Slitherine

I agree that the "player versus environment"-aspect of the game already is fleshed out pretty well and that further focus on this aspect could make it even more interesting in that regard.
I'm not too sure about random-events that specifically hit the leader. Some people play this game quite competative and stuff like that could not be welcomed by everyone.
From a development-point of view I would say: Adding aliens and events that destroy some stuff shouldn't be too hard. As long as they don't need new graphics or popup-texts that is.
But I'd say that at least for events popups pretty much are a necessity.

Xilmi wrote:I agree that the "player versus environment"-aspect of the game already is fleshed out pretty well and that further focus on this aspect could make it even more interesting in that regard.
I'm not too sure about random-events that specifically hit the leader. Some people play this game quite competative and stuff like that could not be welcomed by everyone.
From a development-point of view I would say: Adding aliens and events that destroy some stuff shouldn't be too hard. As long as they don't need new graphics or popup-texts that is.
But I'd say that at least for events popups pretty much are a necessity.

That is a good point. The effect of more major random events in multiplayer could be devastating and very frustrating especially on smaller maps...

Single player "player vs environment" would give it even more replayability, and it already has plenty. Story mode!

Big congratulations, Ail. And a bow of respect for you having chosen to take this up. Only good things can come from this for all parties.

Pandora is a gem of a game and I'm glad to learn it will receive the polishing I have complete confidence you will deliver unto it. Whatever comes of this, I hope you will enjoy your time working on it.

I was a beta tester on this game. I very much liked many of the game mechanics, like the way cities could be specialized and the operations that could be researched -- these things were different from any 4X game I'd played before. (And I've played many, many 4X games, beginning with CIV I; it's my favorite game genre.)

But when Pandora was released I didn't buy it, even with the beta-tester discount I had available. Because as much as I liked the game mechanics, I knew from the beta that the AI was weak. And I just wasn't interested in spending time on yet another 4X game with weak AI.

But today I checked this forum and discovered the AI improvement work that's now going on. I read everything on the forums about it. And then I bought the game and the DLC. Solely because of the work that's being done here. (I'm familiar with Ail from the Endless Space forums, and him being allowed to work with the source code here is the best thing that could have happened with this game.)

There are no words to express how happy I am about this turn of events. (Because I really did always like the game mechanics in this game. And now it seems that the AI will finally live up to the game's potential.) And Proxy has made a new sale, simply by being open to this offer of excellent outside assistance.

Thank you, you two.
These are extremely encouraging words!

All thanks go to you for being willing to put in the time and effort to improve this game.

The fact that in less than two months you've been able to improve the AI as much as you have just crystallizes for me in my mind something I've pondered before on several occasions. Which is this: Why can't the designers of a particular game themselves create an AI which plays the game like a reasonably good Human player? I mean, it's their game -- they designed it.

Now, I know that the economic realities typically get in the way. (I'm a lawyer by trade; I have experience with helping business people deal with economic realities, especially the negative ones.) There's never enough time to get everything finished before that release date comes knocking. But then, even after release, there rarely seems to be any fundamental improvement in the AI of most games, even when the post-release support can be characterized as "good" or even "excellent." Sure, there's often some improvement, but it's rare to unheard of that anyone ends up saying, "That AI played the game kind of the way I would."

But now, in less than two months, you've made improvements here that qualify as "playing the game kind of like a human would."

Clearly, you have significant experience playing the game, and you've put significant thought into how to play it well. But I would anticipate (hope?) that the designers themselves have also done this as they've worked on the game all this time. So why you, rather than them, being the one to finally bring this level of improvement?

I think it boils down to this: you possess a rare cognitive ability to analyze complex systems and interactions and then perceive what's necessary for improvement and how to put it into effect. And I think it's an ability that few people in the game "industry" possess.

(It's kind of like this: I work as a prosecutor now. I do a lot of jury trials. And I work with a lot of other prosecutors who also do a lot of jury trials. Many have as much or more trial experience than I do. But almost none of them know how to effectively communicate their cases to the audience we call a "jury." Because it's not a skill that's dependent solely on past experience. In a fundamental way, a person either has the ability or they don't. If they do, experience will improve their level of ability. But if they don't possess the ability at a fundamental level in the first place, no amount of experience will grant it to them.)

TL;DR: You have a level of ability that few people seem to possess. And I'm ecstatic that you're working on this game, which I've always wished might rise to meet the level of potential that's inherent in the game mechanics. And Proxy is very, very fortunate to have your assistance.

Theodotus wrote: The fact that in less than two months you've been able to improve the AI as much as you have just crystallizes for me in my mind something I've pondered before on several occasions. Which is this: Why can't the designers of a particular game themselves create an AI which plays the game like a reasonably good Human player? I mean, it's their game -- they designed it.

In my opinion it is all a matter of how much time and effort you want and are allowed to put at it.

In order to be motivated to put that effort in, you have to consider it fun. This is extremely important in everything you do. If something is no fun for you, you'll never be dedicated enough to do it as good as your ability allows.
And for me AI-programming is a lot of fun. The challenge to teach a machine how to properly play a game has something very fascinating for me.
Also: There's no real pressure from the outside. I can spend days at experimenting with stuff and then scrap it.
I think the absence of pressure is extremely helpful.

Take tile-improvement-management for example. From all things I've worked with, it's the one that I've spent the most time on. And I'm still not satisfied.
Had I been pressured and told I have 5 days and not more, then a lot of tests could not have been conducted.
It's really tough to take everything there is into consideration.
Last sunday I redid the mechanism how the need for mines/fungi is determined because the AI was planting way too many fungi for my taste.
Fungi-planting looked fine then but I sometimes would see suburbs and purifiers on mountaintops because my new evaluation formula valued the base-mineral value of the mountains in addition to the value of the other improvements.
In order to find out what went wrong, I needed more time and patience.

Both things that I imagine are limited when you are under pressure.
I eventually found out it was a copy&paste error, where I mixed up food and minerals at one point, which meant mineral-need was linked to food need and the more food there was the less eager it was to increase mineral-income.

Theodotus wrote: Clearly, you have significant experience playing the game, and you've put significant thought into how to play it well. But I would anticipate (hope?) that the designers themselves have also done this as they've worked on the game all this time. So why you, rather than them, being the one to finally bring this level of improvement?

We've actually played with the Devs in Multiplayer. They are not bad at the game. No Zak0r level but quite decent.
So I can say that lack of experience and understanding about their game is not the reason.

Theodotus wrote: I think it boils down to this: you possess a rare cognitive ability to analyze complex systems and interactions and then perceive what's necessary for improvement and how to put it into effect. And I think it's an ability that few people in the game "industry" possess.

That is a nice way to put it.
I personally would have said that what I'm good at is inventing algorithms that recreate my behaviour.

Like: What considerations go into choosing how I would distribute my population to the different jobs and how do I get to my result. A lot of this is done subconsciously from experience. So I have to think about what impregnated my subconscious and be able to word it.
Do I need Units? Do I have a lot of unbuilt buildings/advancements? Am I ahead or behind in science? Does this city have special boni? How do I prevent getting too many or too few miners in comparison to workers, when I don't want to use scientists anymore? Should I use mineral-sources that produce only one mineral? If yes, under what circumstances and how many? Should I consider pollution when doing these assignements, if yes how?
Once again it took many iterations to get it where it is now. Checking that the way the thoughts were implemented actually does what the thought intended for any of the imaginable scenarios takes time and patience.

TL;DR: Lack of ability imho is not the main-reason why AI seems such a neglected field in the gaming-industry. The requirements in order to write a good AI besides the obvious ability are:
time, freedom from pressure and most importantly fun in doing it

I blame capitalism. When games are made to make a living and not because you consider doing it as a great way to spend your free-time, you'll never have the time and dedication to make a game as good as you can imagine.

Heyo Xilmi,

I just started playing this game recently, and I signed up just to say I am very much enjoying the improved AI in the beta. I've also watched a lot of your video streams, which have helped me learn more about the various strategic tradeoffs.

So please do keep up the excellent fun you've been having! It is definitely appreciated.

Hello Xilmi.
I have registered in the forum just to say I very much appreciate your efforts and developers' wise decision to enlist you. "...in the long run, my ambition is to actually have the AI come to the same conclusions that I would in each given situation." - I look forward to possible future, where Pandora, being already decent game, can become best AI-wise 4X game on the market. Please keep up!

Made a lengthy-post in the steam-forums about the balancing-issue caused by income-sources that scale with the power of your opponents.

http://steamcommunity.com/app/287580/di ... 159568222/

If you don't have access to write there, you can also discuss this here. I'll check here aswell.

Hey!
Just hopping in to show my support to your project: it’s very nice from you to dedicate time to improve this game, and from the developers to give you the tools allowing you to do it.
Pandora is the only 4X game I’m still playing, alongside Alpha Centauri (I’ve been playing this last one since it came out). And being allowed to hope for further improvements while I originally thought support was ended is a great surprise!

Even if we’re only a few to have posted in this thread, I bet there is a lot of people waiting for your upgrades, knowingly or not

++1

First off, thank you for taking on this project, Xilmi, I really do adore Pandora for its systems and can only see good thins coming from this.

Obviously you have a good handle on the AI and the spying thing is already being talked about being reworked officially, so I thought I'd share my thoughts on a particular system that doesn't seem to get quite as much attention, that being the diplomatic one.

The main problem I can see with the diplomatic game mostly comes down to transparency, in my opinion. We can see rough approximations of what the AI thinks of us with the Cooperative, Generous and so forth descriptors. We can see WHAT they think of us, but especially for new players it can be confusing to figure out WHY they think that way of us, which is not helped with the rather poor introduction the game gives about how diplomacy works. I realize that this aspect might be very difficult to correct, but some way of exposing how your actions relate to the opinions of the AI might go a long way towards making the diplomatic game more approachable.

In a perfect world for me this would mean that on the main game GUI a section is dedicated to giving a light overview of what other met factions think of you with an indicator as to how their opinion about you is trending, as in a green up arrow for positive growth or a red down for negative growth which you would ideally be able to mouse over for a tooltip pop-up of what is currently affecting your opinion trend, also color coded. Going to the diplomacy screen would yield far more hard data about your relationship status, perhaps even to the point of giving hard numbers about how your actions and state of military are affecting their opinion exactly.

That might be overkill or too radical of a change, but I think making it so we can see more of the 'why' the AI changes its opinions about the player will go a long way towards dispelling the feeling that the AI is simply bi-polar and erratic with its decisions.

One last aspect about the current diplomacy game that I would like to see changed is how we are made aware of the opportunity to use diplomatic actions. Currently we can either praise or denounce publicly, and also privately denounce, both of these actions being on a hidden eleven turn cooldown. I've noticed that when I keep track and use these options every opportunity, I can maintain a much better control of how the diplomatic game is played. Because these actions cannot be used freely, I'd consider diplomatic actions a resource of its own, but one that is not very well communicated to the player. What I would like to see is some sort of notification to make the player aware that they can take certain diplomatic actions so that optimal play does not require the player to keep track of the number of turns passed since the last action taken, which can be very tedious.

Anyways, those are my current thoughts, thank you again for all your hard work so far!

Uhm, well, diplomacy is kind of a bane for me.

Since the last "official" beta-release I've been constantly changing how stuff works there.

Right now, in my current development-version, I'm using three different things for diplomacy.

One is the relationship as we see it right now in the game.
But in addition to that there is a method that determines your qualities as a trade partner depending on guesses and once you actually trade with them also on facts.

The last one, is the most evil thing I've coded so far. The AI learned to estimate difficulty-level differences by comparing population and, if available, trade-pact-yield. So if you mix up AIs of different difficulty-levels (by using advanced-game-setup), the higher-level-AIs will consider the lower level as much easier targets than other higher-level AIs...
Doesn't sound so bad this far? Well, from the point of view of an AI, the Player is an Easy-Level AI and thus would be treated accordingly.

This "feature" comes combined with a new rush-preparation-mode, the AI can perform. Which means instead of pretty uncoordinated military-buildup, that eventually gets used for a war, the AI now actively observes its neighbours and when coming to the conclusion that they are weak, they secretly perpare for a war and once they feel confident enough, will declare and attack.

Another thing this means is that AIs that don't feel like there is much gain in war, will play way more economy- and research-focused than they used to do.

Until about a week I could still beat the game on Very-Hard. But now... It seems impossible. Which pretty much was the purpose of the whole excercise anyways. ^^

But I'm not even sure I should leave it this way. I'd actually like some feedback.

So what do other players think:

Is it enough to give the AI a bonus?
Or should it, in addition to that, be made fully aware about the implications from this circumstance?

I'll make a new Let's play to showcase the issue for better evaluation.

Here it is:
https://www.youtube.com/playlist?list=P ... 3A0mncAyUd

After being disappointed with Civ Beyond Earth I also finally just bought Pandora and DLC because of your announcement and this game finally having a good AI because of your work and using the Easter discount! Excited to see your new work! Is your current AI mod updated to the new beta 1.5.5?

There might be some misunderstanding.
The AI mod you can see in the modding-forums is not for use with the 1.5.5+ betas.

I've been granted access to the source-code of the game which gives me possibilities that go way beyond the possibilities of modding.
The 1.5.5 beta that is automatically downloaded is the state of my work at the middle of february.

Unfortunately I cannot persoanally update what will be downloaded via the updater/steam.

So I had to come up with other ways to distribute my further development. Currently you can get it here:
http://www.civforum.de/showthread.php?9 ... ost6629677

I (I'm "AIL" there) post in this thread every couple of days. So scroll all the way down and look for posts which attachments.
The attached archives contain 2 executables which have to be extracted to your Pandora/Binaries folder and overwrite the existing files there.

This way you have the newest version of my changes.

Xilmi wrote:There might be some misunderstanding.
The AI mod you can see in the modding-forums is not for use with the 1.5.5+ betas.

I've been granted access to the source-code of the game which gives me possibilities that go way beyond the possibilities of modding.
The 1.5.5 beta that is automatically downloaded is the state of my work at the middle of february.

Unfortunately I cannot persoanally update what will be downloaded via the updater/steam.

So I had to come up with other ways to distribute my further development. Currently you can get it here:
http://www.civforum.de/showthread.php?9 ... ost6629677

I (I'm "AIL" there) post in this thread every couple of days. So scroll all the way down and look for posts which attachments.
The attached archives contain 2 executables which have to be extracted to your Pandora/Binaries folder and overwrite the existing files there.

This way you have the newest version of my changes.

Ah I understand now thanks! And those versions are newer than the current steam beta?

Yes, they are. Steam beta is from february. The newest of those is from yesterday. ^^