(Thought Exercise) Reduce knowledge to reduce AIP scaling.

[TechSY730]

In general, I am happy with the relative scaling that happens, it just happens too quick.

If you could magically stretch the power curves so that what is currently AIP 300 happened at AIP 500 but the relative power curves stayed the same to get there that would be my ideal solution to this issue.

Yes you can do this. For the linear forms, you reduce the slope but leave the intercept untouched on the linear formula (that's not a linear relation, the math guys sort of painted themselves into a corner when they made this nomenclature ).
For the exponential forms, the most straightforward way would be to reduce the coefficient of the AIP term in the exponent. You could also reduce the base of the exponent, which while is a bit more "sensitive" gives similar results over the kinds of AIPs we are looking at.

Actually, I think that regardless of whatever happens the AIP scaling needs to go back to linear for all difficulties. We are talking about mid and high AIP games being not viable and here is a mechanic that deliberately makes mid and high AIP harder because the higher AIP goes, the more ships 1 point of AIP gives the AI for waves. This will be a huge nerf to diffs 9 and 10 but if we are going to make both low and high strats viable, we can't AIP scaling different for the two of them.

D.

Hooray, someone else finally, sort of agrees with me.

I would contend that going all the way back down to linear may not be a great idea, but maybe polynomial would be a good compromise. (Yes this again, but to my credit, I have not harped on this on this particular thread yet )

[Diazo]

In general, I am happy with the relative scaling that happens, it just happens too quick.

If you could magically stretch the power curves so that what is currently AIP 300 happened at AIP 500 but the relative power curves stayed the same to get there that would be my ideal solution to this issue.

Yes you can do this. For the linear forms, you reduce the slope but leave the intercept untouched on the linear formula (that's not a linear relation, the math guys sort of painted themselves into a corner when they made this nomenclature ).
For the exponential forms, the most straightforward way would be to reduce the coefficient of the AIP term in the exponent. You could also reduce the base of the exponent, which while is a bit more "sensitive" gives similar results over the kinds of AIPs we are looking at.

Erm, I was speaking in game terms. On a pure numbers basis it is easy, but the AI's (and the players) "power curve" is a combination of so many factors that you can't just slash AIP (or knoweldge on the player side) and call that 'sliding the power curve'.

Is it doable? Yes. Is it optimal? Opinions will vary.

I still want to see something done along these lines to bring AIP at mid to high levels back in line.

D.

[Toranth]

Just for reference, here's the Wiki page that talks about wave sizes:  Why do Enemy Waves get so Large?
The important equation from there is that base wave size is:

((((AIProgress + modifier) * 0.8 ) ^ 1.1 ) * AIDifficulty ) / ( 13 - AI Difficulty )

Note:  This is so close to linear that the R^2 of a linear regression is over 0.998.  The only noticable difference is very low AIP (<30-ish).


For reinforcements, it's more complicated:  The largest factor is that the AI gets x reinforcement pulses, where x = # non-AI planets / 2.
Strength of each pulse is determined by:

reinforcementStrength = 1.5 * AIDiff * handicapMultiplier
reinforcementStrength += 4 + random(0,AIDifficulty)
reinforcementStrength += ((AIP/10) * (AIDiff/10))
reinforcementStrength *= (tech level multiplier)
reinforcementStrength *= 0.14
reinforcementStrength must be at most 35;
reinforcementStrength *= (1 + guardPostReinforcementPulsesShiftedToCentralPulse)

This means that as long as AIP is less than 10 x AIDiff, the AI's Difficulty will be the largest factor in reinforcement strength.


Special Forces uses another method to determine strength:

baseSizeFactor = 30   (Not sure where this number comes from)
specialDifficultyFactor = Game.Instance.Options.TotalSpecialDifficultyModifier (based on difficulty, homeworld count, handicap)
effectiveAIP = AIP
aiTypeMultiplier = (1 + 2 per Special Forces Captain AI)
specialForcesPostsInNonAITerritoryMultiplier = Mat.One + ( FInt.FromParts( 0, 050 ) * numberOfSpecialForcesPostsInNonAITerritory )
specialForcesStrengthCap = baseSizeFactor * difficultyFactor * effectiveAIP * aiTypeMultiplier * specialForcesPostsInNonAITerritoryMultiplier

Here, AIP is a straight up multiplier:  Double it, the SF cap doubles.


Strategic Reserve is similar to SF:

*Computing MaxStrategicReserve
baseStrengthFactor = 19.8   (Not sure where this number comes from)
specialDifficultyFactor = Game.Instance.Options.TotalSpecialDifficultyModifier (based on difficulty, homeworld count, handicap)
effectiveAIP = AIP
maxStrategicReserve = baseStrengthFactor * difficultyFactor * effectiveAIP
Here as well, AIP is a straight up multiplier:  Double it to double Reserve strength.

[Diazo]

Just for reference, here's the Wiki page that talks about wave sizes:  Why do Enemy Waves get so Large?
The important equation from there is that base wave size is:

((((AIProgress + modifier) * 0.8 ) ^ 1.1 ) * AIDifficulty ) / ( 13 - AI Difficulty )

Note:  This is so close to linear that the R^2 of a linear regression is over 0.998.  The only noticable difference is very low AIP (<30-ish).

For clarity's sake, this is only on 8.0 or higher.

On any difficulty lower then 8.0, the waves base strength calc is:

For Difficulties less than 8: ( (AIProgress + modifier) * AIDifficulty ) / ( 13 - AIDifficulty )

The other calculations stay the same across all difficulties as far as I am aware.

D.

edit: From my post at the end of page 1:

Actually, I think that regardless of whatever happens the AIP scaling needs to go back to linear for all difficulties. We are talking about mid and high AIP games being not viable and here is a mechanic that deliberately makes mid and high AIP harder because the higher AIP goes, the more ships 1 point of AIP gives the AI for waves. This will be a huge nerf to diffs 9 and 10 but if we are going to make both low and high strats viable, we can't AIP scaling different for the two of them.

I am talking about making all difficulties use (AIProgress + modifier) * AIDifficulty ) / ( 13 - AIDifficulty ), rather then have 8.0 or higher use ((AIProgress + modifier) * 0.8 ) ^ 1.1 ) * AIDifficulty ) / ( 13 - AI Difficulty ).

This would be a nerf to 8.0 or higher so they would have to get buffed, probably in the difficulty multiplier, to compesate.

[keith.lamothe]

I think we may be looking at the whole "low-AIP playstyle vs not-low-AIP playstyle" thing the wrong way.  Backing up a bit:

1) For what reason would a player draw more AI attention to themselves than necessary?  In other words, why would they not play low-AIP whenever possible?  There are answers to this, but the main two that come to mind are "because they're more interested in having fun than playing optimally" and "because they're playing Fallen Spire".

2) Should we then be concerned about balancing for both a low-AIP and a not-low-AIP situation during the early/mid game?  Presumably either the low-AIP is going to have an excessively easy time, or the not-low-AIP is going to have an excessively hard time.  Is it, ultimately, a problem to say that you're going to need to keep AIP low during that time?

3) With stuff like the non-Lazy rules, the player (in theory) will have a hard time staying at low-AIP through the endgame.  Combined with the above, this makes for a sort of "common curve" among games: stay below, say, 130 before the end, and then jump up to around 200 during the buildup to the final battle, and then either you win or you lose and probably die soon thereafter (or pull off a desperate gambit involving more warheads than I'm comfortable contemplating).  Maybe adjusting things so those numbers are more like 200 and 300 would be better, with the first number being effectively lower on the higher difficulties.  Or you can turn Lazy-AI on and win at low AIP if that's how you like to play the game.  Or you can play FS and go high AIP from midgame on if that's how you like to play the game.  Aside from those two alternatives, what other "curves" need to work?

Anyway, not trying to step on anyone's toes here, but I'm wondering what our goals really are/need-to-be.

[Diazo]

I'm not sure I have a specific goal beyond what I previously stated.

That being where the current game sits in terms of power at AIP 300 is what it should be at AIP 500.

I think my position is more from the fact that over the long term, the game has been creeping towards low-AIP being the 'optimal' strategy more and more.

So as more and more players favor it, the AIs counters at low AIP got buffed, but those buffs carried though to the higher-AIPs and so low-AIP became more and more optimal until it is almost exclusively the only way to play.

Right now because I play with such a high difficulty, taking all worlds adjacent to my homeworld for defense is not an option as it would cost too much AIP. It's not a case of me making that choice to keep AIP low for an easier game, it's the simple fact that if I take every world adjacent to my homeworld, I've lost. (For the record I'm talking down at diff 9 here, that's not a diff 10 game I'm talking about.)

Now, if it's intended that at AIP 9 I can't take adjacent worlds for defensive depth I can accept that, but in my opinion it should be an option to be able to take at least a few worlds beyond what the CSGs make you take.

Actually, what should be the expected planets a player takes? I've been assuming that because the CSGs make you take 8 worlds, that 10 to 12 worlds is 'expected' for an average low-aip to mid-aip game. Am I assuming correctly?

D.

[keith.lamothe]

That being where the current game sits in terms of power at AIP 300 is what it should be at AIP 500.

For both humans and the AI, right?  I'd be fine with that.

I think my position is more from the fact that over the long term, the game has been creeping towards low-AIP being the 'optimal' strategy more and more.

And what I'm getting at, in the interests of provoking thought rather than promoting any particular agenda, is what's actually wrong with that being the optimal strategy?  On the face of it, is it not obviously optimal to spend the minimum amount of AIP possible to reach a position (both astrographically and in terms of offensive power) where you can kill the AI, and then kill the AI?  How could that not be optimal?

And doesn't the game need to provide a sufficient challenge to optimal play (in the macro/strategic sense, I don't mean absolutely optimal tactical micro or whatever since that's not what this audience wants to spend its leisure hours doing)?

If someone doesn't want to have to play optimally (which I fully understand, as I am one of those people), that's what the lower difficulty levels are for.  Right now Diff 7 is somewhat harder than I want it to be, largely due to some side effects of the reinforcements overhaul I did last year.  But it still allows a lot more wiggle room in AIP if you don't feel like sticking to minimal-AIP.

Actually, what should be the expected planets a player takes? I've been assuming that because the CSGs make you take 8 worlds, that 10 to 12 worlds is 'expected'

That's kind of what I'd expect at diff 9, yea.

But that brings up another question: how many planets do you actually need to take to win?  I think CSGs are actually masking a problem here: the player has the actual force-of-arms to win after taking... what, 3 planets? 2?  Without the deepstriking rule could you win without taking a single planet?  You're just forced to take down the CSG network because nothing can kill an invincible home command station.

If you only need 20,000 K to win on Diff 9 (or Diff 7, for that matter), then there's the problem, right?  Of course you'll go low-AIP, why wouldn't you?

"To get defensive depth" is a good response, I think, but there we're back to the tension where if you can afford that defensive depth and still win, then someone who's willing to go without that depth (and not take other planets to compensate) can win much more easily.  Which is fine depending on the difficulty level, of course.

[Toranth]

Actually, what should be the expected planets a player takes? I've been assuming that because the CSGs make you take 8 worlds, that 10 to 12 worlds is 'expected'

That's kind of what I'd expect at diff 9, yea.

But that brings up another question: how many planets do you actually need to take to win?  I think CSGs are actually masking a problem here: the player has the actual force-of-arms to win after taking... what, 3 planets? 2?  Without the deepstriking rule could you win without taking a single planet?  You're just forced to take down the CSG network because nothing can kill an invincible home command station.

I played a 9/9 game back in the AS beta without CSGs, using 2 starting homeworlds, where I won without capturing another planet.  If there was no deepstrike, I wouldn't have killed any AI command stations.  I did it once with Champions, then once without.  It took a long time, and that was pre-Reserve, so I'm not sure I could do it again today... but I suspect better players than I could.

Was it fun?  Not really.  I spent a LONG time rebuilding between each cross-map suicide wave.  Once I'd neutered a path to the AI HWs, it wasn't even in doubt.  Just grindy.  That's the reason I almost always play with CSGs on - it makes the game more challenging.

But if we ignore CSGs, how many planets should the player need to take before being able to kill the AI?  Players who are very skilled at doing more with less like Diazo and Faulty would be able to win with fewer systems than players like me that specialize in defense and 'persistance'.  That makes it really a matter of skill and style.
Personally, I think I would like to see the human strength required to be somewhere around 50 times the original starting strength.  (That's counting a cap of Mk II as 2 caps of Mk I, etc).
That basically means getting 4 ARSs and unlocking all ships to Mk II, plus 2-3 Mk IIIs, a few starships, and maybe a Fab or two.
Knowledge wise, that works out to about only 25,000-35,000 K.  Do we think this is enough strength before the game is winnable?  Should the AI HWs be hard enough to basically REQUIRE multiple Mk IV fleetships or starships?

(yes, I know Mk II =/= 2 Mk I.  It looks Mk I equivalents actually goes 1 - 3 - 4 - 6 - 7.)

Truth be told, as long as the game is balanced enough that the theoeretical required strength is not so high that the AIP becomes too hard to handle, then I don't really mind where the balanced point is.

If you only need 20,000 K to win on Diff 9 (or Diff 7, for that matter), then there's the problem, right?  Of course you'll go low-AIP, why wouldn't you?

One of the things I noticed looking back at earlier versions of the game, was not only that AIP was so much higher ("400-600 by the end of an 80 planet map"), but that there was so much more Knowledge available (20-30 planets worth).  Since then, costs for some things have come down (Starships, from 15,000K to 4,000K) others have gone up (Mk I -> Mk III/IV fleetships went from 7,000K to 8,500K).  At the same time, there's now more available to spend Knowledge on.

If K was made less available, even for a lower AIP rise in exchange, I think I would find myself playing with even fewer unlocks.  Even in Fallen Spire, 33% additional systems to take adds up.

[Cinth]

I think the game needs a center point to balance around.   With a lot of stuff being thrown at low AIP games, mid-high AIP gets crazier faster, and sooner or later will be unplayable.

I think that we can make low AIP challenging enough though the use of mechanics like the CSG, Strat reserves, and the "lazy eye" changes that just went in.  To me that would work better than having the AI just throw more at the player.

[chemical_art]

And what I'm getting at, in the interests of provoking thought rather than promoting any particular agenda, is what's actually wrong with that being the optimal strategy?  On the face of it, is it not obviously optimal to spend the minimum amount of AIP possible to reach a position (both astrographically and in terms of offensive power) where you can kill the AI, and then kill the AI?  How could that not be optimal?

I think what is desired is the "difficulty" range between ultra low and moderate AIP to be less extreme. To accomplish this is what has been done to strategic reserves and the proposed idea of the hacking...actually making AIP in of itself not so key to everything.

The current "model" of AIP scaling is such that it trumps all. In part because the AI's offenses scale so rapidly at all stages of the game. In part because to win in of itself for so long was so very low. The strategic reserves helps to address the later. Hacking helps could help counter the effects of the former.

I really am becoming a fan of the polynomial growth model. It allows early game to scale quickly so the absolute base is not destructive, allows high AIP to hit really hard for high aip. But it could allow "mid" AIP to get a bit of reprieve so you could have more situations of it being beneficial without also benefiting low aip games. On the other hand, adding external things to skew the graph, like baseline AI checks or assests that scale more with mid level AIP works too.


EDIT: New reinforcements certainly in the right direction. I truly think it helps: It increases the stakes for everyone involved in the big picture. Getting AIP suddenly is a more viable tactic. For higher aip games the offenses of both players and AI increases faster then defense, which makes high stakes and in general a conclusion.

[Diazo]

That being where the current game sits in terms of power at AIP 300 is what it should be at AIP 500.

For both humans and the AI, right?  I'd be fine with that.

I think my position is more from the fact that over the long term, the game has been creeping towards low-AIP being the 'optimal' strategy more and more.

And what I'm getting at, in the interests of provoking thought rather than promoting any particular agenda, is what's actually wrong with that being the optimal strategy?  On the face of it, is it not obviously optimal to spend the minimum amount of AIP possible to reach a position (both astrographically and in terms of offensive power) where you can kill the AI, and then kill the AI?  How could that not be optimal?

I suppose part of this is looking at the AI's unlocks and tier values.

The AI gets a new ship every 200 AIP. That means there's only a 50/50 chance the AI will unlock another bonus ship type in a given game before you are attacking the AI homeworlds. That tells me that somewhere in the mid to mid-late game it is expected that you will hit 200 AIP so the AI gets a new ship.

And the AI goes to tier II waves at 230 AIP for Diff 7. Again, I would expect that in the mid game is when you are expected to be facing tier II waves and tier III waves not unheard of in the end game. Now, as of a few patches ago this interpolates the mark levels in the wave so this is not a huge deal.

Looking at those two numbers makes me thing "I can be attacking the AI HW's at 120 AIP, both of those mechanics are kind of superfluous." Then add in the balance creep we've had towards low-aip play over the past year(s) and that is pretty much what prompted me to start this thread.

Moving the AI Tiers is a big deal, and already happens on the higher diffs, but maybe the AI should get a new ship type ever 150 AIP to reflect today's fact that AIP is lower on average?

D.

[KDR_11k]

Just keep in mind when rebalancing that planets offer more than just knowledge.

The new non-lazy logic definitely suggests that Keith doesn't really want people to stick to sub-100 AIPs.

[keith.lamothe]

The new non-lazy logic definitely suggests that Keith doesn't really want people to stick to sub-100 AIPs.

Well, no, I don't want people to stay that low the whole game.  Maybe through the midgame, but for the final assault I feel like something's wrong if AIP is < 100 when the first core guard post dies.

But if someone wants to play that way, I don't want to take it away.  I don't say I'd make it possible now if it wasn't already, but I won't take it away.  Hence the lazy toggle.  But I'm not balancing the game to be challenging with lazy on: a player wanting that on and a challenge can compensate through the other lobby options.

But with lazy off the definite "hint" to the player is "you know, if you have less than 200 AIP you really want to gather more strength before attacking the AI HWs".  The rules aren't particularly subtle   But not into CSG territory.  We'll see if people actually feel a need to go that high or if they just cheese their way through the strategic reserves and still pull off a double-kill in a short enough timeframe that the floor increases from the core posts don't kill them.

[kasnavada]

The AI gets a new ship every 200 AIP. That means there's only a 50/50 chance the AI will unlock another bonus ship type in a given game before you are attacking the AI homeworlds. That tells me that somewhere in the mid to mid-late game it is expected that you will hit 200 AIP so the AI gets a new ship.

And the AI goes to tier II waves at 230 AIP for Diff 7. Again, I would expect that in the mid game is when you are expected to be facing tier II waves and tier III waves not unheard of in the end game. Now, as of a few patches ago this interpolates the mark levels in the wave so this is not a huge deal.

Low-AIP games just circumvent those mechanics completly. From those values you'd think that "early game" (initial expansion) would be below 200, mid-game (preparing first assault) would be around 400, first assault up to 600, final assault around 800.

That would mean that the waves get higher in level while bonus ships are being unlocked, making the AI evolve from our actions.

Instead with the low AIP route, players manage to fend off... waves of level one, sometimes level 2 ships during the entire game and get one, maybe two bonus ship at most for the entire game. Ok, I kinda agree with the "try not to draw more attention than possible"... but wouldn't the game be more fun if at end-game players would fight waves of lvl 3 ships / starships ? And get level 4 ships as an extra challenge or when mistakes are made and +AIP stuff you're supposed to be protecting get destroyed ? And wouldn't the game be more interesting if the AI had more variety in the ships it uses ?

What I've seen during last years are lots of updates to make the AI stronger by adding mechanics to them, shouldn't we make the AI stronger by using mechanics that exists already ?

[Fluffiest]

Actually, what should be the expected planets a player takes? I've been assuming that because the CSGs make you take 8 worlds, that 10 to 12 worlds is 'expected'

That's kind of what I'd expect at diff 9, yea.

But that brings up another question: how many planets do you actually need to take to win?  I think CSGs are actually masking a problem here: the player has the actual force-of-arms to win after taking... what, 3 planets? 2?  Without the deepstriking rule could you win without taking a single planet?  You're just forced to take down the CSG network because nothing can kill an invincible home command station.

If you only need 20,000 K to win on Diff 9 (or Diff 7, for that matter), then there's the problem, right?  Of course you'll go low-AIP, why wouldn't you?

"To get defensive depth" is a good response, I think, but there we're back to the tension where if you can afford that defensive depth and still win, then someone who's willing to go without that depth (and not take other planets to compensate) can win much more easily.  Which is fine depending on the difficulty level, of course.

Forgive me if I'm missing something obvious, but doesn't this suggest that the best fix would be to reduce the amount of knowledge available at the start of the game? This makes all game types more difficult, but hits low-AIP proportionally harder. Each extra planet taken is still worth the same amount to the player; you just start off weaker and therefore have further to climb.