View Issue Details

IDProjectCategoryLast Update
0021278AI War 2SuggestionJun 6, 2019 10:41 am
ReporterMatthewYCR Assigned ToChris_McElligottPark  
Status assignedResolutionopen 
Product VersionBETA 0.866 Hotfix 
Summary0021278: The Importance of Time
DescriptionHere is my foreshadowed suggestion, It's very well linked to the design doc recently linked to me here.

https://docs.google.com/document/d/108hBRHf3viOgyIS-Wp6l9bBMWPqsjkS1uUjNKrv-2CM/

My opinion on the document is roughly, "you're getting close to the heart of the issue, without truly seeing it". Anyway on to the point.

"The AI needs its 30 minute increment from AI war 1" OR another thing that does the same job.

What's the increment do, you might ask. In the average game of AIwar1, it only accounts for 5-10% of the total AI progress or an average game of AI war, its statistically insignificant.

So first, i'll provide an experience I've had with every one of my heavy gamer friends, who have had the experience of playing Klei's Don't Starve.

When the player starts, he sees that he is slowly starving, he gets food, berries, everything to alleviate the hunger statistic in the game, until they make a bunny trap. Once they catch a bunny, and eat the bunny correctly at a crock-pot (meatballs). They pause for a second, furrow their brow "Oh..." they say, their motivation drops a lot, the second that action happens.

What happened you might ask? At that point, the player realises that the food situation is no longer an issue, there's nothing in the game that prevents the player from getting more grass, making a bunch of those traps, and creating renewable food that way, they look for a game's way to prevent this action. Trap durability, sanity, etc, are there, true enough, so there are future problems, but the main threat is gone.

What was the threat? That's the best question asked. The threat truly could be written as "if the situation continues, the player will starve to death". Let's rewrite this, using more baser terms.

"As Time progresses, the player will eventually lose".

A big thing that you need to take away from this is that, "it doesn't matter how long it would truly take my friend to starve to death, as long as it eventually happens". The effect is mostly psychological, the player sees a problem, and tries to fix this.

One more differing example, using another Klei game (they are 0000001 on my list of 'games that offend this rule and suffer for it', though they constantly half-assedly address this and it's enough to get by), this time on the opposite side of the spectrum.

In Oxygen Not Included, the player is given a bunch of people, with the goal to discover what's out there, and must survive to do it. Problem is, they are running out of air and food, whilst you might have trouble with other resources the first few times you play the game, especially toilets!, food is a problem, that you'll constantly run into and the true threat is running out of air. Honestly, they do a pretty good job at making this as difficult to solve as possible, with stop-gap on stop-gap solutions to stall the issue out as long as possible.

This game can be quite hard (on harder settings) until you realise something. The only reason you run out of air and food is because you have a lot of people. The next question you ask yourself is "Wait, but with more people, you do proportionally more work, so, it shouldn't matter how many people I have right?" And then you look for the things that are different with spending 2x as long to do something, against having 2x the amount of people to do it. For a bit, this exchange seems neutral, and both are justified, but then you notice some discrepancies.

- Plants will grow with 2x as much time, compared to 2x as many people. The amount of seeds you have do not really increase with more people.
- Animals will produce more useful byproducts, such as coal, with more time.
- Using one person for a task, results in a higher level person, giving greater efficiency, as opposed to 2 people with lower level.

Unless the game has been changed in the last month or so (played somewhat recently, but not totally recently). The best strategy i found to beating the game is.

- Don't get more people
- Kill off all but 1-2 of your starting people (by hunger), and dispose of the bodies properly to prevent stress

And true enough. The game was a lot easier when I did this. compared to the amount of times I starved to death trying to feed a large colony, food wasn't even an issue.

So what could you draw from this.

In Oxygen Not Included, Time FAVOURS the player. That is, the more time that passes, the greater the advantage the player has, therefore the player should attempt to maximise the amount of time needed to do tasks. This i found... very... silly. As the only reason you want more people, is actually impatience, until much, much later, when you can find other justifications. (too far to walk, etc). This is very irritating when playing the game, as you're literally fighting your impatience to get more people, as it worsens your position.

And that it's very hard to make a time-neutral game. It's almost always one way or the other.

Now back to AI war.

What happens with raw time in AI war 1 and 2. Does your position get better or worse?

Let's think about what happens if the player just sits in base for 50 hours, and only defends himself

In both AI war 1 and AI war 2, well, eventually the CPA will hit and the player will lose, so...

Lets change the scenario slightly.

Now the player has 10 bases, what happens if we wait now?

AI war 1 - The player will eventually, inevitably, lose, even if he can survive the CPA, AI progress is increases. So - The player is incentified to go and do something about it

AI war 2 - Assuming the player survives the CPA. he's probably still alive 50 hours later, nothing has changed.

So we can see the game looks to be "Time Neutral", time neutral is 'fine', neither the AI or player got stronger 50 hours later. However, if we take a while to think about it a bit more, there's a problem.

Nothing prevents the player from taking the following actions
- Performing raids to neuter planets, making future conquest easier
- Attempting repeatedly, to try and win a fight that he has low probability of winning
- Hunting down - ANY - amount of hunter fleet, or rather, since more is probably generated, hunting it down to the lowest possible number that he can reasonably get it to
- Reducing warden fleet size
- Repositioning metal buildings into better positions. (hey look, metal costs coming into play)

So it could be argued that time... is on the player's side.

No matter how tedious is may be, there's always something you can do that contributes towards winning without increasing your loss chance. This eats at any experienced gamer, even if they don't know what's causing the disconnect.

Now you could truly hunt down every single extraneous way the player could get an advantage whilst keeping the game neutral... however, its a near endless list, but that's the thing, in most games, time is not neutral at all, its VERY difficult to maintain time neutrality. The best we can do, is always make sure the player is incentified to move forward, rather than keeping the current status quo.

AI war 1 prevents the most tedious cases, with the statement "well you can try to get that advantage, see how it goes!, ill be sitting here increasing my progress while you do that".

A way to get around doing this exact action would be for the AI to do something else that fills the following conditions
- The player can visibly see that the game is becoming harder
- The change needs to be permanent, aka, the player cannot remove it, as, once the threat is removed, we're back to time favouring the player again.

An example - The AI can tech towards Plots which the player can check at any time - with each one changing how the AI behaves in a way that's generally reguarded as harder, such as.
- The AI waits and builds bigger waves before sending them
- The AI generates more buildings over time, (I see you instigator station)
- Guardposts now stock more guards.

So long as the AI cannot run out of plots, this would fill the same job (just make the plots repeatable). It's just AI progress is the simplest, time-wise to code right? ;)

The thing I love about AI war 1, is the fact that, no matter what advantage you manage to get over the AI, no matter how far ahead you feel. You know, the AI can, and will, inevitably still win. Unless you, the player, takes action.

TagsNo tags attached.

Activities

MatthewYCR

Jun 6, 2019 6:32 am

reporter   ~0051854

Note: I haven't played EVERY neutral faction, I was intending to do that first, but I felt like waiting a week to make sure that the OPTIONAL features could not fulfil a feature the MAIN game needs, was a problem in itself.

The neutral faction that increases AIP with time, but reduces it if the player captures it does not have permanence, so I'd argue It does not fulfil this role

Chris_McElligottPark

Jun 6, 2019 9:00 am

administrator   ~0051857

This is something that I've thought a lot about as well, although for the moment I've kind of left it to be something that I deal with later. As I was writing the design doc that you linked, I was thinking about how that does also affect the player's ability to "just sit and wait and use clock time (or fast-forward) to win."

The problem with the AIP auto-increases, which might be something I bring back, we'll see (certainly as an option it doesn't hurt) is that those can be overly stressful to players, who feel like they need to either be rushing around or have the game paused.

When I look at the metal changes as they are outlined in that document, my feeling on that was that at a low level that has no real impact on your ability to "spend wall clock time" to win against the AI in the end, so I decided not to muddy the waters trying to chase that down at the time. But you're very right, it is an issue that needs to be dealt with at some point.

With the AI in AI War 2, it's probably that past a certain point you will lose because of the AI getting reinforcements constantly while you are hunting the warden fleet and so on, and so the balance will eventually shift to the AI even as it stands right now. Is this enough to offset what you're talking about as a problem? I'm not sure. For lower level or new players, probably it is. The main problem is that it isn't communicated to the player. Possibly the "total strength of the AI" needs to be shown somewhere in the UI, fairly prominently, so that players can be aware of the fact that this is creeping up over time.

Hunting down all the hunter and warden fleet members and killing them is fine, but you're going to have allowed for a lot of reinforcements to happen during that time, and potentially you're going to be just feeding them even MORE units as you stir up threat that you don't kill, or as the AI hits caps on its various planets and starts devoting more of its budget to the hunter or warden rather than bolstering plants that are filled up.

So I think that the mechanics you're looking for ARE here at the moment in AI War 2, but the problem is that they may not be strong enough, and not communicated well enough.

The problem that I run into is that I am wary of using the "stick" instead of the carrot. When players feel intense time pressure... that's something that some people like, but a lot of people just get stressed out. I've already talked some (mostly on a podcast) about a missions structure that I want to implement, and having that be optional but guiding and more carrot-ish as a reason to do certain things, sometimes in a certain amount of time, I think would be a win for the average player.

But the game definitely isn't any one thing: different people have different stress tolerances and goals, and so I'm trying to find a baseline that won't freak out most people, but that will keep the advanced players feeling like they need to move. In their case they might be tempted just to waste wall clock time to get an advantage, and that's never good.

MatthewYCR

Jun 6, 2019 9:27 am

reporter   ~0051860

Its half and half, I was aware that might be the case. (ai getting better offscreen, but cannot see it) I thought that may be the case, but yes, it's not communicated at all.

There's about 10 things which I could have said "it might be... etc" and some i placed in the note, but i didnt want to write in like 20 if cases that were in my mind. And ended up placing towards the end that it must be communicated.

But I'll throw a cycle in here to show the problem in detail

Let's say you're a competent player enough that in the early-mid game, the AI has no real chance of killing you. But you're worried about its performance in the late game, which I think is the standard mindset of the average player.

There's a base you want, its got the best units on the map, by a lot, so you want it. Problem: its a Mark IV base, its heavily defended, and you're right at the start and have little to no upgrades. So really, the odds of you destroying even a guardpost in that base is like 1% odds.

But... you can destroy it. You know you can, its just a matter of time.

1. So you try, you fail
2. the AI builds up a reprisal fleet
3. You rebuild your army (navy?), In time to stop the attack
4. Since you failed, theres some more hunters and wardens around, So now you hunt those down
5. Rebuild your fleet again, but this time, AI presence has dropped. down to its starting amount
6. Go back to step 1.

Yes, there's a chance that you'll lose to the AI during this time, but the odds of you losing to early game AI is abysmally low, your 1% chance of slowly breaking down that mark IV base is more likely.

Therefore in the current AI War2 The best play is to keep trying that 1% chance that you'll make some damage

Most players, (99%+) will just bite the bullet of frustration and take a new base. However the better players... will still do so anyway, but they also KNOW that taking the base is the wrong play at a tactical level. It's more that they look at the clock and argue with themself whether they want to spend the next 10 hours trying to crack this base.

If you have the smallest permanent advancement for the AI during this time, the "awfully unlikely" play is thrown into debate. If its that stupidly unlikely, then you can justify that "I could capture this base 5 hours faster, therefore its the better play"

Its unknown whether attacking it would have been the better play or not with AI progress, but at least now you can justify breaking the unlikely stalemate. So the change is mostly psychological.

Note: This is a very common situation in AI war.

Yes, you put a whip on them, but even when i first played AI war 1, i noted how little is was, and felt like I had a lot of freedom at the rate it was increasing at. The freedom to try, and fail. and feel like the game isnt over.

Often you can just say "well, their forces have dropped a bit, there's less turrets, They lost a guard post, hey the warden fleet was there, maybe next time". And still feel like you've made progress. There's so many things to justify continuing despite losing in AI war that you don't normally feel like you've wasted your time.

MatthewYCR

Jun 6, 2019 9:32 am

reporter   ~0051862

And hey, I did say, you we're probably aware of it, but i never saw it in a design doc, and i left it for ages because I was thinking you hadn't gotten around to it. But when you're asking "how the game feels", my answer is.

It feels off because the AI gets stuck, and cant change the status quo

MatthewYCR

Jun 6, 2019 9:37 am

reporter   ~0051866

And ESPECIALLY remember, that despite all attempts to keep the game as Time Neutral as possible. Which is the goal of not having a whip.

There's a 101 ways to change a time neutrality into "time favouring the player".

Like right now. Best Play: Destroy every single guard post possible before taking a base. (assuming weak/no CPA)

Tedious. but it does place you in a better position.

Chris_McElligottPark

Jun 6, 2019 10:28 am

administrator   ~0051867

Oh, none of my response was meant to be critical or really even defensive. I was mainly thinking out loud. It's an issue I've thought about, but I haven't had anyone to really talk to about it directly, so this has been hugely useful of a conversation already. I agree with you that there's a definite problem here. I was simply musing on the nature of it and trying to come up with some ideas in a generalized sense, as well as trying to isolate this discussion from the metal discussion. I think they're both important, but unrelated (based on my own design goals, anyway).

MatthewYCR

Jun 6, 2019 10:38 am

reporter   ~0051868

I'm thinking out loud too, im more hoping that this doesn't seem like an attack myself.

I've posted "What is metal?" While you were replying to this.

At this point, I think i've said all on my mind related to resources.

Chris_McElligottPark

Jun 6, 2019 10:41 am

administrator   ~0051870

None of it feels like an attack, it's all good brianstorming. I definitely appreciate it. I've written up a response to the what is metal one(0021279). so I'll be interested in your thoughts.

Issue History

Date Modified Username Field Change
Jun 6, 2019 6:21 am MatthewYCR New Issue
Jun 6, 2019 6:24 am MatthewYCR Description Updated
Jun 6, 2019 6:27 am MatthewYCR Description Updated
Jun 6, 2019 6:32 am MatthewYCR Description Updated
Jun 6, 2019 6:32 am MatthewYCR Note Added: 0051854
Jun 6, 2019 9:00 am Chris_McElligottPark Note Added: 0051857
Jun 6, 2019 9:00 am Chris_McElligottPark Assigned To => Chris_McElligottPark
Jun 6, 2019 9:00 am Chris_McElligottPark Status new => assigned
Jun 6, 2019 9:27 am MatthewYCR Note Added: 0051860
Jun 6, 2019 9:32 am MatthewYCR Note Added: 0051862
Jun 6, 2019 9:37 am MatthewYCR Note Added: 0051866
Jun 6, 2019 10:28 am Chris_McElligottPark Note Added: 0051867
Jun 6, 2019 10:38 am MatthewYCR Note Added: 0051868
Jun 6, 2019 10:41 am Chris_McElligottPark Note Added: 0051870