Let's Play Dominus Galaxia

[RefSteel]

It seems like there are some really neat things about this game, and I'm looking forward to seeing how it develops!  Some thoughts on the discussion, in case they're helpful.

Keep in mind that, although we don't yet, there's nothing stopping us from modifying crits and flanking via design. For example, a weapon that has a large crit range, but is less reliable. Or a special that reduces overall HP slightly (or something else, like movement speed, initiative, etc.) but reduces flanking bonus against it. Of course, the hard part is balancing these so that they aren't any no brainers, that they have counters, etc.

So.  I love MoO, and I love its simple but many-layered combat resolution, but my least favorite thing about that combat system is the amount of time it takes certain weapons and specials (worst offender: Repulsor Beam) to complete their animations, because lengthy tactical combat encounters take time away from playing the actual strategy game.  It's not unusual (in a <50-star standard MoO galaxy!) to have combat at half a dozen or more different stars in a single turn, and many of those will involve one or two stacks of much more than six ships in the mid and late game.  Outside of the case mentioned above (repulsors against an AI that doesn't know it should be retreating) that's not too bad in MoO because I can resolve most battles in half a dozen clicks or less.  But in DG, if I manually resolve combat, even if the distances (relative to ship combat speed) are the same as in MoO - and it looks to me like in fact the distances involved are much greater - it will take me three to six times as many clicks because of automatic stack-splitting alone!

Now, that analysis isn't really fair to DG:  If the combat AI really does know how to play ~optimally (and I'm willing to trust that it does) I'll only need to manually resolve fights where I have a strategic objective the auto-resolver doesn't understand.  (E.g. preserve one particular ship at all costs, or take any casualties necessary to destroy all a planet's missile bases or say the sporers in the enemy fleet...)  But if that's the case, why bother to include fiddly mechanics like flanking that just add a piece to a tactical puzzle that the auto-AI has already solved, and little or nothing to the strategic game?  Changing existing (or adding new) researchable weapons and systems to better exploit those tactical mechanics only exacerbates the problem.  When I'm planning my empire's research, I want to be aiming for high-impact, big-picture goals, not calculating which special system I need to best exploit the special cases of flanking or crits.

(Another stack-splitting issue that auto-resolve doesn't fix:  Effective goaltending or just the map's geometry can limit the number of approaches to a target that can all arrive in firing distance within the minimum number of turns.  What if I don't want to split my stack?  Do I have to send a token ship apiece of five other designs just to be sure my bombers will stick together and won't hit a traffic jam with each other when they try to get into bombing range?  This is easy to fix though if you offer the option for the player to split stacks prior to combat manually.)

As for crits:  They increase variance, with the amount of extra variance depending on how often they come, how much extra damage they cause, and on what those two factors depend.  If crits are rare and smallish in effect, they can prevent a single shield-4 fighter from holding off an arbitrarily large laser fleet, and can create huge luck-based swings in important early battles (because later battles, if important, should involve a lot more ships to spread the variance around) but otherwise won't have much impact either way.  If crits are common, it's basically the same thing as increasing the damage range of different weapons, except that there's another fiddly mechanic to worry about on the strategic level (and I suppose from a marketing perspective, some players presumably think crits are cool).

This might be a bit on the simplistic and rigid side, but would probably be a bit better than MoO 1 where you have this inescapable graduation to larger hull sizes being superior (except for a few explicit roles like bombing of course!).

Actually, for skilled MoO 1 players, small hull sizes are superior to larger hull sizes in almost all circumstances:  Mediums are for carrying critical weapons you can't quite fit on a fighter, Large hulls are used when needed to carry specific pieces of equipment like colony bases or repulsors (or pulsars to deal with the other side's clouds of fighters) and Huges (or often just Larges even for this role) are for battles you're going to win easily, where you might as well avoid taking any attrition.  The Small ship advantage is the main reason pulsars exist, and it's difficult to make even these an effective counter to a well-designed small-hulled fleet. [EDIT: To be clear, there are of course many exceptions for specific game situations - one of the great things about MoO is that there's almost never a One Right Answer - but the rule to which these exceptions exist is the power of a swarm of well-designed small ships.]

[*]Another idea is to move things over to equipment. You could have equipment that only works on certain hull sizes, and you could do this either organically (eg. the equipment is too large to be practical in smaller hulls, or scales in way that it becomes impractically spacious in larger hulls, this later bit working better for specials than weapons for obvious reason), or rigidly (weapons or specials have an explicit size requirement), or could be based on the attributes of the equipment itself. For example, just as pulsars work best against smaller ships, you could have a weapon that is very good against larger hulls, but has a big penalty when used against smaller ones.

MoO does the organic scaling thing for many (really all) types of equipment.  It also has weapons that are great against large hulls and lame against small ones:  Basically ever big non-streaming, non-MRVing weapon in the game, from Heavy Blast Cannons and Stinger Missiles to Death Rays, Zeon Missiles, and Plasma Torpedoes.  They don't need any special drawbacks; they just have to hit really hard - a problem for big ships, but the equivalent against armorless, unshielded smalls of "3 points of damage, flat, per hit."  This is a big part of the reason smalls are so powerful, in fact.

[[[Various implementations of regenerating shields with hitpoints]]]

This looks kind of cool and different from up close, but if you take a couple steps back and look at the big picture, the effect is more or less the same as auto-repair.

TL/DR:  I'm a mechanics minimalist - one of the reasons I love MoO - so maybe take this with a grain of salt.  But increasing the complexity of tactical combat makes balancing the game harder, a good AI harder to program without missing unexpected strategic or tactical holes, and the learning curve for new players steeper.  It's possible that potential "cool factors" can make this worthwhile anyway - I dunno; that's not my department.

[Reformations]

Really great points Ref. I'm glad I'm not alone is being skeptical of flanking/crits. Your comment about AI being able to optimize flanking positions suggests it is a solved problem and the rest are left as exercises for the player during combat (my words, not yours) was interesting. I'm perhaps more cynical in that the AI won't be able to effectively flank/crit and therefore this mechanic is introducing another hiccup for the player to exploit against AI (repulsor etc). I don't really feel superior when beating an AI with repulsors. I imagine I'll feel the same after the 10th time I flanked an unwitting AI ship.

I love backstabbing crits in Icewindale as much as the next guy but it falls out of scope when galactic battles are going on. This is what the law of large numbers is for.

[quote='RefSteel' pid='619548' dateline='1482984008']
If crits are common, it's basically the same thing as increasing the damage range of different weapons, except that there's another fiddly mechanic to worry about on the strategic level (and I suppose from a marketing perspective, some players presumably think crits are cool).
[quote]

I think the game has high damage variance already. First off, I love love love that the starting chance to hit is 50% and remains so throughout the game with all things considered equal. I don't want my computer tech to take me from 90% to 92% chance to hit. (Also, half of all shots missing is maximizing variance of a Bernoulli.) Second, the game has relatively high base damage ranges with low minimum damage. Those minimum damage values grow very slowly (lots of 1's, 3's, and 4's). This makes for great interaction with shields by introducing even more '0 damage' hits --> more variance.

Putting all this together, I believe crits already exist in the game as the designers knew it would take a 'good' roll to not only hit, but also get through shields and do damage. Law of large numbers has absorbed the individual crit process into the damage calculations while maintaining the integrity of shields.

Although I'm not a fan of either, I think crits are more damaging to MoO mechanics than flanking. What about giving flanked attacks a +ATT bonus? This increases the expected damage of the attack and in some cases increases the minimum damage possible. Or perhaps define a crit as an additional % chance to do max damage of the weapon (but not exceed the max damage). Both of these keep shields mechanic in familiar territory. I would be most interested in seeing these mechanics as weapon modifications rather than tactics overload, though.

[Reformations]

Sorry for double post, but this is slightly related.

On your game engine, would it be possible to display (in small numbers that float away quickly) a random sample of the actual damage done by each shot? There would still be the overall sum of damage displayed in the numbers as you have now.

The numbers would be: white * indicating a miss, blue 0's indicating shield blocked, and orange numbers indicating raw damage actually done to HP of the ship. (the large total damage of all the attacks would be red).

I think this type of feedback to the player would go a long way in informing them how the battle computers and shields should influence their ship design.

[Jeff Graw]

So.  I love MoO, and I love its simple but many-layered combat resolution, but my least favorite thing about that combat system is the amount of time it takes certain weapons and specials (worst offender: Repulsor Beam) to complete their animations, because lengthy tactical combat encounters take time away from playing the actual strategy game.  It's not unusual (in a <50-star standard MoO galaxy!) to have combat at half a dozen or more different stars in a single turn, and many of those will involve one or two stacks of much more than six ships in the mid and late game.  Outside of the case mentioned above (repulsors against an AI that doesn't know it should be retreating) that's not too bad in MoO because I can resolve most battles in half a dozen clicks or less.  But in DG, if I manually resolve combat, even if the distances (relative to ship combat speed) are the same as in MoO - and it looks to me like in fact the distances involved are much greater - it will take me three to six times as many clicks because of automatic stack-splitting alone!

Now, that analysis isn't really fair to DG:  If the combat AI really does know how to play ~optimally (and I'm willing to trust that it does) I'll only need to manually resolve fights where I have a strategic objective the auto-resolver doesn't understand.  (E.g. preserve one particular ship at all costs, or take any casualties necessary to destroy all a planet's missile bases or say the sporers in the enemy fleet...)  But if that's the case, why bother to include fiddly mechanics like flanking that just add a piece to a tactical puzzle that the auto-AI has already solved, and little or nothing to the strategic game?  Changing existing (or adding new) researchable weapons and systems to better exploit those tactical mechanics only exacerbates the problem.  When I'm planning my empire's research, I want to be aiming for high-impact, big-picture goals, not calculating which special system I need to best exploit the special cases of flanking or crits.

(Another stack-splitting issue that auto-resolve doesn't fix:  Effective goaltending or just the map's geometry can limit the number of approaches to a target that can all arrive in firing distance within the minimum number of turns.  What if I don't want to split my stack?  Do I have to send a token ship apiece of five other designs just to be sure my bombers will stick together and won't hit a traffic jam with each other when they try to get into bombing range?  This is easy to fix though if you offer the option for the player to split stacks prior to combat manually.)

I don't think that MoO 1's combat would stand up particularly well with a hyper-competent AI of Alex's variety. And of course a combat system that only exists as a puzzle for players to solve would make it very difficult to create a non-cheating AI that was challenging, even if a lot of players might find it entertaining. We're in a kind of uncharted territory in that regard because developers typically don't try to make competitive tactical AIs, especially not for turn based games, and to be honest it's going to require a lot of trial and error to get it right -- assuming we can. At the very least, we'll have a very nice foundation for auto-resolve that the player can go and play around in, which by itself should feel much better and more authentic than the likes of GalCiv or Endless Space that don't even really try. I hope we can do better though.

But you've identified the conundrum. A simple puzzle isn't good. A solvable thing where the player can only ever hope to match the AI, at which point they may as well just autoresolve, isn't good either (probably worse than the former, but that's a matter of opinion). That's why a lot of thought is going in the direction of increasing the complexity of the system, even when everywhere else we've been reducing complexity. Many different variables interacting in non-linear ways leading to many unique and emergent situations that the AI handles "good enough," but with too much variety to ever fall into the realm of rote response for the player. It is of course really difficult to get right, and there's the ever present danger of adding so much complexity that the thing becomes unapproachable. The other choice is to settle (more or less) on a puzzle, like MoO 1, because many players will like that, or go very abstract. The former isn't out of the realm of possibility, but the later likely is since it would require massive reworking of just about everything.

Of course keep in mind that way less time has been spent on the tactical side of the game than it has the strategic. It's going to see a lot more work in the future, and ideally, we'd have a dev dedicated to that system. One thing I'd like to reassure you about though, is that we're pretty good at avoiding the sunk cost fallacy. If something terrible comes out as a result of our efforts, our team is the kind that will back away from it and try again. Also keep in mind that Alex is the kind of guy that would plug any kind of AI exploit someone finds as a matter of pride, even way after release. In fact, a certain regular here at RB recently found one involving luring AI ships with missile boats to areas of the map lacking substantial cover, which Alex is working on fixing.

As far as stack splitting, that's actually been a matter of debate within the team. I'm pretty much the solitary guy who doesn't really like it. I wouldn't mind making it a toggle, but "game mechanic toggles" are generally evil, fragment the playerbase, and weaken overall vision. I would make an exception, but there's some resistance to the idea. A lot of it will depend on player feedback, but disabling the auto-stack splitting would take all of 5 minutes. Of course, we might also double down on it and give the player agency over stack splitting and unit positioning at the start of each encounter.

As far as pacing goes, for me at least, it is much improved versus MoO 1. That said, it might be hard to tell in the Let's Play because Alex is going along at a deliberately slow pace. Not only can you autoresolve at any point in a combat event, but the AI isn't likely to send forces that are too weak, and will move its fleets away before you even arrive to avoid a battle it knows it can't win. While I've experienced fatigue from constant battles in MoO 1, I haven't ever had that same experience with DG.

As for crits:  They increase variance, with the amount of extra variance depending on how often they come, how much extra damage they cause, and on what those two factors depend.  If crits are rare and smallish in effect, they can prevent a single shield-4 fighter from holding off an arbitrarily large laser fleet, and can create huge luck-based swings in important early battles (because later battles, if important, should involve a lot more ships to spread the variance around) but otherwise won't have much impact either way.  If crits are common, it's basically the same thing as increasing the damage range of different weapons, except that there's another fiddly mechanic to worry about on the strategic level (and I suppose from a marketing perspective, some players presumably think crits are cool).

They err more towards the uncommon end of the spectrum. But I imagine that the values will be tweaked quite a lot before launch either way.

Actually, for skilled MoO 1 players, small hull sizes are superior to larger hull sizes in almost all circumstances:  Mediums are for carrying critical weapons you can't quite fit on a fighter, Large hulls are used when needed to carry specific pieces of equipment like colony bases or repulsors (or pulsars to deal with the other side's clouds of fighters) and Huges (or often just Larges even for this role) are for battles you're going to win easily, where you might as well avoid taking any attrition.  The Small ship advantage is the main reason pulsars exist, and it's difficult to make even these an effective counter to a well-designed small-hulled fleet. [EDIT: To be clear, there are of course many exceptions for specific game situations - one of the great things about MoO is that there's almost never a One Right Answer - but the rule to which these exceptions exist is the power of a swarm of well-designed small ships.]

Right, bad example, and while I'm not the greatest MoO 1 player (I do win on impossible from time to time, so I'm not terrible), I have seen that approach and tried it a few times myself. The takeaway point is that more distinct roles guard against one hull being overbearing or the default, simply because no hull can fill all of the necessary roles (or at least, not well) by itself.

[[[Various implementations of regenerating shields with hitpoints]]]

This looks kind of cool and different from up close, but if you take a couple steps back and look at the big picture, the effect is more or less the same as auto-repair.

Similar from a high level, but very different (or at least, has the potential to be) when you sit back and think about it. For starters, auto repair doesn't have much cost besides space, and the special slot, and its relevance is very binary. On the other hand, the shield mechanics being talked about exist on a three dimensional continuum where strength in an area(s) can have a weakening effect in another/others. There are three separate but interacting variables at play; a shield with high DR, low regen, and medium HP is significantly different than a shield with medium DR, medium regen, and medium HP, which is significantly different than a shield with low DR, high regen, and high HP, which is significantly different than a shield with high DR, high regen, and low HP, etc. etc. etc. which is before even getting into cost and size. That and different weapons combined with different compositions of fleets and size of the engagement can result in more scenario combinations than one could hope to memorise or figure out a way to exploit. Well, theoretically, at least. But again, it's just something that has been tossed around, not something that is actually planned at this point.

TL/DR:  I'm a mechanics minimalist - one of the reasons I love MoO - so maybe take this with a grain of salt.  But increasing the complexity of tactical combat makes balancing the game harder, a good AI harder to program without missing unexpected strategic or tactical holes, and the learning curve for new players steeper.  It's possible that potential "cool factors" can make this worthwhile anyway - I dunno; that's not my department.

I'm also a mechanics minimalist, as is the rest of the team! We actually did things like getting rid of colony ships, explicit waste cleanup, research interest, the power attribute, and even reserve fuel tanks. Tactical combat is the exception where we've added a bit of complexity. Not for the cool factor, but again to try to find that elusive sweet spot where it's not a puzzle that devolves into rote response, yet player agency is still important. And no, we aren't there yet. And admittedly it's a very difficult problem that is also very easy to mess up!

Really great points Ref. I'm glad I'm not alone is being skeptical of flanking/crits. Your comment about AI being able to optimize flanking positions suggests it is a solved problem and the rest are left as exercises for the player during combat (my words, not yours) was interesting. I'm perhaps more cynical in that the AI won't be able to effectively flank/crit and therefore this mechanic is introducing another hiccup for the player to exploit against AI (repulsor etc). I don't really feel superior when beating an AI with repulsors. I imagine I'll feel the same after the 10th time I flanked an unwitting AI ship.

I'd say it's in between the two extremes. The AI hasn't solved it, nor is it easy to exploit. From experience, flanking did actually feel good in a perverse way when the AI wasn't very good at it, because it gave the player the impression of solving a puzzle and easily beating a superior force is self affirming... but that's not the game we're trying to make.

I think the game has high damage variance already. First off, I love love love that the starting chance to hit is 50% and remains so throughout the game with all things considered equal. I don't want my computer tech to take me from 90% to 92% chance to hit. (Also, half of all shots missing is maximizing variance of a Bernoulli.) Second, the game has relatively high base damage ranges with low minimum damage. Those minimum damage values grow very slowly (lots of 1's, 3's, and 4's). This makes for great interaction with shields by introducing even more '0 damage' hits --> more variance.

Putting all this together, I believe crits already exist in the game as the designers knew it would take a 'good' roll to not only hit, but also get through shields and do damage. Law of large numbers has absorbed the individual crit process into the damage calculations while maintaining the integrity of shields.

Well, it does and it doesn't. It really depends on the number of hits. 5 hits? Sure, lots of variance. 500 hits? The variance is largely smoothed out. In DG, either every single weapon in that bank on every single ship in that stack crits, or none of them do. The variance crits afford isn't smoothed out by a larger sample size the same way that damage range is. Of course, this also makes crits feel very significant.

Although I'm not a fan of either, I think crits are more damaging to MoO mechanics than flanking. What about giving flanked attacks a +ATT bonus? This increases the expected damage of the attack and in some cases increases the minimum damage possible. Or perhaps define a crit as an additional % chance to do max damage of the weapon (but not exceed the max damage). Both of these keep shields mechanic in familiar territory. I would be most interested in seeing these mechanics as weapon modifications rather than tactics overload, though.

The first is a bit unlikely, since I imagine we'll want to keep attack rating as integer, and flanking works better along a spectrum. So things like +x% damage, x% crit chance, etc. work best. The second sounds like a possible direction we could go in, although I'm actually the reverse of you - I'm far more confident in the current implementation of crits (I think they feel quite good) than I am the implementation of flanking!

On your game engine, would it be possible to display (in small numbers that float away quickly) a random sample of the actual damage done by each shot? There would still be the overall sum of damage displayed in the numbers as you have now.

The numbers would be: white * indicating a miss, blue 0's indicating shield blocked, and orange numbers indicating raw damage actually done to HP of the ship. (the large total damage of all the attacks would be red).

I think this type of feedback to the player would go a long way in informing them how the battle computers and shields should influence their ship design.

Yup, totally agreed, or at least something along the same lines. This has been pointed out by others too. Haven't had the time to get around to it yet though.

[Reformations]

I really am excited about the project and am impressed with galaxy interface/playstyle (be careful with buildings though).

I'm surprised your team is going in seemingly opposite direction with combat. You had mentioned %damage as a modifier. I think that is a terrible idea for the same issue as crits. Raw mitigation (moo1 shields) fall flat against %multipliers. Then you have to talk about %dr ratings of shields and the arms race is on. MoO1 weapons are very much intended to be obseleted by new shields. That binary result of 0 damage is the motivation to design a new ship.

[bjg]

Is there some special requirement to still be using a 1993 computer to qualify as a "MoO 1 diehard" ?

2003 would be nice. Or (not top of the line) 2010. Or (modern) ultrabook.
Anyway, seems like this is not my cup of tea. It's too bad I've already paid for it.

[RefSteel]

I don't think that MoO 1's combat would stand up particularly well with a hyper-competent AI of Alex's variety. And of course a combat system that only exists as a puzzle for players to solve would make it very difficult to create a non-cheating AI that was challenging, even if a lot of players might find it entertaining. We're in a kind of uncharted territory in that regard because developers typically don't try to make competitive tactical AIs, especially not for turn based games, and to be honest it's going to require a lot of trial and error to get it right -- assuming we can.

I'd love to see a hyper-competent combat AI in MoO 1. What I may not have made clear is that I want to hit "Auto-resolve" every time unless I have a special strategic goal for which I'm willing to sacrifice "optimal" results in a single combat. I don't really like tactical puzzles because I think they ultimately will necessarily boil down to repeatable exploits or be impossible to play better than the AI - except in those special cases when you have an unusual goal for the battle. I may be quite unusual in this respect though.

Of course, we might also double down on it and give the player agency over stack splitting and unit positioning at the start of each encounter.

This is what I'd love to see from stack splitting, yeah: As long as there are two key buttons when (or before) the splitting/positioning options appear: 1) Auto-split, and 2) Do Not Split, so that the whole step can be skipped quickly if a player so wishes for an individual battle.

While I've experienced fatigue from constant battles in MoO 1, I haven't ever had that same experience with DG.

I believe it - and I think the reason is Alex's combat AI. If Auto-resolve is a good option and displays quickly, for me that's the best case scenario for most battles - especially because PBeM multiplayer might be possible with all battles auto-resolving, perhaps with an option for players to review their events during their next turn (after the fact).

Similar from a high level, but very different (or at least, has the potential to be) when you sit back and think about it.

I do see some differences on the margins, yeah, the idea being that all ships are vulnerable to lasers-in-sufficient-numbers but some ships can handle them better than others relative to heavier beams. I am concerned that any idea that makes starting weapons viable against late game defenses (in sufficient numbers of course) will seriously skew the strategy of the game though, particularly in the value of weapon and shield technologies. Among other possibilities, I could see everything else in two trees becoming ~filler for ground combat bonuses and scatter packs, with planetary shields becoming a punchline, though of course this depends on implementation.

Well, it does and it doesn't. It really depends on the number of hits. 5 hits? Sure, lots of variance. 500 hits? The variance is largely smoothed out.

Exactly! As the game goes on and more resources are committed to a battle, ship design and production decides the battle, and die rolls are largely smoothed out of the equation.

In DG, either every single weapon in that bank on every single ship in that stack crits, or none of them do.

Wait.

What.

Okay, so this is indeed high variance! This is very high variance, especially if the probability of crits is low and their impact on damage is large! This is ... potentially variance so high that it swamps everything else in the combat system?

Why is it necessary or appropriate to make crits "feel significant"? (I guess there's Nero Wolfe's answer: "We are all vainer of our luck than of our merits.")

I'm far more confident in the current implementation of crits (I think they feel quite good) than I am the implementation of flanking!

This has been true ever since the first time some middle-school kid or someone first introduced the idea of critical hits as a house rule in a D&D game: Crits feel really good! ... when you roll one. Crits are the worst! ... when one is rolled against you.

I mean. I'm probably not your target audience, and it's very possible that many people will love those massive crits, and may or may not build their star fleets and research plans around abusing or avoiding them. From my lonely perspective though, full-stack crits would be enough by themselves to turn me off from the game. I'm just not interested in seeing an important battle turn on whether some stack of obsolete pea-shooters gets a lucky die roll (for me or for the enemy).

[Jeff Graw]

I'd love to see a hyper-competent combat AI in MoO 1.  What I may not have made clear is that I want to hit "Auto-resolve" every time unless I have a special strategic goal for which I'm willing to sacrifice "optimal" results in a single combat.  I don't really like tactical puzzles because I think they ultimately will necessarily boil down to repeatable exploits or be impossible to play better than the AI - except in those special cases when you have an unusual goal for the battle.  I may be quite unusual in this respect though.

Ah, I get where you're going. I don't think you're unusual, but there are many different camps when it comes to tactical combat. Given what you've said, you'd probably actually quite enjoy the current iteration of combat. When I play for fun, I hit Auto-Resolve 90% of the time. If we can't meet the lofty goal of tactical combat where there's both room for the player to outsmart the AI (and be outsmarted by the AI), and not have it devolve into a repeatable puzzle that only need be solved once and then repeated, we'll most likely settle on what you describe as your ideal as then next best thing. Heck, Alex would probably revolt if we tried to make him design an AI that was there to be exploited by the player!

A thing to keep in mind, and apologies if I'm sounding like a broken record on this point, is that we haven't put nearly as much time into the combat system as we have the strategic layer. It might ease you to know that we implemented a number of ideas and went down a number of different paths on the strategic layer, that sounded good in principle but worked poorly in practice, and we backed away from them and went in different directions. I have no doubt that when we shift our focus to the combat side of things that history will repeat itself to some extent. Some ideas won't work well and we'll need rethink them. But it can be difficult to know for sure until there's a working model to evaluate.

I believe it - and I think the reason is Alex's combat AI.  If Auto-resolve is a good option and displays quickly, for me that's the best case scenario for most battles - especially because PBeM multiplayer might be possible with all battles auto-resolving, perhaps with an option for players to review their events during their next turn (after the fact).

As far as multiplayer goes, we're trying to take a more sensible approach. The sad truth is that, unless we make a really good game, no one is going to want to play it multiplayer anyway. A game like SoTS 2 would probably have came out a lot better if they didn't waste so much effort on the multiplayer mode no one ended up using. Even a big budget title like nuMoO has practically no multiplayer community.

That said, multiplayer is something we absolutely want to do. With that in mind, we're structuring things with multiplayer in mind so if we ever do get to the point where multiplayer makes sense, there won't be too many road blocks in the way. Brent actually had a pseudo hot-seat mode up and running way back, although I think that's been broken for awhile.

I do see some differences on the margins, yeah, the idea being that all ships are vulnerable to lasers-in-sufficient-numbers but some ships can handle them better than others relative to heavier beams.  I am concerned that any idea that makes starting weapons viable against late game defenses (in sufficient numbers of course) will seriously skew the strategy of the game though, particularly in the value of weapon and shield technologies.  Among other possibilities, I could see everything else in two trees becoming ~filler for ground combat bonuses and scatter packs, with planetary shields becoming a punchline, though of course this depends on implementation.

Yup, strikes me as more of a balance than a systems thing. Of course, we're only talking about a mechanic that has been discussed rather than one that has been decided on. The main argument, for me, against that kind of shield mechanic is the overall complexity of the thing. I'm not sure if the ends justify the means. But I'd be surprised if there wasn't a way to balance it.

Okay, so this is indeed high variance!  This is very high variance, especially if the probability of crits is low and their impact on damage is large!  This is ... potentially variance so high that it swamps everything else in the combat system?

In practice, I think I can pretty confidently state that it doesn't. It can make a difference, but typically not an outsized one. Criticals are a part of combat, but don't dominate it.

This has been true ever since the first time some middle-school kid or someone first introduced the idea of critical hits as a house rule in a D&D game: Crits feel really good! ... when you roll one. Crits are the worst! ... when one is rolled against you.

From my lonely perspective though, full-stack crits would be enough by themselves to turn me off from the game.  I'm just not interested in seeing an important battle turn on whether some stack of obsolete pea-shooters gets a lucky die roll (for me or for the enemy).

Again in practice, with our current values I don't think crits feel that lucky. A critical adds 50% extra damage, and the chance is very low (5%) when attacking an enemy facing you head on. When one flanks an enemy (which starts outside of its 60 degree forward arc), the chances to score a critical go up (right now this is the only bonus you get from flanking), and are maximised directly aft of the enemy at which point there's a moderate chance of scoring a critical (at the current values, it increases the odds by 33%). For the most part, when a critical is scored against you, you're able to trace it back to "I let myself get flanked." Rather than feeling unlucky when a crit is scored against me, I feel relief when I get flanked and the shot doesn't crit. Even though the chances are still better than even that there will be no critical, my default state of mind when I see myself get flanked is to prepare for consequences of the critical hit.

My bigger issue is that flanking should be more difficult to pull off, or have a higher opportunity cost (more interesting and varied types of terrain might help, or controlling effects of some sort). What happens when you flank or get flanked feels good, but the actual act of flanking could be better.

That binary result of 0 damage is the motivation to design a new ship.

I don't think that's quite right. Certainly, it happens from time to time (especially with early bombers, and laser craft), but for the most part the decision to build a new ship (and/or scrap a previous design) is a somewhat complex one involving the viability of old designs versus maintenance fees and the payout from salvage, plus the need to create a new design to fill a new role or better perform an existing one, plus the consideration of what technologies are coming in the future that might let you create an even better design and whether it's worth the wait.

[RefSteel]

If we can't meet the lofty goal of tactical combat where there's both room for the player to outsmart the AI (and be outsmarted by the AI), and not have it devolve into a repeatable puzzle that only need be solved once and then repeated

To me, accomplishing this lofty goal means simplifying combat as much as possible (so that human players can make reasonable calculations reasonably quickly) and then making sure that combat can serve a wide range of strategic goals: The best way to minimize your casualties in the fight isn't the same as the best way to maximize enemy casualties, which isn't the same as the best way to maximize the number of spore ships destroyed at their muster point before they can retreat, etc.

As far as multiplayer goes, we're trying to take a more sensible approach. The sad truth is that, unless we make a really good game, no one is going to want to play it multiplayer anyway. A game like SoTS 2 would probably have came out a lot better if they didn't waste so much effort on the multiplayer mode no one ended up using. Even a big budget title like nuMoO has practically no multiplayer community.

Good point, and I'm glad to hear you're nonetheless planning things so that making multiplayer possible won't be prohibitively difficult.

Yup, strikes me as more of a balance than a systems thing. Of course, we're only talking about a mechanic that has been discussed rather than one that has been decided on. The main argument, for me, against that kind of shield mechanic is the overall complexity of the thing. I'm not sure if the ends justify the means. But I'd be surprised if there wasn't a way to balance it.

I agree pretty much completely.

Again in practice, with our current values I don't think crits feel that lucky.

Really? You've never seen an early crit take out a large, heavily armed and shielded ship that wouldn't otherwise have been lost? The individual crit might or might not feel like a big deal in the moment it happens, but the carry-over effects in the course of a battle can be enormous.

A critical adds 50% extra damage and the chance is very low (5%) when attacking an enemy facing you head on. When one flanks an enemy (which starts outside of its 60 degree forward arc), the chances to score a critical go up (right now this is the only bonus you get from flanking), and are maximised directly aft of the enemy at which point there's a moderate chance of scoring a critical (at the current values, it increases the odds by 33%).

The low chance is exactly the problem. If it were a high chance - and in fact if flanking is as easy as you imply below, it might be high enough already - it would just mean weapons have a higher damage range and kind of weird damage distribution. You'd just have to adjust the effectiveness of shields accordingly. (But this is in fact quite important!) That 5% chance of a huge damage boost for enemy ships you have completely outmaneuvered (and larger but still smallish chances of a huge damage boost in most normal situations) basically just gives the AI an unusually-visible artificial combat boost because there are more of them, so they have more opportunities to get lucky and slag a ship they had no business hurting seriously if at all.

Unless I'm misunderstanding the impact of a crit: 50% of what, exactly?

Possibility 1: The attack and damage results for each weapon bank occur normally, subtracting shields. Total the damage, then roll to crit. If crit, then increase total damage by 50%.

This seems more convoluted than it has to be, but okay; either in the form of crits or with simpler deterministic +% damage (i.e. up to +17% for a rear attack, without an extra die roll needed) it makes flanks valuable if you want that, and shouldn't break anything.

Possibility 2: Roll to crit. If crit, then increase the damage of each hit by 50% before applying shields.

This is the one I was assuming you meant, and it would mean effective damage against shielded targets would increase enormously. A hull with shield 6 isn't immune to Ion Cannons in MoO, but it can largely shrug them off - and it should be able to; that's a huge amount of shielding against one of the earliest fighter weapons! Let's say you have a big cloud of ion fighters that can still pling it down in four or five turns if it ignores them - but it's heavily armed, with excellent targeting, so they'll lose a bunch to attrition and the shielded ship will survive easily.

Until they get a lucky crit on their first attack and kill it immediately. Depending on rounding, "+50%" in this system would be between +380% and +480% to actual damage output for Ion Cannons against shield 6 - and of course +(divide-by-zero-error) damage for basic Lasers against Shield 4.

Possibility 3: A wide range of other possibilities, up to and including crits auto-hitting and dealing damage equal to 1.5 times the weapons' normal max damage.

I'm assuming these aren't the case, but they would obviously be even worse (from my "I don't like large variance" perspective).

Of course none of this affects my beloved fighter and bomber fleets. Crit 'em as much as you want. Instead of massively overkilling one measley unshielded armorless small per hit, you're really massively overkilling one measley unshielded armorless small per hit!

For the most part, when a critical is scored against you, you're able to trace it back to "I let myself get flanked."

Or "The auto-AI let itself get flanked after I hit Auto; ugh do I seriously have to manually control every ship and watch for flanks in every combat round of every fight where I have a ship I care about?" Or "they rolled that 1 in 20 chance, which is by no means negligible, as the designers must know or else they'd have made it zero, when I wasn't flanked at all."

Rather than feeling unlucky when a crit is scored against me, I feel relief when I get flanked and the shot doesn't crit. Even though the chances are still better than even that there will be no critical, my default state of mind when I see myself get flanked is to prepare for consequences of the critical hit.

I mean, which one someone feels is a matter of personality (lots of people are upset when they lose >60% fights in Civ, and would be upset when they get crit on the max <40% chance from behind) but it doesn't matter which the player feels: Whether your ships are lost because the enemy was lucky, or saved because you were lucky, it's all still a Big Huge Dieroll deciding the fate of your fleet.

My bigger issue is that flanking should be more difficult to pull off, or have a higher opportunity cost (more interesting and varied types of terrain might help, or controlling effects of some sort). What happens when you flank or get flanked feels good, but the actual act of flanking could be better.

While respectfully disagreeing with flanking as a game concept, I could suggest: What if you only get the bonus for flanking when your target is also "pincered" by another of your ships? I would hate this mechanic (like everything that adds tactical complexity ~independent of the strategic game) but again that doesn't mean it wouldn't be fun for others.

[Reformations]

Ref has hit on a lot of reasons why I think crits are more dangerous than flanking. There actually is a small bit of positioning that influences damage in MoO1 combat: a defender gets +1 DEF for each range beyond 1 he is attacked from. Is this present in DG? If so I wouldn't mind seeing it exaggerated a bit more where basic weapons have range of 2 but firing from far away is quite inaccurate through ATT/DEF calculations. However, I would still suggest leaving angles (flanking) out of it. Close should be dangerous no matter which side of ship you are facing.

The ALL or NOTHING crit of a weapon bank also threw me for a loop. What a weird way to do it (besides satiating the desire for big numbers). How does streaming damage work when it crits? What about enveloping? Do multishot weapons cirt on all of thier hits. OOOOOWWWW!! (Multi-hit weapons are particularly resistant to shields and hence the damage increase of a multi-hit crit is getting obscene).