Author Topic: (Thought Exercise) Reduce knowledge to reduce AIP scaling.  (Read 7772 times)

Offline Diazo

  • Master Member Mark II
  • *****
  • Posts: 1,717
  • I love/hate Diff 10
(Thought Exercise) Reduce knowledge to reduce AIP scaling.
« on: April 10, 2013, 11:52:52 AM »
Okay. First, this is a thought exercise, not an actual suggestion (yet).

From another thread:

The basic problem is you don't get anything from having 400 AIP that you didn't get by 200 AIP, but the AI is twice as hard.

10 planets worth of knowledge is nothing?

There should also be at least a few fabs in those planets (on most maps anyway).
Yes, 10 planets worth of K is nothing, because I've already unlocked everything meaningful.  Nothing, not K or Fabs, can give me enough power to match doubling the AI strength.  I already end 9/9 games with ~5-10k  Knowledge unspent, and I'm doing 10-12 planets.

Now, playstyle differences strike again, I certainly don't end a 9/9 game with that much K unspent.

Having said that it did prompt a thought.

Over time, and especially for starships, knoweldge cost have come down, and down a lot.

What are peoples thoughts on reducing Knoweldge (both starting K and/or K per planet) in return for cutting back on how fast AIP scales?

My logic behind this is that Knowledge is a relative resources. The more K you have, the less more K is worth, relatively.

By reducing the amount of K you get, that critical point of "I can take the AI HWs" is reached slower and requires taking more planets.

This would hurt low-aip players more then high-aip players, but at the moment the low-aip strat is so much more viable then the high-aip strat that I am okay with that.

Indeed, the point of this change is to make the high-aip strat more viable. Right now if you are facing Mk III waves (excluding stuff like superweapons), you've probably already lost the game.

However, AIP right now scales to give low-aip a challenge and so scales in the high-aip range too much (imo).

This all comes out of my stance that low-aip games have become too powerful and so the AIP scaling to give low-AIP games a challenge has resulted in high-AIP games becoming too challenging. A 9.0 or 9.3 diff game should be winnable with a high-aip strategy. Right now I'm not sure it is. (Currently testing this in an AAR though).

D.

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #1 on: April 10, 2013, 12:03:24 PM »
I'd first really rather K unlocks get a pass to make them all viable.  There are a ton I just don't bother touching.  Like Transport IIs are okay, but for 4k Knowledge, I'd really rather something a lot more impressive in terms of what it opens up.  If it were a beastly Assault Transport that held maybe 50 units instead of 200 but released them all at once, and had a ton more health and armor...then I might use them.  Throw in Missile immunity and I could see them really being useful for deep Raiding.


Offline Cinth

  • Core Member Mark II
  • *****
  • Posts: 2,527
  • Resident Zombie
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #2 on: April 10, 2013, 12:29:03 PM »
Just going to throw out that I use pretty much all but 13-14 of the unlocks.
Quote from: keith.lamothe
Opened your save. My computer wept. Switched to the ST planet and ship icons filled my screen, so I zoomed out. Game told me that it _was_ totally zoomed out. You could seriously walk from one end of the inner grav well to the other without getting your feet cold.

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #3 on: April 10, 2013, 12:37:19 PM »
You're playing FS with a ton of AIP.  Basically, you might as well use them.  But I'm betting you could win games with 20,000 K left over if you wanted (or even more).

Offline Cinth

  • Core Member Mark II
  • *****
  • Posts: 2,527
  • Resident Zombie
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #4 on: April 10, 2013, 12:40:59 PM »
Not at higher difficulties.  With all the modifiers to AI strength I play with, I need everything I can get to hit exos with.  Stopping 1 million FP is no joke and you have to do it without getting caught in an economic death spiral. 
Quote from: keith.lamothe
Opened your save. My computer wept. Switched to the ST planet and ship icons filled my screen, so I zoomed out. Game told me that it _was_ totally zoomed out. You could seriously walk from one end of the inner grav well to the other without getting your feet cold.

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,562
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #5 on: April 10, 2013, 12:51:54 PM »
You're playing FS with a ton of AIP.  Basically, you might as well use them.  But I'm betting you could win games with 20,000 K left over if you wanted (or even more).

I'm sure he could. Players could probably also win the game (maybe with "Lazy AI" turned on only) with nothing but 1, maybe 2 planets worth of econ (that count includes the HW), good defensive setup to handle "backwash", good micro skills, and lots and lots of patience on all but the hardest of difficulties.
The issue is, is that fun? Why would you do that to yourself when taking a few more planets would open up a ton of new "funner" approaches? When I mean more planets, I don't mean taking a third of galaxy or anything like that, but maybe 4, 5, or 6 planets (which is still pretty small scale, especially when data centers enter the picture).

While I would agree that this "tipping point" of where the AI's growth overcomes the benefit of a new planet (even with awesome stuff on it) does come a bit too soon for lower difficulty games (like around 400 AIP for difficulty 7, which seems rather low), the fact that you don't need a ton of stuff to win if you have crazy skill, patience, "tolerance for grind", and willingness to "sacrifice fun" doesn't mean the balance is necessarily broken (unless that sort of technique works on ultra high difficulties).


Now I agree that some of the techs need knowledge balance adjustments.
I also find it interesting to look at the K per planet balance over the years.
Way back in the day (like around 3.0 and before), it was 2000 knowledge / planet. Then it went up to 2500 knowledge. Then finally up to the current 3000 knowledge.
Each time, the prompting was that there is more stuff to try out, and the AI is getting more stuff to use too. (Not saying I agree or disagree with the reasoning or whether it overcompensated, just stating the original rational)
« Last Edit: April 10, 2013, 01:35:40 PM by TechSY730 »

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #6 on: April 10, 2013, 01:05:18 PM »
Taking 6 more planets can easily result in pushing AI strength up to x3.  There is nothing those 6 planets offer that compares.  So what if I unlock Mark IIIs when the AI hits me with 2250 strong waves instead of 750.  I'd love for their to be funner options when taking more planets.  But there really aren't.  There was discussion of changing fleet ship K prices and that might put them into the realm of worthwhile.

But stuff like turrets are a losing proposition.  To unlock them costs K, which means you need to take more planets.  Which increases AIP which means the turrets are facing tough opposition.  It is a race that turrets lose.  There are a few Mark II turret unlocks that are worthwhile (Mark II Basic) but after that, it really isn't.  Because even if the K spent on turrets perfectly balanced out the increased defenses required, you are also increasing reinforcements which means your offensive fleet needs to deal with that, and you just spent all that K to solely increase your defenses.

Offline Diazo

  • Master Member Mark II
  • *****
  • Posts: 1,717
  • I love/hate Diff 10
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #7 on: April 10, 2013, 01:08:37 PM »
I also find it interesting to look at the K per planet balance over the years.
Way back in the day (like around 3.0 and before), it was 2000 knowledge / planet. Then it went up to 2500 knowledge. Then finally up to the current 3000 knowledge.

I suppose this is really what I was getting at, without having a clear idea what I was getting at.

A single HW gets you 13,000K. That's one Mk III and 2 other Mk II fleet ships. Combine that with the stuff you get given, I can crack a HW quite easily with that at low-aip, up to a pretty high difficulty. (I'm ignoring stuff like the strategic reserve here.)

Now that stuff like the special forces, the strategic reserve and other balance changes to nerf the low-aip strategies, mid to high-aip strategies are in theory a more powerful option then they were.

Expect because AIP scales so steeply they still can't really compare.

I suppose my formal position on this is that AIP scaling needs to come down somehow, and for balance reasons the player is looking at a nerf to accomplish this.

Doubling the AIP doubles waves sizes and I assume reinforcements also.

From AIP 10 to AIP 50 the player gets tens of times stronger, while the AI gets 5 times stronger.

From AIP 100 to AIP 200, the player gets 20% stronger (maybe 30% stronger) but the AI gets 100% stronger (doubles).

The players power goes up sharply in the early game and then levels off, while the AIs power starts off slowly and then accelerates as the AIP goes up.

This mechanic needs to stay for the game to stay recognisable as AI war, but I'd like the curves leveled off, or at least the crossing point where the power is 'equal' to move significantly higher up the AIP scale.

I suppose I'm talking about how the game 'feels' as much as anything at this point, but low-aip has certainly been too dominant for quite a while now.

The latest changes in the last couple patches will probably help, I'm trying a game now, but we'll see where those changes get us.

D.

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,562
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #8 on: April 10, 2013, 01:25:29 PM »
Taking 6 more planets can easily result in pushing AI strength up to x3.  There is nothing those 6 planets offer that compares.  So what if I unlock Mark IIIs when the AI hits me with 2250 strong waves instead of 750.  I'd love for their to be funner options when taking more planets.  But there really aren't.  There was discussion of changing fleet ship K prices and that might put them into the realm of worthwhile.

I was talking about taking 3 more planets to bring you up from, say, 2 or 3 planets to 6 planets.

So basically, yea, don't take 6 more planets if the AI is already sending you waves of like 750. Maybe you should if the AI is only sending you waves of like 75.

But in any case, Nx of what?
Consider starting the game and taking two more planets.
At AIP 10, that goes to AIP 50  (10 + 2*10), which is indeed a 5x increase in AI strength. But, at AIP 10, the waves and other stuff are so weak, that even 5x of that isn't very threatening either. At that point, it would be worth it, as getting that first planet gives could quite possibly give the player more strength such that you still have "gained" on the AI, even considering that the AI still grew 5x strength to your new planets' worth of stuff.

Going from 50 to 250 AIP is of course very different, as the player is starting to "decelerate" in the benefit per planet (enough so to offset that it you get 5 planets to get from 50 to 250, rather than the two from before), but the AI is still growing in strength at the same rate (5x AIP for 5x relative strength). It might or might not be worth it in that case.
This is where the difficulty in the game is, finding those "tipping" points for your set of difficulties and options.


I is worth mentioning that my examples sort of fall apart in very high difficulties, where the AI growth from 10 to 50 could be more than 5x, and from 50 to 250 be much more than 5x, thanks to the different formula used (an exponential one in the higher difficulties). In that case, this "tipping point" would start coming much, MUCH earlier.

EDIT: To the mathematically inclined among you, yes, I am aware that even on the lower to mid difficulty equations, AI "strength" over AIP is not technically a linear relation, thanks to the formulas indicating a >0 strength at a theoretical AIP 0. (strangely, the linear equation, y = mx + a is NOT a linear relation, go figure :P)
So all of those 5x growths I mentioned should actually be "around 5x growth".
In fact, this is the key to being able to adjust low AIP "strength" and higher AIP "strength" in different directions, the intercept (a) will adjust how much strength the AI has at low AIPs (and all the way through, but its contribution becomes less noticeable as AIP goes up), while the slope (m) will adjust how "punishing" each additional point of AIP is.

EDIT2: So, maybe the thing is to adjust when that "deceleration" to the player occurs? How would that happen? Knowledge cost rebalance? Economy rebalance? Energy rebalance?
« Last Edit: April 10, 2013, 01:44:24 PM by TechSY730 »

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,562
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #9 on: April 10, 2013, 02:04:45 PM »
Actually, now that I think about it some, a AIP 10 to 50 is almost certainly NOT an 5x growth, it is probably much less, at least in the lower difficulty, "slope intercept" form. This is due to that constant factor in the equation.

So, for difficulties <= 8, for most cases, the AI "strength" boils down to
strength = m*AIP + C.
Yes, m and C vary by difficulty, what mechanic (waves, reinforcements, etc), and many other factors, but it boils down to this form.

So, if we take a, say, 2x increase in AIP, and measure the relative growth, we get (m*2*AIP + C) / (m*AIP + C). So, as AIP -> 0, the limit is C/C, or 1 (provided C is non-0). The limit as AIP -> infinity is indeed 2. (this works for any Nx comparison, with the limit to infinity being N).
If C > 0, then what you see in terms of 2x increases is not that much growth ratio wise at lower AIPs, but much starts approaching 2x as AIP goes up.
If C < 0, then all sorts of weirdness can happen, like the AI growth actually decelerating for a bit, then spiking up, and the decelerating again, but never getting below 2x.


It is also worth comparing this to what happens if you add a fixed amount of AIP, so we get (m*(AIP + N) + C) / (m*AIP + C), which has its own curve behaviors and shows off a different aspect to consider when talking about AI growth with its own implications. (at AIP 0, this results in (m*N + C) / C = m*N / C + C/C = m*N/C + 1, and as N goes to 1, this goes to m/C + 1 (which is itself a constant purely independent of game state), but for as AIP -> infinity and N is any constant, this approaches 0, so each absolute AIP point gained becomes less significant, even though multiplying it by a constant becomes about the same significance as AIP gets high (and indeed, grows quite a bit at low AIPs))

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #10 on: April 10, 2013, 02:09:19 PM »
At three planets in my last game I was getting hit with waves of 750 :) .  Since I'm playing with CSGs, I must take 6 more.  Taking another 6 on top of that would be suicide.

Also, exponential growth at high difficulties is generally irrelevant unless you are playing FS.  On 9/9 I'm taking the first homeworld at 120 AIP tops (I prefer around 80 thanks to a ST) and the second falls before another Exo/CPA hits and at most one wave lands from each AI.  It's adding only 32 AIP.  If I get to the homeworld at 80 AIP, that's just +17 AIP.  It would matter if mid-range AIP games were possible at high difficulty, but since they aren't, it's like playing with one invisible system capped you get nothing in return for.

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,562
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #11 on: April 10, 2013, 02:16:56 PM »
At three planets in my last game I was getting hit with waves of 750 :) .  Since I'm playing with CSGs, I must take 6 more.  Taking another 6 on top of that would be suicide.

Also, exponential growth at high difficulties is generally irrelevant unless you are playing FS.  On 9/9 I'm taking the first homeworld at 120 AIP tops (I prefer around 80 thanks to a ST) and the second falls before another Exo/CPA hits and at most one wave lands from each AI.  It's adding only 32 AIP.  If I get to the homeworld at 80 AIP, that's just +17 AIP.  It would matter if mid-range AIP games were possible at high difficulty, but since they aren't, it's like playing with one invisible system capped you get nothing in return for.

So at AIP 50, that gives waves of 750 size? Does that sound a bit unreasonable to anyone else, even for difficulty 9?
What waves sizes were you getting at AIP 10? That would help answer whether that sort of crazy growth can be expected throughout the game or whether that is just due to waves starting big at difficulty 9 (and thus, you won't get hit by the exponenential "explosion" yet, and the transition from 50 to, say, 90 will be much more reasonable), or if indeed the "explosion of growth" is starting too early at difficulty 9.
(a*b^x + c*x + d has a similar sort of change in relative growth and acceleration vs behavior and magnitude at low values that mx + a does)

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #12 on: April 10, 2013, 03:54:23 PM »
I think there may be something to the general idea of "rolling back the arms race" so that both the players and the AI get less from each planet taken.

Just going back to 2000 K per planet and multiplying the AI's scaling from AIP by 2/3rds would be a start.

There are some unlocks that need revisiting, like the transport-II.  I did give it an extra planet's worth of range (away from supply) recently.
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,562
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #13 on: April 10, 2013, 04:04:48 PM »
I think there may be something to the general idea of "rolling back the arms race" so that both the players and the AI get less from each planet taken.

Just going back to 2000 K per planet and multiplying the AI's scaling from AIP by 2/3rds would be a start.

There are some unlocks that need revisiting, like the transport-II.  I did give it an extra planet's worth of range (away from supply) recently.

Something like this may be a good first step.
Not sure yet if this would encourage use of more planets' resources or not.I guess it depends on if this is more of a nerf for the player or the AI, relatively speaking.
If it is more of a nerf for the AI, then yea, we wouldn't have to worry about taking as many planets, but at the same time, ultra low AIP games are now even more in the player's favor.
If it is more of a nerf for the human, then the "tipping point" over the number of planets gained will actually come faster, even though the AIP point may be higher, which is actually the opposite of the effect that the OP was looking for.
I was kind of hoping for something that would make the early game a bit harder, but decrease the "scaling" over AIP.
Like, if the current formula is strength = m*x + a, instead of making it strength = 2/3*(m*x + a), make it strength = 2/3*m*x + 3/2*a (or some other, >1 multiplier to the old value of a). Or possibly just keep a as it is now, but only scale m.
With properly chosen values, this would mean the player is now a bit more behind at very early AIP, putting pressure to get more planets early on, without having to worry about the AI growing to much early on in the game for it to be worth it.
(A similar transform would occur to the exponential form of the equation)

This would be in addition to making planets only have 2k or 2.5k knowledge each.


Keep in mind, some adjustments to the recent changes of the HW may be needed if this happens (like a new cap for the effective AIP for strategic reserve, or held value if lazy AI is off)
« Last Edit: April 10, 2013, 04:06:44 PM by TechSY730 »

Offline Diazo

  • Master Member Mark II
  • *****
  • Posts: 1,717
  • I love/hate Diff 10
Re: (Thought Exercise) Reduce knowledge to reduce AIP scaling.
« Reply #14 on: April 10, 2013, 04:25:00 PM »
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.

As that is not possible, we are stuck talking about how exactly to go about making AIP scaling less of a cliff and that gets messy regardless of what the suggestion is.

On the wave sizes, going from AIP 10 to AIP 100 on diff 7 is a 10x increase in wave size. On diff 9 it is a 12.5x increase when you hit AIP 100.  So waves on Diff 9 scale 25% faster in size to reach the AIP 100 mark.

Except because scaling is not linear on diff 8+, as AIP gets higher, the scaling gets worse. Diff 9, AIP 200 is 27x increase over AIP 10 as compared to diff 7's increase of 20x. So at AIP 200, diff 9 has scaled 35% faster then diff 7.

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.