AI Progress Reducers

[Tridus]

In AIWC, there are a number of ways to do AI Progress Reducton. Some are always there. Some are lobby options. Some are interesting, and some are less so. Lets talk about the different things in play on these.

Should they exist at all?

My opinion on this is absolutely yes, in some form. These create high priority targets for players, which creates things for players to want to do, which creates interesting situations (sometimes). If we had an objective/tutorial type system, these are targets that it could point to and say "see that? You want to blow that up. It's extremely good if you blow that up."

Those are all good things. Doubly so when the last co-processor is a mile away from where you want to be and you either have to choose to ignore all of them, change your plans, or go learn how to deepstrike raid (and take the consequences of that).

How many should there be?

Fundamentally a balance/tweaking question, and I don't care a lot. I think the sentiment is that in AIWC right now, there are too many worlds with things on them to capture/hack/whatever, because it's gradually expanded over time. So this is a chance to scale that back some while keeping the better ones. (Design Backup Servers are a candidate to vanish entirely because with a new game, the same mechanic could exist within the hacking interface itself and not require a planetary target.)

Which types of reducers are good, and why?

I think this is where it gets tricky. My favorite one is the SuperTerminal while my least favorite are Data Centers, and Toranth already explained why in the other thread:

I prefer the SuperTerminal over them in terms of implementation because Data Centers are so easy to hit, but the SuperTerminal can get very out of hand if you get careless with it. It's got a higher risk/reward ratio.

SuperTerminal, I like - It's fairly limited in effect, but it is a lot of fun (and !!FUN!!) to use.
The Data Centers, though... So few HP that even a single shot from a Raid Starship can kill them, and Raids shoot through Forcefields, so not even that can protect them.  There are games I don't touch the SuperTerminal.  I don't think there been a game that I didn't get all the Data Centers, though.

Fundamentally, the SuperTerminal is an interesting choice. There's one of them (at most, it's not always on a map at all), and it does something you want. But it's not easy to use, and carries the risk of going very wrong if you lean on it too hard. That risk of pushing it just a little bit more adds a lot of fun to using it, and when it goes wrong, it leads to spectacular stories.

Data Centers are dull, comparatively. They're extremely easy to raid, the AI has very few ways to stop you from raiding them (because Raid Starships), and there's never a reason to *not* get them. If I were going to chop one, it'd be these.

Co-processors are more interesting because you don't get any benefit until you can get the fourth one, and sometimes that can be a problem. They also interfere with supply, which means if one is in the way, you have to make a decision about how to proceed: take out the one that's in the way and eat the AIP until you can clear the rest of them, go out of your way to clear them all to get that one out of the way with no penalty, or alter your plans to work around the one that's in the way. That can force the player to have to react to the hand they're dealt, which creates some interesting choices.

On the lobby options side, you have Spire Civilian Leaders, which conceptually I like, but in practice could use a rework. They have to be protected for a VERY long time to pay off, they have virtually no limit on how much AIP they can reduce (which contributes to fiddly complications like the AIP floor), and some games the hand you get dealt with them is extremely advantageous for you (if you can capture 4 within 2 hours because they happened to spawn in the right places, you have 6 AIP reduction every hour for the entire rest of the game, even if you never see another one). In a "conquer the galaxy" style of game, they're absurdly overpowered and become a "well if I go slower, the AI HW assault will get easier" mechanic. That isn't something we want to encourage, so a rework is in order before I'd ever consider bringing them back.


For me, the overall goal of reducers is to create interesting things in the game, and give you a way to reduce AIP some. Both the SuperTerminal and the Co-Processors give us those things, and just having those makes the reduction a smaller amount (unless you really push the SuperTerminal), so IMO I think that's adding the most to the game, while significantly cutting down on the number of planets with special things on them.

Thoughts?

[PokerChen]

I'd chop the civilian leaders first, easily before data's centres. Unlimited +unconditional reductions here is a no-go, especially as it encourages longer game time.

Overall, I'm not all that keen on having AIP reducers, but will roll with it. Each seems like a "must-do" checklist, and there's little penalty to them. Three Mark-I raid starships in Classic was generally sufficient to deep strike a few hops in, which is not much effort.
Suggestions:

• Give Data Centres set pieces to protect them., And buff their HP somewhat.
• Give players reasons to NOT kill every data centre they see. For example, they are needed to gain access to AI database (it's a freaking data centre, after all...).
• For LoLs, let each living data centre increase the vulnerability of AI to your hacking, then make hacking somewhat uniquely useful against AI homeworlds.
• Have two sets of co-processors, of 3 each, target than one of 4.

[Tridus]

I'd chop the civilian leaders first, easily before data's centres. Unlimited +unconditional reductions here is a no-go, especially as it encourages longer game time.

Agreed. They need a rework at the least, and given time constraints for 1.0, doesn't seem worth doing initially. Maybe a "one day in the future" thing.

Overall, I'm not all that keen on having AIP reducers, but will roll with it. Each seems like a "must-do" checklist, and there's little penalty to them.

Do you think that applies to the SuperTerminal, though? Especially the current incarnation that costs hacking to activate?

Suggestions:

• Give Data Centres set pieces to protect them., And buff their HP somewhat.
• Give players reasons to NOT kill every data centre they see. For example, they are needed to gain access to AI database (it's a freaking data centre, after all...).
• For LoLs, let each living data centre increase the vulnerability of AI to your hacking, then make hacking somewhat uniquely useful against AI homeworlds.
• Have two sets of co-processors, of 3 each, target than one of 4.

Neat ideas. The Co-Processor one is pretty interesting, since each AI could have one. I wonder if you could integrate the hacking idea into Co-Processors though, instead of data centers (it is doing processing, after all!). Then you could leave them up for hacking opportunity, or take them out for the AIP reduction, and just chop the data centers entirely as well.

[Draco18s]

I would remodel the Civilian Leaders before removing them.

Data Centers I think should stay in some form, as the "quick and easy" popcorn variant.  But I'm thinking more like "two per galaxy" than however many there are now.  It's the one that even new players go "oh that's easy" and have a goal that doesn't require too much planning.

Re Leaders:
Fast change method (i.e. I didn't think about this very hard):

• +10 AIP on game start
• +10 AIP when world first goes on alert [once only]
• -30 AIP when [controlled by AI and destroyed by player] or world flips to player control
• +30 AIP when [destroyed any other way] or world flips to AI control

The "on flip" handles the new AI recapturing territory.  The initial AIP on game start handles the previous count-up mechanic, plus 10 more if the planet goes on alert.  So deep-striking it would be a valid tactical decision.  Note: there is currently missing the "long term" benefit of protecting these (resources? new ships?).  The plus 30 when "destroyed at other times" makes sure that the player doesn't take the world, then murder the civilians to prevent recapture.

[Toranth]

As I said before, I like the SuperTerminal.  I've had huge amounts of fun over the years with that thing - setting up a massive beachhead, killing tens of thousands of ships during the hack, sometimes failing and watching the galaxy be wiped by an AI fleet of hundreds of thousands...  Basically, as AIP reduction, it's mixed and dangerous, and involves the player doing stuff (often, a LOT of stuff) to use.

Co-Processors I don't think much of either way.  As a want-to-win player, all AIP reduction is good, and a for-fun player, they do require a little complication (mostly just exploration and planning) before being used.  On the other hand, they're basically datacenters that need to be killed all four at once, and raise the AIP Floor a bit when doing so.

But data centers are just "Of course".  In most games, I've found and destroyed all the data centers before I conquer my first planet.  Certainly before I take the second one.  Considering how essential it is, how brainless the decision is, and how easy it is... I think something could be done.  The ideal choice, I think, would be to make it a opportunity cost sort of situation.

Perhaps, you could be required to hack the AI to find out which data center holds the AI's "human threat history" database, so that destroying it would cause the AI to think you less dangerous.  This would effectively turn HaP into AIP reduction, although I'm sure it should be at a not-very-good ratio.
Perhaps the choice should be between the data center being destroyed for AIP reduction - or captured for an upgrade, or resources, or some such.
Also, fewer of them.  Maybe they should also only be generated as the game goes on (like Backup Design servers are) so that you can't ride the AIP Floor with -100 AIP early in the game.

[Toranth]

I would remodel the Civilian Leaders before removing them.
...
Re Leaders:
Fast change method (i.e. I didn't think about this very hard):

• +10 AIP on game start
• +10 AIP when world first goes on alert [once only]
• -30 AIP when [controlled by AI and destroyed by player] or world flips to player control
• +30 AIP when [destroyed any other way] or world flips to AI control

The "on flip" handles the new AI recapturing territory.  The initial AIP on game start handles the previous count-up mechanic, plus 10 more if the planet goes on alert.  So deep-striking it would be a valid tactical decision.  Note: there is currently missing the "long term" benefit of protecting these (resources? new ships?).  The plus 30 when "destroyed at other times" makes sure that the player doesn't take the world, then murder the civilians to prevent recapture.

I think Spire Civilian Leaders need a rework, but it may be hard to get a good new system done for 1.0.  I don't have any good ideas, myself, either.

As for your suggestion...  Assuming whatever AIP variety exists in AIW2 is similar in behavior to AIP from Classic, that +10 per Leader at the beginning of the game would be HELL.  There're normally 10 Leaders, IIRC, so that's starting the game at 110 AIP at second 1, or 11 times the base AIP you normally begin with.  That's roughly 14x-15x as much AI response, in Classic.
Also, the +10 to start, +10 when alerted vs a max -30 benefit leaves, at best, -10 AIP... Which isn't bad, but it basically means that "Find and kill all the Spire Civilians" is the default response.  Safe, easy, decent reward... and utterly heartless and cruel.

[Draco18s]

Numbers were kind of arbitrary and again the idea was put together quickly.

[PokerChen]

Do you think that applies to the SuperTerminal, though? Especially the current incarnation that costs hacking to activate?

As long as you're not gunning for minimum AIP or at some extreme 9+ difficulty, then yes it should go on the must-do list.
*Goes through patch notes again to check HaP cost and mechanics again*
At a minimum players can deal with -60 AIP ( 20 ticks at -20, then 20 more ticks as -40), and from this you obtain 3 free planets (9K knowledge, anywhere between 0~18 rocks, plus captureable or securing extra planets to make your Mk-IV factories safer). I did it last week in my test game @ Diff. 7, fairly early on, and the final response is easily manageable with mark-II triangles plus (base) full turret caps.

[ewokonfire]

Quick suggestion for Civ Leaders:  What if we made holding on to them a bit more of a tradeoff?  When the AI holds them, you go and execute them if you don't want to capture the planet.  What if the AI gained exowaves (or some other nasty offensive ability) that would try and hunt down your captured leaders?  If balanced right, they could be an interesting tradeoff between reducing AIP but gaining an additional irreplacable target to defend, and become kind of like a 'slow-motion superterminal'.  As they are now, I always enable them, but that's because I have slightly esoteric preferences with regard to lobby settings.

[havikryan]

I actually really like civilian leaders, it was basically a early game trade off for late game power, though them infinitely reducing AI progress is too strong yes.

I really like such mechanics, you get better X but it makes Y more difficult to do, in this case the early game got allot more difficult(for me at least) with ai progress going up so quickly and really being encouraged to explore to find them.

I frankly have no idea how to balance, you can limit ti how far they can reduce it, so say they will never reduce ai progress below 100 or will never reduce ai progress below X% of the total ai progress you've accumulated.

another added difficulty would be the ai realising the threat of them once you have more then 1 and sending strike forces to try and deal with them.

I personally think not reducing ai progress below x% of total accumulated ai progress seems like the best solution, this way they are useful but not to the extent it neuters the ai if you simply play defensively for a couple of hours

[tadrinth]

In most games, I've found and destroyed all the data centers before I conquer my first planet.

Doesn't that trigger deepstrike threat?  Is the deepstrike response just trivial at 10 AIP?

Sounds like deep strike spawns need to be buffed in some way, to make this play style less tempting.  Maybe if you go more hops out, they start acting like an AIP of 200 instead of the current, or something.

What if the AI gained exowaves (or some other nasty offensive ability) that would try and hunt down your captured leaders? 

I like that idea.  I would also make each civilian leader only able to reduce AIP by a certain amount before they despawn.  Like a spire archive, basically, where you only have to protect them for a certain amount of time.  That gets rid of the infinite reduction.  They can spawn a cool military unit when they're done. 

As long as you're not gunning for minimum AIP or at some extreme 9+ difficulty, then yes it should go on the must-do list.

The ST should become *more* important as diff goes up, because AIP is more dangerous and Data Centers offer less reduction. 

Howerver, I've STILL never done a Superterminal hack; my last two games were Diff 10 and Diff 9.  The former is in the AAR section (Limburger vs the Angry Gods); I was basically riding the floor which makes the ST much less worthwhile.  The latter, I was GOING to do an ST hack, but a nomad conjunction right as I was about to start the ST hack forced me to do an early AI HW assault (because Core Raid Engine). Then the floor was high enough to make it not super worthwhile, and another Nomad conjunction gave a a backdoor to the second AI HW which I used to win. 

I think an interesting addition would be a mini-terminal, which can be hacked to reduce AIP by a certain fixed amount with no floor increase.  You could even gradually replace data centers with these as the difficulty increases. 

[Cinth]

Doesn't that trigger deepstrike threat?  Is the deepstrike response just trivial at 10 AIP?

Sounds like deep strike spawns need to be buffed in some way, to make this play style less tempting.  Maybe if you go more hops out, they start acting like an AIP of 200 instead of the current, or something.

Deepstrike threat is just threat that comes from a deepstrike.  It's nothing special.  It usually ends up as part of the threat fleet, so it can be dealt with later.

I think fuel is mean to tone down how much you can do with deepstrikes.

[tadrinth]

If the threat generated can be dealt with later, that may just mean that there isn't enough of it being generated.

I'm not sure this kind of sneaky raiding would be all that fuel intensive. It might depend on having a champion, though, and I don't remember seeing those in the design doc.

I do wonder what kind of AIP auto-progress Toranth is playing with, though.  Auto-progress is an important counterbalance to strats which are very safe but also very very time-consuming.

If careful data center raiding takes long enough, you're at risk of CPAs getting bigger faster than you're getting stronger, but the first CPA is usually quite tame. 

[kasnavada]

My opinion is that AI reducer must be reduced... until the AIP Floor mechanic does not exist anymore. To do so I'd have removed data centers for sure. Too easy to burst. Too brain-dead as an objective. Also, the last versions of AI War had enough objectives already. Both others have more interesting mechanics, so...

Which also means that game has to provide better incentive to actually have the player raise his AIP. CSG forced you to do some capturing, but apart from that... Frankly, assuming default options (1 AIP / 30 minutes I think ?), unless floored, there is little reason to take an additional planet. The AIP raise from taking one is it's the equivalent of playing 10 hours more. There are very few planets in a game, if any, which give you enough of a boost to be equivalent to playing 10 more hours.

Possibly, "forcing" the player to want to raise his AIP needs another thread though.

(Things like someone winning with a single planet... please, only with heavy tweak of the game options / mods, please, please, please...)


PS: wasn't fuel just power that applied in to ships in the latest version of the design doc ?

[Mánagarmr]

My problem with AIP reducers is that they are, without fault, the optimal strategy. If you don't go for them, you are essentially shooting yourself in the foot. That makes them a lot less fun for me.