extreme challenge to take on enemy fleets is terrific fun in MoO! Given all the methods of acquiring the tech in spite of these holes I described, the possibility just adds challenge and yes, strategic depth, to the game.)

[Jeff Graw]

The only way for the AI to ensure that it never fights a battle it can't win is to win the strategic game so thoroughly that it can simultaneously be prepared to repel any attack the human can muster at any time and be developing its economy fast enough to retain this advantage.

Not quite, an AI that can "precog" whether or not it will win any theoretical tactical confrontation hasn't necessarily beaten you strategically, and won't necessarily win because of that ability. All it means is that it knows which battles to avoid, and which ones to fight. You as the player might still have an overwhelming advantage in terms of forces, production base, and technology, that makes it all but impossible for the AI to prevail. His precognisance of the outcome of any potential battle won't help him except maybe to put up a good fight before he kicks the bucket.

Your description of Chess and the way players randomize it is perfect, by the way, except that skill at Chess tactics and memorization of the openings are independent variables, except insofar as studying them helps a player to understand the game's mechanics better. A low-skill player can use (or blindly memorize) a book of openings and still lose every time because of failing at the mid-game. But to the point about randomness: All those known openings exist because the game is played on the same "map" every time, with the same pieces in the same positions. The player who has memorized them is gaining the benefit of millions of playthroughs from those same starting positions. The solution, swapping pieces around so the defacto layout of the map is different, is analogous to rolling a different random map every time you play MoO or DG. Randomizing starts (including tech tree and opposition) is a great way to increase replayability and reduce or remove the relevance of rote openings. But increasing the randomness of combat would be like playing a version of chess where moving your piece onto a square occupied by your opponent's forces a die roll: On a 1-5 you take the piece as usual, but on a 6, that piece takes yours instead! This would not add to the strategic depth of the game.

That's all true! Randomness simply wouldn't make sense in terms of piece taking because it upsets the very binary nature of the game, and would work against the established rules instead of working alongside or complementing them. But in areas that lead to rote response, randomness helps. Opening composition, starting locations, or even board geometry could all be good uses of randomisation applied to a game like Chess.

Why ever use random variables like to-hit chances in combat then? There are a few reasons, which I think we've both touched on at different times, but the main thing is, 4X isn't Chess. There is more to the game - as narrative, as simulation of space-opera-style interstellar conflict, which must have random factors, etc. - than a highly-complex strategic logic puzzle. And look, if the balance of these factors makes DG more fun as a game when combat is more random, great! Bring on the dice!

The "4X isn't Chess" is worth expanding on. The reason removing all randomisation from a typical 4X combat system would be a bad idea is because the underlying mechanics aren't nearly as robust as a game like Chess or Go. They aren't robust enough, in large part, because it's extremely difficult (if not impossible) to make them that way without a large amount of abstraction, so you're left with pure math, assuming no AI exploits of course. If anyone is able to devise a 4X combat system that would work in a chess-like fashion, that makes at least some amount of sense in terms of immersion, has no randomness, and doesn't break when you throw a unit designer in the mix, I'd begin work on it in a heartbeat

(Just one other thing though: I'm not sure how clear this was, but finding myself with a gaping hole in my tech tree that makes it an extreme challenge to take on enemy fleets is terrific fun in MoO! Given all the methods of acquiring the tech in spite of these holes I described, the possibility just adds challenge and yes, strategic depth, to the game.)

Definitely agree with that, which is why I mentioned that it's a precipitous balancing act to get it right. MoO 1 has a few tricks for tech tree generation as well to make sure no one gets anything too horrid (like no warp drives).

[RefSteel]

Ah - I misunderstood what you meant about an AI never fighting battles it can't win. Actually if an AI did precog every fight and refused to fight the ones it couldn't win, that would be a small but exploitable weakness. Taking tactical losses for strategic reasons (mainly attrition of enemy forces) can be the correct play in many situations. In any case, with a sufficiently simple combat system, the math is doable for a human too - and more so when the results have fewer random factors. I can do a decent job of "precog" on MoO battles, but it would be way easier if there were no die rolls involved so I wouldn't have to account for swingy breaks on first attacks. Computers, being way better at math than me, can "precog" the probable outcome of each battle, and the probability curve for the various outcomes, even with numerous potential luck swings. Thus, it could certainly refuse a fight in which the likelihood of victory falls below an "acceptable" threshold as surely as it could refuse a no-chance fight in a deterministic system. Or, if well designed, it could choose a "Hail Mary" when it knows that keeping or losing a given planet would make or break its game. And to me, this would be cool, and an argument in favor of randomness even though (or in part because) the number of AIs who might try such a strategy mean the small-but-extant chance of success does favor the AI.

(Chess and Go are also pure math; the math is just so complex that even our most powerful computers haven't solved them yet; though they can outplay any human at Chess and Google is developing one that either is or soon will be the world's best Go player, their play isn't strictly perfect. Whereas checkers, as I understand it, given the normal starting positions, is a solved problem.)

[Reformations]

Glad you like NSW. Not all my ideas are good, but now and then I find a nut.

Another mechanic I am proud of: Overcrowding. This may either be a racial trait (Sakkra) and/or planetary special such as rich.

Overcrowding allows a planet to exceed the population max. OC population does not generate production and cannot control factories. They do not defend in combat and do not count toward galactic vote population. Perhaps another thematic angle is that the particular star has multiple planets and this mechanic abstracts a way for the smaller planets to contribute to population growth but not actual production.

OC population do however form a very useful population buffer that is pulled from first when sending transports. A (maxed) planet with 10 overcrowding may send out 15 transports but only lose 5 active population. These 5 are replaced first (and the production capacity is quickly maximized again) by regular population growth, but then up to another 10 OC population may accumulate on the planet. If 10 or fewer OC population is sent out then the planet does not lose any production capacity. The planet still pays 1 credit per population sent regardless if it was OC or a regular active pop.

It would be interesting to see how this would play out with DG transports as Colship and the lack of cloning. Also, most of my ideas are meant to be modular so this is something that could be turned off/on for more racial (or galaxy generation) flavor.

[Jeff Graw]

Ah - I misunderstood what you meant about an AI never fighting battles it can't win.  Actually if an AI did precog every fight and refused to fight the ones it couldn't win, that would be a small but exploitable weakness.  Taking tactical losses for strategic reasons (mainly attrition of enemy forces) can be the correct play in many situations.  In any case, with a sufficiently simple combat system, the math is doable for a human too - and more so when the results have fewer random factors.  I can do a decent job of "precog" on MoO battles, but it would be way easier if there were no die rolls involved so I wouldn't have to account for swingy breaks on first attacks.  Computers, being way better at math than me, can "precog" the probable outcome of each battle, and the probability curve for the various outcomes, even with numerous potential luck swings.  Thus, it could certainly refuse a fight in which the likelihood of victory falls below an "acceptable" threshold as surely as it could refuse a no-chance fight in a deterministic system.  Or, if well designed, it could choose a "Hail Mary" when it knows that keeping or losing a given planet would make or break its game.  And to me, this would be cool, and an argument in favor of randomness even though (or in part because) the number of AIs who might try such a strategy mean the small-but-extant chance of success does favor the AI.

While I'd argue that your ability to precog is hurt less than the AI's (Might depend on how you look at it. Going from a full 100% accuracy to something less than 100% can be represented as inf), I'm willing to let things lie here since I think we're probably way past the realm of practical with these extreme examples that are admittedly entirely my fault for fielding

(Chess and Go are also pure math; the math is just so complex that even our most powerful computers haven't solved them yet; though they can outplay any human at Chess and Google is developing one that either is or soon will be the world's best Go player, their play isn't strictly perfect.  Whereas checkers, as I understand it, given the normal starting positions, is a solved problem.)

I suppose there's multiple ways of looking at it, but I'd say at the most fundamental level they're memorisation, since you could theoretically store every single position to solve the game. Of course, on the larger Go boards you'd need a device housing several times the number of bits as there are atoms in the physical universe to truly solve the game, so the fundamental is quite removed from the practical. At least until we can create some kind of magic subatomic storage device and an insanely fast processor to calculate all positions.

Another mechanic I am proud of: Overcrowding. This may either be a racial trait (Sakkra) and/or planetary special such as rich.

While I'm happy to discuss the idea, at first blush I'm not much of a fan. I've always felt that moving population around in MoO was largely (relatively simple) math and micro. In DG we've reduced the micro in population movement by an order of magnitude or so via a better UI solution, but haven't really injected much strategy (although removing cloning helps a bit in a roundabout way). Having those "buffer zones" I think would only serve to increase the emphasis on correct micro when it comes to player performance. Eg. if you aren't using the buffers to increase your growth rate, you're playing sub optimally, but the process of using the buffers is not overly strategic and becomes a chore.

[Reformations]

Huh. Don't you have to worry about planets at max pop with or without OC mod in place? I personally don't play that way (pedantic about pop growth) so perhaps you have a point. But if you're obsessed with moving population around at 99/100 or 109/100 -- is there a difference?

I think MoO3 experimented with automatic population spread where nearby colonies would send their excessive growth to neighboring (not maxed) planets in some organic/behind-the-scenes way? Would we see something like that to fix the mathy problem?

Maybe there's a solid argument as a late game mechanic for spamming shock troops with less damage to production in the established colonies? That role makes sense as a racial trait and less as a planetary bonus (unless it was a really big bonus).

[RefSteel]

While I'd argue that your ability to precog is hurt less than the AI's (Might depend on how you look at it. Going from a full 100% accuracy to something less than 100% can be represented as inf), I'm willing to let things lie here since I think we're probably way past the realm of practical with these extreme examples that are admittedly entirely my fault for fielding :)

Basically all of that is patent nonsense. The further randomness in a battle gets from the edge cases (100% deterministic or straight-up coinflip) the harder it is to calculate and keep track of the likely results and the likelihood of each result. With battles that are highly random (with a wide range of possible outcomes with non-negligible probabilities) but also highly weighted, a human's ability to guess the winner and especially how many survivors the victorious fleet will have quickly approaches zero, while a computer can predict all the odds perfectly. You can only get "inf" as your answer by asking a question that is irrelevant or meaningless, like "Okay, but IF the computer only cares about the most likely outcome and makes no allowances for the range of probabilities, then what is the RATIO of the likelihood that it will be wrong if there IS randomness to the likelihood that it will be wrong if there ISN'T?" None of that matters to the ultimate outcome of the game (except for the hypothesized AI being an idiot).

[EDIT: Trying this again so it's a little clearer than mud: What matters is the difference between the abilities of each side to infer how the battle may, can, or will end. The simpler and especially the more intuitive the combat system, and the more nearly deterministic (or more nearly pure-dead-random; it's easy to predict the probability range of a coin flip) the easier it is for humans to get their brains around it. The more you add weirdly-interacting random factors, the more the human is left shrugging and saying, "I have no idea," barring a deep understanding of the rules behind the system and a spreadsheet with which to let a computer do the work instead - which means that even a rudimentary calculation by a computer will do a better job and have a big advantage. Even oversimplifying massively to a binary outcome (which is a mistake) 100% certainty isn't much of an advantage over 95% certainty. 80% certainty has a HUGE advantage over 30% certainty. Using random numbers to make calculations harder doesn't bother computers, which are literally calculating machines, but it can really tie humans in knots unless you come up with a way to make it somehow naturally intuitive.]

I suppose there's multiple ways of looking at it, but I'd say at the most fundamental level they're memorisation, since you could theoretically store every single position to solve the game.

This doesn't matter to the game, but it annoys me because either we're speaking completely different languages or it's strictly wrong. You can't even theoretically memorize information to which you don't have access, so with no database of all possible moves in a game, it isn't memorizable. Even if there were a database of every possible position, you'd have to calculate how to get from each position to the next, which is math, not memorization. Also note the best Chess computers use "memorization" only for opening moves and endgames, and as I understand it the best Go computers barely rely on "memorization" for anything, if at all.

While I'm happy to discuss the idea, at first blush I'm not much of a fan. I've always felt that moving population around in MoO was largely (relatively simple) math and micro. In DG we've reduced the micro in population movement by an order of magnitude or so via a better UI solution, but haven't really injected much strategy (although removing cloning helps a bit in a roundabout way).

The UI I saw in the video would be a good addition to MoO's, but is it meant to be a replacement? When the number of planets with spaceports gets large, I feel like it could be daunting. I'm also not convinced that removing the ability to grow pop helps, especially in combination with the interface as I understood it from the video: You have to remember which half-dozen (or more) planets you want to use to send a little pop each so they'll stay in the sweet spot of growth/factories/etc. and that seems way more micro-intensive in the later game than what I do in MoO: Pick a nearby planet or two whose production I've decided I can spare, and have them hurl a big raft of transports over to the new planet and regrow immediately.

(There is some super-micro-intensive stuff in MoO, like the pop-growth super-optimization Thrawn talked about in our SG thread, and getting rid of forced-pop-growth kills half of that, but that would still leave the transport half on the table unless - as I'd hope - DG uses a growth formula that works a little more smoothly and so isn't as easy to game ... which should solve both halves of that problem, I would think.)

Huh. Don't you have to worry about planets at max pop with or without OC mod in place? I personally don't play that way (pedantic about pop growth) so perhaps you have a point. But if you're obsessed with moving population around at 99/100 or 109/100 -- is there a difference?

For me, the micro point with OC would be around 100/110, and I think it would be a lot more important than 99/100: The trouble is that you never want to have anyone in OC until you've maxed out all the potentially-productive population in your entire empire, and then with more space to grow into (if similar formulae are used) you'll grow useless-unless-transported population faster than you'd regrow from 99 in MoO, so you likely have to pay attention to every planet close to the OC point more or less every turn.