Thoughts on Golems (Rant of Sorts)

[Doddler]

I've messed around with golems a lot recently, and I have to agree that they are powerful forces.  They have little in the way of weaknesses, they don't count as heavy units and hence aren't vulnerable to bombers or other similar heavy damage ships.  But they cost you, and they keep costing you as long as you keep them around.  The very action of using one incurs additional costs, and none of them can really do anything you can't do with a good fleet of a few hundred ships.

Golems cost a lot.  It's obvious that they cost AI progress.  100 AI progress isn't a drop in the hat, its like 15%-20% of the average progress I'll rake up in a game.  The other costs aren't really that obvious until you actually quantify them. The actual cost of repairing a golem I found is as follows:

Level 1 Engineer: 180 minutes for a total cost of 600,000 metal/crystal
Level 2 Engineer: 135 minutes for a total cost of 450,000 metal/crystal
Level 3 Engineer: 90 minutes for a total cost of 300,000 metal/crystal

You can stack more engineers to increase the speed, but the cost will remain roughly the same.

Additionally, a Golem costs an absurd amount of energy.  Energy is deceptively pricy.  At 400k energy usage, if your energy comes from only Mk II Reactors, that costs you 150 metal/crystal per second, or 540,000 per hour.  Likely though you will need to use Mk III Reactors to reach 400k power, which will cost you 200 metal/crystal per second, or 720,000 per hour.  That's really expensive.

When you add it all up, if you keep a golem online for 3 hours of gameplay, you pay 100 initial AI Progress, and presuming you use Mk II engineers and Mk III energy to power it, you spend a total of 2.6 million metal/crystal keeping it running.  That's ignoring the fact that any repairs you do to the golem costs just as much as the initial repair and just as much AI progress.

For that price you'd think that the golem must be an AI stomping machine that delights in nothing but tearing huge swaths out of the AI fleet.  But that's also not really the case.  Sending any of the golems into your average Mk IV planet (500-600 ships) will probably get it killed, or at least, cost you another 100 AI progress and half million resources to repair the damage.

But what makes this monstrous investment even worse is that now that you've sacrificed it all to resurrect an AI killing machine, and what happens?  The AI increases the number of ships it sends against you, both in waves, reinforcements, and general (AI progress bonus).  I mean, it sort of makes sense, but the fact that the AI directly counteracts your progress (huge weapon of mass destruction) with way more ships kind of cheapens the whole deal.

For the incredible investment, they are of limited use in the end game because of the AI's usage of anti-golem/starship weaponry on its core/homeworld.  You can take out mass drivers but for the effort it's usually easier to target the command station.  As such, golems are best used defensively and in the mid-game. 

Problem here is that if you want defensive power, you can do way better than golems (some golems are useless on defense as well, such as cursed or artillary).  First, with the AI progress you incur from repairing a golem and the triple wave size against systems with golems, you might actually have an easier time defending without the golem.  Heavy Beam Cannons, Fortresses, or even the coveted Super Fortress will probably serve your defensive purposes better than a golem, and probably end up costing you less and expose you to less risk (except black widow, that thing is a portal camping superhero).

Of course, I respect x4000's opinion about the existing golem balance, making golems cheaper would probably require the golems to be adjusted as well.  Out of everything that is limiting about golem usage, I would point to the AI progress cost as the biggest culpret.  The game to me is entirely about managing opportunity and risk in relation to AI progress, the one number that binds the whole game experience.  Largely in the game you weigh opportunity versus increasing AI progress, and taking on more risk at the cost of keeping AI progress low.  Golems fail this second check pretty hard, while golems do increase opportunity they also present risk and AI progress.  The game design at a fundamental level encourages you to avoid both risk and AI progress.

So... that's why I don't use golems. 

For the sake of discussion, if I were the grand poobah of game balance, I would do the following:

- Golems cost the same to repair as they do now.
- Energy cost to maintain a golem is half, at 200,000 energy.
- A golem that is out of energy loses all functionality to move or attack.
- Repairing a broken golem doesn't cost AI progress.  When the golem comes under your control however, you incur 40 AI progress.
- Golems are vulnerable to attacks from bomber type ships in the same way that heavy defenses are.
- Golems cannot be repaired once operational.  The golem is an incredibly complex alien weapon, and while you can activate a disabled golem, repairing combat damage is impossible with the current technology available to you.
- In the case of the Cursed Golem, there is no AI progress penalty for use of his vampirism.
- Golems under control of the AI do not increase reinforcements to the current planet.

The AI progress cost of the golem is way down, as is energy cost.  The penalties of AI wave size and reinforcements still exist, as well as the new vulnerability to bombers (as much of a penalty for players as a sanity check for fighting golemite AI type).  The fact that you cannot repair the golem changes the golem into a limited resource which can be exausted.  You pay less harshly for them however in terms of progress.  I would even consider a lower value for progress, as even with 0 progress, use of the golem is still both expensive, limited in its usage, and presents risk for the player.

Anyways it's just an idea.  Anyone else want to share their opinions on golems and ideas?

[x4000]

Not to be overly brief, but I will skip making point by point statements here.  Basically, I do think you make a number of good points.  I think you're undershooting the AIP a big, but aside from that I could see the model you are proposing as working.  I will definitely think about this, and if more changes are made they will probably be toward that direction, honestly -- those are quite good points.  But, I'm not yet comfortable moving too quickly with changing the golems.  I already feel like I've made borderline too many changes too fast with them, as that can be game-breakingly-overpowerful if one is not careful.  But -- I do hear what you are saying.

[Doddler]

Not to be overly brief, but I will skip making point by point statements here.  Basically, I do think you make a number of good points.  I think you're undershooting the AIP a big, but aside from that I could see the model you are proposing as working.  I will definitely think about this, and if more changes are made they will probably be toward that direction, honestly -- those are quite good points.  But, I'm not yet comfortable moving too quickly with changing the golems.  I already feel like I've made borderline too many changes too fast with them, as that can be game-breakingly-overpowerful if one is not careful.  But -- I do hear what you are saying.

I don't think I've ever mentioned how awesome you are.  

I might be undershooting the AIP, but I think the theory I was kind of putting forward is that the player pays for the golem in other ways other than AIP, because people are compelled to ignore content that costs AIP.  When I tested the golems, none of them could take out a Mk IV planet on their own that had 500-600 ships, they would kill lots of the ships but the AI would throw 300 or so in reinforcements every few minutes (Putting 4 golems on a planet at once caused the AI to instantly reinforce 6000 ships).  I admit in an actual game situation where you use golems as part of a fighting force things might be a very different story.  The game changing effects of say, the Dyson Sphere seems to overshadow the other golems by a considerable amount at a very low level of risk.  Admittedly though Dyson is an optional faction that makes things easier and golems exist in all expansion games.  I guess I just want the golems to be something I want to get and work towards getting rather than intentionally ignoring.

Some of the points came from a game that I fought against the Golemite AI.  I tried to take out an Artillery golem with 1000 ships on a planet that had a negligible amount of Mk II ships defending, I thought I could easily kill it because simply put the artillery golem is bad against groups of ships, and I couldn't even really hurt it because it had no weaknesses, and I eventually I got wiped by mass reinforcements cause AI golems give reinforcement boost.

[Trezamere]

I can't say I'm a fan of the idea that you can't repair them... especially considering what you said, they can't even take out a mk4 planet (sans the cursed golem).  The problem I see with not being able to repair them is that you can't load them on a transport and they're pretty slow, 4/5 times you have to go out of your way to actually acquire a golem, so to make use of it you would then have to babysit it back to your "safe zone" of planets and then babysit as you move it where you actually want it to go (because god help you if it takes any significant damage).  Either way (the current way or this new idea) it ends up being a burden and not really that helpful.

I will agree that the main drawback for them is AI progress (who disagree's really..).  I made a thread about it too but to add some content/other stuff to the table without just shooting idea's down I'll repeat it here, thought there are likely better solutions and I'm not trying to steal anyones thunder.  One alternative solution would be to leave all of the costs as is, and instead give it a huge initial AIP cost, something you can actually quantify and weigh the pro's and con's of whether or not it's worth capturing the golem for instead of all of a sudden finding yourself with a couple thousand AIP.  You could also make it so they only appear on mk4 planets, so you really have to fight for it if you want it.  If they need to be made weaker then so be it (the bomber idea is a good one anyway).

[x4000]

Here's some more food for thought, from a discussion between Keith and myself offline:

Chris,

I've totally left the golem balance stuff
alone since it's very much something to be kept in line with your
philosophy for them.  But one thing came to mind:

For 100 AIP, you can take 5 systems.  A strategically placed planet on
most map types is connected to at least 3 other planets. so that's 15
other planets that you can get supply on and thus knowledge raid.  (5
+ 15) * 2000 = 40,000 knowledge.  Is one full-health to 1-hp use of
even a Black Widow or Armored or Artillery golem even roughly
equivalent to the ability to unlock the mk II and mk III versions of 5
base ship types (and mk II of the 6th)? Or Flag + Zenith + Spire +
Raid I + Raid II + Raid III + Leech I + Leech II + Leech III + Dread I
+ Dread II starships?

Granted, there's a lot of dynamics that make this apples and oranges
(clearing to and taking a strategic system, and then knowledge raiding
all neighbors is a non-trivial act, the time consumed = AIP, may
involve destroying some other AIP-causing structures, etc).

Anyway, in my own balancing I tend to primarily focus on knowledge
cost as the only non-renewable resource, though AIP is also
essentially non-renewable and they are in some ways convertible.  I'm
curious to know more about the nuts and bolts of how you make balance
decisions, if/when you have time.

Keith

Keith,

You make excellent points there.  In my own balance decisions, as far as costs go, I tend to focus on ship caps and knowledge costs as being the only true limiters, same as you.  But, when it comes to capturables I also focus heavily on AI Progress cost.

When it comes to smaller ships, of course, I use the results of the simulation to make sure they are balanced acceptably against their peers, but I also use some mental weighting for ships that I want to keep weaker but with specialized functions, or which the simulation is likely to undervalue.  It's very much an art, not a science, and I make further adjustments over time based on my own play experiences and those reported by other players.

When it comes to the golems in particular, I can't really use any of that stuff, aside from AI Progress.  Thinking of territory in terms of knowledge seems fairly fruitless to me, because you can always trick the system in various ways by doing knowledge raiding on systems adjacent to a highly-connected planet, or things of that nature.  I mean, I see where you are coming from, but in terms of golems I think that the data is too restricted to make for meaningful decisions.

When it comes to golems specifically, but also starships, there's really only one main factor I look at, and then I work backwards from there: how much they can destroy.  The balance for the golems was based on how I was able to use them in testing, what I was able to accomplish knowing all the ins and outs of them at the time, and knowing when to skirt a planet and when to attack it.  In other words, unlike a lot of these other players, I was not trying to use a golem for any purpose beyond just wanton destruction of as many planets as I could.  From that perspective, I was able to cut huge swathes through enemy territory with even the worst golem: cursed.  My balancing was basically geared at reducing it so that it could only take four or five low level planets -- surprisingly, perhaps, with that mechanic it's a fine line between it being perpetually powerful and it petering out overly quickly.

In all cases, I erred on the side of making the golems less useful, or too expensive in all respects, rather than making them overly useful.  I've seen it far too many times before, repeatedly and repeatedly in various alpha and beta and even official versions of this game.  Mainly with starships, which were something that took eons to get right.  There was a time when spire starships could take out two or three Mark III AI planets with just the three spires clustered.  There was then a time when the spires were practically useless, and the Zenith starships could just clobber things.  Raid starships were so useful for a while that people were fools to use anything else (or were being "good" and not using the exploit -- or not having yet found it).

So then it was on the wake of all this that I started working on golems.  Theoretically entirely possible and plausible, these were an enormous balancing challenge and broke most of the normal conventions I would use.  A "super starship" that can eat high-level planets at a time by design is just crazy to make workable.  In a normal pvp RTS that is shorter, the designers at least have the luxury of making it so that both players can do this thing, and it can take a crazy long time, and based on the resources used to build it one player can beat the crud out of the other in the meantime.  Here, based on the rules of the game and the AI, the AI simply can't respond as a normal human player would if the players get a golem in certain map positions; in others it's fine, of course.

But even in those other games, it led to starship-like scenarios.  Know what killed SupCom for me, ultimately?  Even playing at 1/3 speed because of sorians AI mod wasn't quite enough to kill it for me.  The game died for me when we discovered that if we built a flying saucer experimental, then just kept it hovering over the enemy commander, it would kill him 100% of the time simply by crashing on him and catching him in that explosion.  It didn't even have to use its weapons to do it!  Thus games that were interesting slogs, close to stalemates, or even heading for a loss, could be won if we simply managed to get that one unit built.

In AI War, given the stuff with the starships, I saw golems as a similar sort of risk.  Supposing that the players find and build all three golems -- granted, a single player might have a hard time doing that, but it's no problem for a gang of four -- and then just sent that against the AI home planets?  The mass drivers make it so that initial raiding parties have to be sent in to destroy those, at least, but the extreme risk is still there.  If "long slog" planets can be easily broken by a single golem, then that breaks the entire flow of the game, and practically eliminates the endgame, allowing for short-circuit victories.  Knowledge cost, or any other costs, haven't even entered my head yet at this stage. It's all about what their potential damage path is, and how to make that appropriately costly in the edge cases that are so exploitative.  And we'll work backwards from there to the more normal cases, of course, was my thinking.

Okay, so if we have this huge weapon that can do all that, then the AI needs to be able to react, and quickly.  The huge AI costs are one thing, and are important -- that way you have to really mean it to use one of these golems, and if you use it to quickly kill one AI, the other AI is likely to be so entrenched as to not be assailable with them, thus restoring the game to balance.  But that still doesn't solve cases where players swarm both AI home planets at once -- there are more golems than home planets, after all.  So in order to handle that, I needed to make it so that at least the AI home planets would be able to respond more quickly.

But really, that same logic holds true for all high level planets, doesn't it?  Another exploit is if the players locate a golem early, and then immediately rush for it and build it.  Then all of the AI planets are weaker, not having had time to reinforce.  And now the player has this weapon of doom far beyond what they should be able to muster at that point in the game.  Thus the player then rampages out and sunders all the mark IV and/or mark III planets that they need for ARS/factory/etc purposes with a minimum of fuss, and then goes back later and actually collects those prizes.  Even if the golem dies, they have still managed to exert enormous benefit by rushing to the golem and using it to catch the AI off guard.

I combated this in two ways: firstly, by making the AI send larger reinforcements, and send reinforcements immediately, when the golem arrives.  Secondly, I also made the golems really expensive in resources (thus slow to build), and really expensive in energy (thus impossible to build until the game has progressed to a point where the player has a certain number of planets and thus the AI has a certain amount of reinforcement points and has had a good amount of time to actually do reinforcing.  This limits the amount of "attacks of opportunity" that can be made much as SupCom or other games makes it so that their big guns can only be built in the last segment of the game.

Thus with this design, it covers all the bases of the fatal exploits that I have seen and have used myself, and that I can think of.  With that, the golems were still able to do a lot of damage in my tests, but nothing that was game breaking.  Did it make the golems to weak?  Yes.  By far, yes, as it turned out.  But given that I was never really that excited about using them in real games myself anyway, I didn't find this out first hand too well.  And most of the scattered reports from a few players who were having trouble with them were clearly misusing them, or using them in nonoptimal ways.  And there weren't that many reports, not compared to how many people bought the expansion.  I think that was due to the sheer volume of stuff in the expansion making it so that most players had plenty to do without golems, and since they -- like me -- are averse to anything that high risk, they tended not to find out how underpowered they were, either.  At any rate, I think that's why golems were able to persist in such an underpowered state for so long.

And, it's all these exploits that make me so wary of making too many changes at once.  You know what?  This is a really good conversation, and filled with info that I'm sure would be useful to the other players who are suggesting various rebalancements and complaining about even the buffed golems.  I understand their points -- and yours -- but the opposition of your points just breaks the golems.  Opposition of mine is more likely to break the entire game.  I'm still coming around to more that direction, but it's going to take a lot of testing and gradual shifts, and some outside-the-box thinking on all parts, to really get an ideal solution.  Many of the proposed changes involve taking out the safeguards that keep the golems from being game-breaking.  But, those same safeguards are what make it so darn hard to make golems worthwhile in the first place.

It's a really challenging thing to balance, and honestly I think it's beyond the power of any one designer to do; I've never seen it successfully done in another game, and many players seem to agree with me on that.  But, I believe that it can be done, and with the accumulated brainpower of all the grognards in a community like this, I think that the crucial bits can be figured out.  I think the recent changes were a step in the right direction.  But, I also feel like more testing time needs to be given to these, just to see if players are able to come up with anything exploitative, and to also give people the time to come up with increasingly inventive solutions (as happens).  And I'm still hoping that something that magically fixes it all will occur to me, as well, which happens from time to time.

In short, I'm not in a rush.  I'm not happy with the balance of the golems at present, but I am happy that they aren't destroying the game.  And I'm also not willing to take them out even if an ultimate good solution is never found, because -- hey, they exist, and they are fun in some situations, and I am still quite sure that if used in really creative / borderline-abusive ways, they are well worth the cost.  The problem is, players have been playing really clean with these, for some reason, which is curious.  Usually with a ship like parasites or something, they figure out all sorts of exploits.  I guess the extreme cost of the golems is a deterrent to to much experimentation, but that also means that by the time we get them "in balance" in normal circumstances, then a rash of exploits will come along.  I'm proceeding slowly in hopes of making it so that we can hit a true point of balance without having that series of exploits, which I fear will end the interest in the game for a number of innocent players (many of whom no doubt are not active on the forums) if it happens.  Starships that can too easily take out a Mark III or even IV planet were one thing, but golems are a magnitude worse.

Anyway, that is my thinking.  This was a really useful discussion, so I'm going to log this to the forums, too -- your notes and my response.

Chris

[x4000]

That last post of mine was way over-long, but hopefully it will provide some good areas for thinking amongst the community, and will maybe even provide some incentive for people to try abusing the golems to see what they can really do with them.  I think that the only way to really use golems, in the end, will be in those sorts of extreme manners, because otherwise the "normal" manner will be really powerful, and the "extreme" version will be game-breaking.  In other words, my best hope is for the normal manner to just be okay, or even not that great, but the extreme version to be really powerful but not to the point of breaking the game.

That would make golems an experts-only unit, not a beginner-friendly unit as is the case in most other RTS games, but I'm okay with that.  Logically, at this juncture, I can't think of another way they could function.  As I noted above, there is not a good model to look at for this in other games, as the superweapons tend to be game breaking for expert play, or lame except in beginner play.  So that's the steep challenge we're working with here.

And now... unfortunately, I need to step away from this conversation for a bit, as interesting as it is.  I really have a ton of stuff to get done for next week's beta of Tidalis, and I've spent a lot of today on the forums because a lot of really interesting posts have come up, and it's fun to get sucked into that.  And there are conversations I want to have, and problems I want to solve.  But what I have learned is that I have to guard my time a bit when it comes to AI War, because this game will just suck up all my time if I let it, and I'll never get anything else done.

So I won't comment overmuch or at all for the next week or two on this topic, most likely, but I am interested in seeing where the discussion goes, and hopefully this is something that we can resolve in time.  Also, I'm determined not to go making any more rash decisions on golems at the moment, as I'm still wanting to see if people can successfully exploit -- or at least make competent use of -- them in their current form.  That will also heavily influence what further changes we ultimately make to them next month or whenever makes sense (after the next official release, which will hopefully be in a week or two, anyway).

Thanks for working with me on this!

[Dmdunn]

this game will just suck up all my time if I let it, and I'll never get anything else done.

What exactly is the problem here?

 

[x4000]

[RCIX]

this game will just suck up all my time if I let it, and I'll never get anything else done.

What exactly is the problem here?

 

That he wouldn't be able to put out the Tidalis beta of course!

[Blahness]

This topic makes me want to generate golems on a medium difficulty and try them out for exploits.

Brb playing with cursed golems

[allmybase]

I think one way to look at golems is compare what you get out of a nuke.

I'm not sure golems are good if they can't solo a 500+ shipcount mk4 planet, think of it this way:

1) You can always build nukes, whereas you have to find the golem amid the myriad of planets
2) Nukes will completely jilt a MKIV planet that has thousands of ships at the cost of significantly less than 2.6 million resources and less AI progress

I guess golems are reuseable and they do spare the planet's ability to produce resources/supply, but it seems to me they do a lot less damage, a lot less reliably, require support, and cost a lot more O_o Is it just some sort of dumb luck unit that if you happen to find one early and you happen to have a number of conditions like really defensible position and overflowing resources and it happens to be the right golem that fits in your strategy then you can maybe use it? (and even with this number of events coinciding its still what over 100 AI progress that you might otherwise not have had if you were able to beat the game before without golems?)

Totally just my opinion but I think golems should be a thing you go 100% after like adv research and that first factory. It's the expansion, you should get used to using to using the new cool unit, not that I want it overpowered. Golems should be interesting in that they have some drawback whereas adv research/factory has no drawbacks (other than possibly being located in a bad spot), but it should be a pretty clearcut oh hellz yes unit like most capturables.

It doesn't need to be better than the nuke, but it should be a fair equivalent in some sort of niche, not some horribly overpriced thing that pees its pants once it sees a mk4 planet with 600+ ship count.

[Draxis]

Another problem is single player and multiplayer play very differently.  Where single player is all about slowly and carefully gathering enough resources and knowledge to find and destroy the AI, multiplayer...isnt - Its more like Genghis Khan in space, with Custer as your personal adviser.  The last 2 player game I finished had 5 worlds destroyed by nukes, and 1 destroyed by the mining golem, and that was only a 40 planet map.

[x4000]

I think that is not a factor of multiplayer, but of who is playing. I play 70% multiplayer, and I plays out just like solo, for the most part.

[Draxis]

We find there is very little need to be cautious in multiplayer.  The AI doesnt seem to defending its worlds more, and with combined fleets, taking a single world is never difficult.  In single player, the AI homeworlds can reinforce so fast that the human forces simply cant get in, we have not had this problem with even just 2 humans.

[superking]

I think giving bombers the hefty anti-defence bonus vs golems would work well as a counterpoint to some buffs to their combat performance