Arcen Games
General Category => AI War Classic => AI War Classic - Strategy Discussion => Topic started by: Nibelung44 on November 04, 2009, 03:23:43 am
-
Hi,
I would like to know if you can do an attrition war to the AI. i.e kill ships somewhere, so that it has to compensate/reinforce the losses, and these reinforcements are not used elsewhere?
The crux of the problem is to know if the reinforcements 'formula' for the AI are only a function of time and AI level, or if auto-compensating mechanisms exist. If say the AI, on average, get 1000 ships every 30 mn when at AI level 100, or if the sum of the reinforcements check also how many are 'missing' and get adjusted dynamically.
An example: I have my southern border, where I do not intent to progress, well guarded by defenses. South of that, it is AI land, and there is 1750 ships in the nearby system. Should I make a raid to trash as many ships as possible, knowing they will have to be compensated (= sucking up the reinforcements of the AI to this place and not elsewhere) or does it has no impact overall?
I think you get the point...
Another question related to this one. When the AI get reinforcements, how is the mix of Mark levels? If AI Mark Level is II, will it receives only Mark II ships, or will it get Mark II ships at least, but if the system is a Mark III+, these will be replaced by their Mark III (or better) equivalent? This would be an auto compensating mechanism somehow.
Thanks in advance, this can develop a new branch of strategy, for me, knowing a bit how reinforcements work in details (I know of the FAQ but I'm still missing some elements...)
-
I can answer your reinforcement mechanism question i think: It's something like max(planet_level, ai_level).
-
In theory it is possible to attrit the AI by killing ships at places where it is likely to reinforce. The AI has a certain priority (home planets, planets that are alerted) and will distribute ships across these priorities, though it will also (but less frequently) reinforce other planets. Once a planet fills up it will stop adding ships there, so attacking filled up planets would lead to less reinforcements elsewhere.
I have my doubt whether attrition will work as strategy: the time and resources that you spent on killing ships is time and resources that are not spent on a more direct approach. And that will give the AI more time to reinforce.
What may work (and what you should be doing anyway when research raiding) is killing AI planets deep inside AI territory with many neighbours. All those neighbours will start sucking up reinforcements from other planets.
-
leave to me if the strategy is viable or not ;D no seriously, thanks for the feedback, but I really must be sure that a ship lost by the AI is not "dynamically compensated" for this strategy to work.
-
leave to me if the strategy is viable or not ;D no seriously, thanks for the feedback, but I really must be sure that a ship lost by the AI is not "dynamically compensated" for this strategy to work.
Haagenti is correct. The AI has a rate of reinforcement (slightly randomized intervals), and chooses where to distribute those ships, with various planets having different levels of priority based on a few factors. The AI will not automatically gain extra reinforcements due to lost ships, so your strategy is theoretically viable. Whether it works well in practice I have no idea.
-
leave to me if the strategy is viable or not ;D no seriously, thanks for the feedback, but I really must be sure that a ship lost by the AI is not "dynamically compensated" for this strategy to work.
Let me know if it works. I try to avoid filled up planets (as these are the only ones that are useful to attrit), since I usually end up the one who gets attritted. But your mileage may vary.
-
The strategy is theoretically viable, in the sense that the AI has no compensation for lost ships (AI Progress excepted, in cases of lost planets, etc, since that eventually causes larger ship volumes in reinforcements and waves). There is no rubber-banding to the AI (ala Mario Kart).
However, I would somewhat question the premise of a war of attrition against a foe that has 100x your strength, and which tends to build ships faster in response to you, anyway. Generally speaking, even if you have a 200% or better Kill-To-Loss ratio in all of your engagements against the enemy, it will still be gaining ships faster than you depending on the AI Progress level, the difficulty level of the AI, the size of the map, etc.
The difference, of course, is that it is gaining ships across a wider space, and it is severely constrained in how it can use those ships. But because it can focus reinforcements in specific areas, you can cause some "misdirection" of its reinforcements -- and by the same token, it can relatively quickly react to focused threats that you throw against it. So in a sense of whittling down the actual number of units of the AI, I think you'd be fighting a losing battle in pretty much all cases on a galactic scale.
That said, a war of attrition is actually an extremely popular tactic on individual planets. Force fields and other big defenses (superfortresses, etc) cannot be repaired by the AI, but have a very slow regen, for exactly this sort of purpose -- so that you can hit them repeatedly with forces and wear them down over time, if you want/have to. So while I do somewhat question the premise of attrition on a galactic scale because of how the AI functions, when you simply look at a portion of the galaxy, or even an individual planet, I think that tactic is highly effective and actually used by a lot of players (myself included).
Then again, depending on what you feel qualifies as attriting the AI, one could also make the argument that the entire game is founded upon attriting the AI -- not the individual fleet ships, which are simply obstacles to bypass or kill, but rather the larger ships -- command stations, warp gates, ARSes and factories, data centers, guard posts, and so forth. The AI has a fixed supply of those at the start of the game, and they never, ever, get more. Unlike you, they can't rebuild their ships of those sorts that are lost. So as you are expanding into the galaxy, you are putting a permanent crimp into where and how the AI can reinforce itself or how it can attack you.
Another way of looking at it is this: the AI can out-produce you in terms of military ships, and if you try to stem the tide of that galactically, I don't think that's likely to work very well. But on a planet or regional basis, that can work very well, mainly by striking at the reinforcement-affecting ships (command stations, guard posts, warp gates, and special forces guard posts). This sometimes leads to a strategy that players here call "neutering" a planet, which is a form of galactic attrition in my opinion.
Now, all of that said -- is there room for you to do something surprising and original that nobody else has yet done? I think yes. :)
-
Thanks for the very detailed answer!
I'll make some tests then... The goal is to have a kill ratio of at least 10:1 in such type of battles. Ideally you lose nothing and the AI lose several hundreds of ships. By targetting enemy systems which are still under your supply, I think you can do such things, potentially.
-
Thanks for the very detailed answer!
I'll make some tests then... The goal is to have a kill ratio of at least 10:1 in such type of battles. Ideally you lose nothing and the AI lose several hundreds of ships. By targetting enemy systems which are still under your supply, I think you can do such things, potentially.
No problem!
And yes, I think if you can make a very good KTL ratio, and also take out guard posts (neutering the planets) as you go, you could have good results and do something very different from what has been tried before -- in terms of scale, at least, and the effect that might have on the overall game flow for you. I look forward to seeing what you find out!
-
Actually at least one form of attrition is a extremely valid strategy. Stick a bunch of turrets(40 or so) on an enemy planet where they can just kill spawns on a wormhole or two(best with some lasers so you can outdistance any turrets that spawn), and every time the ai respawns troops on that planet your turrets will kill the respawns on that wormhole without tieing down any of your mobile troops and with basically no losses if its a tier I or II planet.
-
If you can apply a kill ratio of 10:1, why not apply this kill ratio to a string of planets leading to the enemy home planets? Why apply it to a filled-up backwater planet?
Note: I'm not critical. I'm really just curious on what you are trying to achieve. If I really don't understand something, it's usually because I'm totally missing the point somewhere
-
Actually at least one form of attrition is a extremely valid strategy. Stick a bunch of turrets(40 or so) on an enemy planet where they can just kill spawns on a wormhole or two(best with some lasers so you can outdistance any turrets that spawn), and every time the ai respawns troops on that planet your turrets will kill the respawns on that wormhole without tieing down any of your mobile troops and with basically no losses if its a tier I or II planet.
Yeah, that's true, I still have yet to try that, but I know you've been using it to great effect. That sort of "attrition trap" is pretty interesting. Of course, with some of the new area-damage ships, and the really-long-distance ships, if the AI has them, that strategy might become less effective in the expansion in some maps. But yeah, that's a cool one.
-
If you can apply a kill ratio of 10:1, why not apply this kill ratio to a string of planets leading to the enemy home planets? Why apply it to a filled-up backwater planet?
Note: I'm not critical. I'm really just curious on what you are trying to achieve. If I really don't understand something, it's usually because I'm totally missing the point somewhere
Possibly he's a completionist, but I'm curious as to the motivation as well. Others just like to experiment and do something cool that no one else has ever done before, and achieve victory through nontraditional means. Or maybe it just fits with his playstyle for some other reason -- those have been my main assumptions so far. I think that, for a certain set of strategy players (or gamers in general, perhaps), most-efficient victory is not the main goal. Look at Kalzarius with his conquer-all-100-planets-on-the-map 40-hour game (which I think is pretty cool).
-
After I understood a bit more of the AI mechanics, I had some luck "clearing out" sections.
I had a string of two systems that I'd conquered (I think to get an advanced factory). At first, I just wanted to protect the advanced factory but I was close to jumping to MKII on the AI progress so I didn't want to destroy anything. I formed a fleet of T-Raiders and Etherjets and hopped from system to system clearing out everything in them... This created a sort of "weak" area around that system. Basically, I just found myself hopping into the system with all of my MKII versions of those ships and setting them to RFD mode. Every 15 minutes or so I would come collect them and move them on to a new system.
There were about 6 or 7 systems that I could hop through in a cycle before starting over.
Now, whether or not that was a net win for me is another matter but it kept my factory safe and it was relatively little micro managing.
If you do something like this, don't forget to knowledge raid while you do it ;)
Also, T-Raiders are TERRIBLE at taking down turrets so it may be just that I had an ideal mix of units to do this with.
-
Well, I'm far far from being able to close toward the Core AI planets, as I am on a 80 planets map and owning only 9 systems. But I see concentrations forming around me, and I wanted to try sucking up reinforcements of the AI on one side while I was active on the other.
Anyway, I'm content to know that there is no rubber banding here... all is fair and square :)
-
That said, a war of attrition is actually an extremely popular tactic on individual planets. Force fields and other big defenses (superfortresses, etc) cannot be repaired by the AI, but have a very slow regen, for exactly this sort of purpose -- so that you can hit them repeatedly with forces and wear them down over time, if you want/have to. So while I do somewhat question the premise of attrition on a galactic scale because of how the AI functions, when you simply look at a portion of the galaxy, or even an individual planet, I think that tactic is highly effective and actually used by a lot of players (myself included).
Who needs attrition when you've got 6 raid starships ready and willing to crush anything on a planet? :D
-
If you can't kill something with overwhelming power and kill it relatively quickly, you shouldn't be attacking it. A scenario where you attack a superfortress and your entire attack force gets killed, but you damage it to 50%, and then you send in a 2nd force to finish the job is a very inefficient use of resources.
You should attack with more troops and kill it the first time. If you can't do this (lack of force cap), you have probably already lost the game, as while you are clearing this chokepoint, the AI is preparing a new and even bigger one somewhere else.
-
If you can't kill something with overwhelming power and kill it relatively quickly, you shouldn't be attacking it. A scenario where you attack a superfortress and your entire attack force gets killed, but you damage it to 50%, and then you send in a 2nd force to finish the job is a very inefficient use of resources.
You should attack with more troops and kill it the first time. If you can't do this (lack of force cap), you have probably already lost the game, as while you are clearing this chokepoint, the AI is preparing a new and even bigger one somewhere else.
If find this not to be true at difficulty 7ish and down. Perhaps at the difficulty 8 and up, but at difficulty 7 it is quite reasonable (and often necessary because of ship caps, the position of your fleet for defense, etc), to take down a superfortress or similar in multiple waves, rebuilding in between.
This also comes back a bit to larger strategic choices in general -- I tend to have Mark II Command Stations and focus on high-yield resource planets, etc, so I tend to be quite rich in resources, with ship cap (usually) being the greater limiter if I play too conservatively. So in those cases, there is an impetus to get things done quickly while still maintaining defenses around my wide perimeter of resource-valuable planets. I tend not to be holed up in a sector of the galaxy, but rather strung out as needed to get what I want, so massing my entire fleet at one point is almost never possible. Meaning that I have no choice but to assault a Mark IV planet with a superfortress with just a third or less of my forces, for instance.
I'm not saying you are wrong -- in your style of play, and at the difficulty level you play at, I am sure what you are saying is spot on. I'd just caution against blanket advice of that sort, since the playstyle of an individual player (and difficulty level, to another extent) can really have a huge impact on the strategic imperatives in general. In my games, perimeter defense is challenging and and ever-present threat, but resource efficiency is vastly less of a concern than it is for you. So the smartest choices in how to assault an entrenched target necessarily vary, then.
When I'm at cap with a ton of resources pouring in, and I can't move big parts of my fleet for risk of losing valuable territory, it only makes sense to hit a superfortress with a third of my fleet, quickly rebuild that, hit it again, and repeat as needed. In a temporal sense, it's extending my ship cap based on my greater resource income, which I require a greater (temporal) ship cap in order to defend.
Anyway, so I think that's where the differences lie, is in fundamentally different playstyles (which I think is too cool, incidentally). :)
-
Have you tried using turrets supported by mercenaries for defense? Mercenaries are an ideal sink for resources (due to mercs I never have too many resources and I'm never at cap), and when used judiciously on defense in support of turrets have low losses (not to mention that they can be set to low power when there is no threat).
This frees up the cheap mobile elements, leads to less losses in those mobile elements as you outnumber the enemy, leading to more resources, leading to more mercs etc.etc.
But, you are right, your mileage may vary.
-
Have you tried using turrets supported by mercenaries for defense? Mercenaries are an ideal sink for resources (due to mercs I never have too many resources and I'm never at cap), and when used judiciously on defense in support of turrets have low losses (not to mention that they can be set to low power when there is no threat).
This frees up the cheap mobile elements, leads to less losses in those mobile elements as you outnumber the enemy, leading to more resources, leading to more mercs etc.etc.
But, you are right, your mileage may vary.
To be honest, I never use mercenaries. I feel it is much more resource-efficient over time to just build to your caps, then expend part of those ships in whatever the offensive of the moment is, and then quickly rebuild to your caps and launch a new offensive (or continue the old). This also keeps me from ever getting remotely near my resource caps, and the cost-to-benefit ratio is quite good compared to many alternatives.
In the grand scheme, this means I probably lost more ships (my KTL ratio is generally 1:1 to 3:1 in my favor, whereas I've seen you at 5:1 or more), but it gives me a good deal of flexibility. Mainly, this was the strategy I was already using before mercs were ever added, and I didn't see a compelling reason to switch to mercs when I added them (they were mostly added to support other playstyles than my own).
All that said, I see your points, and that certainly leads to efficient use of your ships (not throwing ships away ever, if you can help it). I view the ships as robotic avatars, by the way, not human lives, so I'm not needlessly killing my men. ;)
I like the sound of your strategy, though -- I probably won't use it much myself, although I might dabble at some point, but it's cool to see the mercs put to such good use as that.
-
All that said, I see your points, and that certainly leads to efficient use of your ships (not throwing ships away ever, if you can help it). I view the ships as robotic avatars, by the way, not human lives, so I'm not needlessly killing my men. ;)
A good deal of my real-life revolves around balance sheets and income statements, so I tend to see ships as money. And therefore translate my strategies in terms of return-on-investment, debt, spending, growth-rate.
So the very concept of attrition is heresy to me. Akin to borrowing money on your credit card against 30% while having plenty of savings in a money market fund at 1%
Heresy, I tell you :)
-
Attrition Strategy is attriting the enemy, not your fleet ;D i.e slowly inflicting losses to him, without suffering much if any on your side. This is the idea here, although 'attrition on the Russian Front' is indeed a mutual attrition where the one with more resources (Russia) win in the end... ;)
-
/begin{pedantic}
This is IMO a non-standard use of the term attrition, as attrition usually implies that both sides are suffering casualties and the one with the most resources wins. See e.g. http://en.wikipedia.org/wiki/Attrition_warfare
/end{pedantic}
Slowly inflicting casualties with low casualties return is in general a worthy goal: however, recent experiences in Afghanistan and Vietnam show that if the enemy is able to accept high casualties while you are unable to accept low casualties, you will still lose.
-
I don't mind your pedantic style :P
Well, it is safe to assume that what we want, with this strategy, is to drain enemy resources slowly without suffering much losses. That why I called that attrition (the slow process). Perhaps 'Erosion strategy' would fit better indeed.
I have yet to try what I envision. There is a nice mark II system with 1800 ships at my southern border. Operation 'Black Hole' will soon be mounted against the forces there, and I'll keep a tally of what I lose in the end (let's add a muhahaha to that) ;D
-
Treating the ships like finances is very interesting, and I can see why you would play that way if you look at them from that angle. I think that's very valid!
My way of thinking about this is more like... hmm, I guess like of like muscular endurance or gas in a tank, or something like that. With a resource flow system, I tend to treat the "rate of flow" as my actual income, and my goal is to maximize inflow while also maximizing outflow to useful purposes. Accumulating wealth (in terms of a mass of ships or literal metal/crystal reserves) is less of a concern for me. But, that's all related into my other larger strategies that I prefer, of course, so that sort of "endurance race" would not work well given various other goals of differing strategies.
I just think it's cool that the mechanics are robust enough that people can apply such wildly different models and strategies to it, honestly. ;)
-
Treating the ships like finances is very interesting, and I can see why you would play that way if you look at them from that angle. I think that's very valid!
My strategies have completely changed over the past few months due to this viewpoint. I almost completely ignore micromanagement, tactics, unit match-ups etc. (apart in the first hour or two) and concentrate only on the economic/strategic side. The theory being that if you outnumber the enemy enough, it does not matter what ships he has or how he deploys them or what he does in combat: he will get killed with few losses on my side.
"Getting there the firstest with the mostest, kill 'em all and let God sort them out" is the general idea.
My way of thinking about this is more like... hmm, I guess like of like muscular endurance or gas in a tank, or something like that. With a resource flow system, I tend to treat the "rate of flow" as my actual income, and my goal is to maximize inflow while also maximizing outflow to useful purposes. Accumulating wealth (in terms of a mass of ships or literal metal/crystal reserves) is less of a concern for me. But, that's all related into my other larger strategies that I prefer, of course, so that sort of "endurance race" would not work well given various other goals of differing strategies.
Undoubtably it works well for you, as your high-level strategy also differs from mine.
From my viewpoint it seems almost like our (Dutch) government: "If I take the most taxes (inflow) I can, and spend (outflow) all of it at the end of the year, it must be good, regardless what I have spent it on"
I just think it's cool that the mechanics are robust enough that people can apply such wildly different models and strategies to it, honestly. ;)
It's cool indeed. Must be one of the effects of the high configurability of the game. And that there is no "campaign", so you are in the end never forced to fight against such impossible odds that there is only one efficient strategy left.
-
From my viewpoint it seems almost like our (Dutch) government: "If I take the most taxes (inflow) I can, and spend (outflow) all of it at the end of the year, it must be good, regardless what I have spent it on"
Yeah, I see what you mean. I think it's mainly a matter of kind of trying to pace myself over a long marathon run, where I'm running as fast as I can (producing ships and fasting) to win the race, but I'm staying below a certain threshold of energy exertion so that I don't have to periodically stop or burn out before the end. If resources were finite in AI War, my strategy would be completely invalid, and I think that's the thing the Dutch government would be overlooking (not that I know anything about the Dutch government). But taxes are more of a finite resource, anyway.
-
Slowly inflicting casualties with low casualties return is in general a worthy goal: however, recent experiences in Afghanistan and Vietnam show that if the enemy is able to accept high casualties while you are unable to accept low casualties, you will still lose.
I think this will be a difficulty for Nibelung's attrition/erosion strategy. Not only do you need a good kill ratio, but you need to destroy the AI's ships faster than it creates new ones globally. Otherwise, you're making no net gain.
On the other hand, I kind of do this while I'm research raiding. In the process of protecting my science ships, my fleet roams the planet destroying guard posts, train stations, special forces posts, and plenty of ships of course. My hope is that that this will delay an eventual cross-planet attack, or weaken its strength. And like x4000, I think that if my ship caps and resource reserves are nearly full, that means I'm not using my fleet enough.
-
You don't actually need to kill them faster than it builds them elsewhere, you just need to be able to kill them while doing something else(even if its just building up your own army for a strike somewhere), and have better than proportional losses, but yeah, non-trivial.
If your ship caps and resources are mainly full, it may mean that you have too many planets(but I tend to play the other extreme :)).
-
And or/you need to attrit the AI in the places you want to go, while causing him to reinforce uselessly in places you have no interest in. Misdirection is definitely worthwhile!
-
You don't actually need to kill them faster than it builds them elsewhere, you just need to be able to kill them while doing something else(even if its just building up your own army for a strike somewhere), and have better than proportional losses, but yeah, non-trivial.
And or/you need to attrit the AI in the places you want to go, while causing him to reinforce uselessly in places you have no interest in. Misdirection is definitely worthwhile!
Yeah, those are good ideas, but Nibelung's attrition seems to suggest doing the opposite: seeking out concentrations of enemy ships even if they're not somewhere you would otherwise want to go, or even if you're not doing anything else useful in the meantime.
Well, it is safe to assume that what we want, with this strategy, is to drain enemy resources slowly without suffering much losses.
That sounds like a recipe for attacking the enemy where he's strongest and ignoring the strategic importance of targets, which is contrary to most of the usual good strategies (like surgical strikes). Maybe I'm misinterpreting.
It makes sense to me to use some focused attrition tactics while playing normally, but based on how the AI gets reinforcements spread across many planets, I think that trying to win by attrition across the whole galaxy is going to get you nowhere. In a typical game, does the AI have more ships on the board at the beginning, or when you win?
-
I tend to cause truly horrific kill-ratios in defense: massed turrets (and I mean *massed*), do a real good number on the AI. So I tend not to care about cross-planet raids. Playing a lot against Vicious Raider makes cross-planet raids quite wimpy anyway :)
-
I tend to cause truly horrific kill-ratios in defense: massed turrets (and I mean *massed*), do a real good number on the AI. So I tend not to care about cross-planet raids. Playing a lot against Vicious Raider makes cross-planet raids quite wimpy anyway :)
I think often of your post regarding having an engineer on standby with a paintbrush in case one of your turrets gets scratched. If that engineer would get bored, then there are enough turrets nearby - otherwise, I build more.