Does AI strength to AIP need adjustment? (aka, is AIP too restrictive?)

[TechSY730]

Based on some feedback, there seems to be a feeling among some players that the AI is still too much of a wimp at low AIPs, that there isn't much to discourage ultra low AIP games (short of Core Shield Generators, but I will get back to those in a moment), there isn't enough to encourage taking the risk to raise AIP to get new planets, and, on high difficulties at least, the increased danger of raising AIP very quickly overshadows almost any benefit you can get from it.

Relevant quote:

And finally, AIP plays into this. I would vote for making the "base" AI strength stronger (aka, the AI puts up more of a fight at low AIPs), but doesn't grow quite at strong as fast as AIP goes up (both the linear of <=8 and the exponential of >8). This should hopefully encourage people to be more willing to take stuff (both because they won't be punished for it quite so hard and because if they stay with low planet strats they will now be further behind due to the base strength increase)
However, the topic of AIP is big enough for its own forum post. (that is this thread)

This I think would help with the issue of this style of play being dominant in the metagame (this sort of minimalistic strat being viable is fine, but it becomes degenerate when in most setups it becomes the most competitively "optimal" way to play).
It would also make players feel less hesitant about taking a planet for a capturable.
It would also feel a much more natural way to "smooth" pacing than the currrent core shield generators do, which I still contend feels like an artificial barrier "band-aid" for a deeper pacing issue.

Also, I would like to see the >8 AIP to strength relationship go from exponential to polynomial. I think this is one of the biggest factors making ultra-low AIP games at high difficulties often times the only viable way to play, in often un-fun ways.
The <= 8 curve should stay linear.

[Faulty Logic]

AIP is supposed to be really restrictive. It feels about perfect to me (maybe even a little too nice).

[chemical_art]

AIP is supposed to be really restrictive. It feels about perfect to me (maybe even a little too nice).

Do you feel this way across the board?

Like, if the jump in difficulty from AIP 10 to 100 wasn't roughly ten times the response, but the jump from 100 to 190 is less then 2 times the response, that is OK?

Would it be better if at say, 250 AIP, the difficulty of AIP is the same, but after 250 AIP the game is harder now, but the jumps pre 250 are not so dramatic and polynomial instead of exponential?

[Faulty Logic]

Like, if the jump in difficulty from AIP 10 to 100 wasn't roughly ten times the response, but the jump from 100 to 190 is less then 2 times the response, that is OK?

Would it be better if at say, 250 AIP, the difficulty of AIP is the same, but after 250 AIP the game is harder now, but the jumps pre 250 are not so dramatic and polynomial instead of exponential?

The difficulty increase for most difficulties is roughly linear, with higher ones getting extra pain piled on. Linear is good because its intuitive, and the feeling of the first planet you take being far more important than the tenth in terms of AI response (proportionally) is balanced by the first planet adding far more to your strength than the tenth, proportionally.

The exponential stuff only happens on higher difficulties, where players shouldn't be able to get a fair deal, so its fine in my book too.

I would say to avoid arbitrary cliffs like the plague, though, which the tech-scaling does quite well now (Heroic and spire hammer notwithstanding).

[_K_]

Well, i'd put the opposite way.

As i understand the AI strength at AIP < 30 was supposed to be extremely low to give the players a better chance to survive during the very early game, when you havent accumulated much of anything.
Once you start expanding, it is supposed to grow according with the growth of your strength.

However, with the AIP reducers, it is quite possible to spend large part of the game in this "spawn protection". In fact, never going outside this zone is pretty much the only way to beat AI on extreme difficulties. Any kind of territorial control is not really viable there, as you cannot afford to have any extra planets.

I see 2 possible solutions there:
1) Set the starting AIP to 20 or maybe 30, and trim the AIP-dependent AI strength curve accordingly

2) Make direct direct AIP redicers in a different way, for example they could reduce current AIP by a percentage, stacking with each other multiplicatively. For example, DCs could decrease it by 10%, and CPs by 20%. The exact values would probably have to be adjusted depending on the point, at which we want the system to be equal with the current one.
With 10% reduction on a DC, it equalises with the current high-diff DC (-10 AIP) at 100 AIP, reducing it to 90 in both cases. For plower AIP, the old DCs are superior, for higher AIP the new ones would be.

[Wingflier]

Seems fine to me.

[Diazo]

I think the big question is factoring in minor factions, notably super-weapons.

Base game only, AIP is restrictive, but it's supposed to be. AIP after all is essentially what you are fighting to keep down.

Once you enable super-weapons though, AIP means a lot less as the big threat (exo-waves) have a time component to them and so expanding faster is worth the higher AIP costs over slow and methodical. (To a point anyway.)

So, going by what we currently have, notably Core Shield Generators.
CSGs require capturing 8 systems to destroy them, that is 160 AIP.
That puts you at 170 AIP as when you should be ready to attack the AI HWs.
I'm ignoring both AIP reducers and AIP from other sources, including capturing more systems and saying they cancel each other out for this.

So, at 170AIP and 37,000 knowledge total, how do the AI attacks feel?

One of the biggest hurdles I've had with this game is accepting the fact that you are going to lose peripheral systems and captured structures, and the fact that that is not a lose condition.

Keeping that in mind, how dangerous is the AI at 170 AIP?

D.

[Hearteater]

Make +1 AIP/15 Minutes mandatory starting at difficulty 7.  Problem solved .

[Diazo]

Heh.

Game over man, game over! 

More specifically, I spend a lot of time at the beginning of the game scouting. If I have several Mk III/IV worlds to bash scouting paths through, I could easily hit AIP 15 before sending my first system capture attack out.

That's 50% bigger waves for the record.

Now, that is far from the majority of my games, but it happens.

Having said that, I'm not sure I'm against the AIP/Time being enabled, I am a slower player and that would make me move quicker.

Now if that is a good thing or not is the question.

D.

[TechSY730]

Updated the OP, if anyone is interested.

[Hearteater]

I've actually turned AIP over time on just to make me play faster.  +1 AIP/15 minutes is my current setting and it really is less AIP than it sounds like.  My games used to end around 10-15 hours, so that's only 40-60 AIP.  I'm still attacking the first homeworlds at < 150 AIP.  I'm trying to get myself up to +1 AIP/10 minutes, but since my games are like 8 hours now that's still only 48 AIP.

[Toranth]

More specifically, I spend a lot of time at the beginning of the game scouting. If I have several Mk III/IV worlds to bash scouting paths through, I could easily hit AIP 15 before sending my first system capture attack out.

That's 50% bigger waves for the record.

Actually, that's 56% bigger waves (effective AIP is non-linear).  Going from 10 AIP to your 170 AIP right before the Homeworld attacks is actually about a 2200% increase.  Waves sizes go from 30 ships base to 750 ships base.
Also, as AIP goes up, reinforcements go up, Special Forces goes up, strategic reserve goes up, and Exowave strength (non-FS) goes up.  IIRC, CPA size also goes up.

So, yes, while the straight AIP-to-Wave conversions don't seem to bad, the fact that EVERYTHING gets a buff means that the downside of an AIP increase is a lot more than it first seems.


Let me pull out an old quote from the Wiki:

Thus the breakdown of planets at the end of a successful game for me might be:
 80 planets total
 20-30 planets belonging to my team
 0-3 planets that were taken from the AI, but that we did not capture for whatever reason (nice to have buffers sometimes)
 15-20 planets that were gate raided, and had their special forces guard posts and train stations killed
 27-45 planets that I never touched, except to perhaps pass through with transports or a convoy, or possibly to knowledge raid or kill data centers on.

Out of my 20-30 planets, usually it is good if somewhere between 8-14 of those are reasonably secure from attack by the end of the game. Those are producing resources and not really at a whole lot of risk except when the AI slips past me on the front lines. If I lose some, I rebuild, and if a planet it just too hotly contested I might let it go fallow and just stay neutral.

This would result in an AIP pushing 500 when approaching the Homeworlds.

Compare that most recent AARs, where the endgame total is more like:
10-12 human systems
4-5 gate raided systems
1-2 systems destroyed but not claimed.
AIP approaching the homeworlds is usually in the neighborhood of 150.


I understand the game has changed a lot in those intervening years.  I just wanted to make the point about how MUCH AIP has come to matter.

[Diazo]

Hmmm.

I was thinking this over and realize we are running in to the chokepoint issue again.

Really it is the number of worlds you have exposed to the AI, and so how much you have to spread your forces out, that determine how restrictive AIP is.

A single exposed fortress world at 100 AIP is significantly easier then a setup with 4 exposed worlds at 50 AIP. (My gut feeling, not actually tested, but you get my point.)

I think there has slowly been power creep over the years as games were won on high difficulty by players who had, while not used cheese, set things up in there favor. Simple map types for easy chokepoints, superweapons, settings AIP auto-progress to 0, etc.

So the AI gets power creep to deal with those situations and the game 'defaults' move towards chokepoints in response, which causes the AI to power creep farther, etc.

I'm thinking there needs to be a bar set, especially for the 10/10 win 'bug report' that if certain conditions are not present, it does not count as evidence for tweaking the AI's response.

Notably I'm thinking that if a game is won where a single chokepoint defense for a lot of the game it does not count.

I'm not actually asking for changes here, I'm pointing out that (in my opinion), a single chokepoint defense is magnitudes easier then even a two chokepoint defense that when a game is won by doing that it can not be used as evidence that the AI needs a buff to avoid further power creep.

D.

edit: Actually, the (potential) changes to the Mk III fleet ship K costs would go a long way towards mitigating this issue as that will increase your late game power.

[keith.lamothe]

Given the feedback I've seen here and before, I think AIP itself is fine, at least with the rest of the game the way it is now.


But there is a significant design challenge I keep bumping up against, that I think gets at the original post's point from a different perspective:


(warning, stream of consciousness incoming)


If I don't make moderate AIP (say, the 100 to 200 range) hurt really bad (at least on the higher difficulties), then many of you never get challenged, because you never go higher than that.

Basically the game has trained you "never ever take AIP unless you absolutely need to" and you have learned that lesson well   Such that I then have to "chase" by bringing the serious AIP consequences down to where you keep it or the game's going to be a cakewalk (unless you have some non-AIP-based threat enabled like hybrids or FS, which can pose moderate to serious problems of their own).


And I don't think there's actually anything wrong with that.  But I do think you aren't being motivated enough to take AIP.  There's of course the CSGs, but let's be honest: that was us forcing you to take more AIP, and it's not really adding much fun to the game.  That kind of sledgehammer approach isn't healthy.  So CSGs were made optional.  Even with them, all it's done is raise the "floor" a bit: you still don't take more than you absolutely positively are definitionally required to by the scenario.  Well, not all of you stick that close to the floor, but you get the idea.  And balancing the game to keep low-AIP wins from being too easy just pushes everyone else towards low-AIP.

Again, not necessarily a problem.  Everyone either plays low-AIP, super-low-AIP, or on a lower difficulty, or whatever.  I don't think that's actually making the game less fun per se, though it does put the crimp on capturables and whatnot because the implicit costs in getting them are so painfully high.


But I do wonder what it would take to motivate people to take more planets.  In theory, simply the need to have enough m+c income should drive you to take a fair bit of territory, in order to be able to produce the high-mark stuff and do it quickly and so on, but instead y'all scatter in basically every other direction conceivable:
a) You avoid building high-mark stuff (unless it's really durable), so you can get by with less income.
b) You pay a ton of knowledge getting either harvester upgrades or econ station upgrades, so you can get more income from fewer planets.
c) You just wait a long time for your lower income, watching netflix or whatever, so you can get more income (overall) from fewer planets.
d) You complain in the forums that the game isn't giving you enough m+c   Not so much anymore, but mainly because we made the harvester upgrades so incredible.

The problem there, I think, is that the game is swooping in with evil-villain cape on, saying  "Hahahahaha, with these dastardly-high costs for effective units you are thereby compelled to take more planets so I can get more AIP and threaten you more! ... wait, what are you doing?  You're not taking planets... you're... watching a movie.  Oh, I guess that is kind of a viable alternative, hrm."  It's just not a very effective villain tactic, to be honest.

There is a way I could fix that: make the AI just keep hitting you fast/hard enough that you can't survive waiting an hour for m+c, or at least that you're spending more m+c on defense than you're getting from your low income.  So when you played on too-high-for-you difficulty you wouldn't be choosing "AIP or tedium", you'd be choosing "AIP or death".  Which would translate to "spectacular death a little later or less-spectacular death pretty soon", so everyone's happy, right?

But something tells me that wouldn't go over so well

[TechSY730]

(long snip)

You know, it's times like this I really respect game designers and software architects in general.

There are so many factors to consider, and goals that people want but you can't "maximize" them all (as that would conflict with the other goals too much if focused on too much), so you have to balance which goals and to which extents based on what the "people" want. Making this more difficult is different ways to implement "goals" impact how viable ways to implement other "goals", and their "effectiveness" once implemented, in complicated, hard to foresee ways.

But what the "people" want can be extremely varied, making it hard to figure that out. Especially difficult is to figure out goal "sets" that conflict with your core design goals/tenants you established for the product, and try to avoid them while still addressing what those people want, if possible.


Basically, since you can't please everyone, who do you please, and in what areas, in ways that are still compatible with the point of your product?
This is a tough question for anything more complicated than, say, tic-tac-toe.