According to the game guide, an Alliance is an obligation to help with War if your ally gets into one.
However, there are some not obvious cases where conflicting interests might blur the action that should be taken.

Currently, the following code parts exist that deal with this :

1. When the Human Player asks the AI ally to declare war on a target, and they are at war with the target themselves, the AI player will always accept the proposal without a roll (they might demand gold at random, and if the player is low on gold this does cause a refusal, simply try again until they accept without demanding gold).
Starting from RC13 and the latest CoM, there is a slight alteration here : if there is an ongoing Peace Treaty between the two AI wizards, the proposal will be refused, to provide consistency with the below effect and prevent abuse by immediately asking this again once the AI players sign peace.

2. When an AI player is allied with someone, and that someone is at war with someone else, they'll declare war on their ally's enemy at a random chance (or break their treaty if there was one).
This cannot happen if there is a Peace Treaty, or another Alliance between the AI player and the target. This code simulates the AI's willingness to help out allies on their own without a request, as well as serving as a replacement for the AI player's request of war declaration between each other. They'll just do so without being asked.
This part works as intended and needs no changes I believe.

3.When an AI player is allied with another AI player who is at war with a third AI player, they'll declare war immediately.
This one was removed because the previous one does the same thing but better, as it works on everyone, and considers peace treaties.

4.When an AI player is allied with two other AI players, and they are at war with each other, both of those AI players will break the Alliance towards the first.
Now this is one which I have large doubts about.
First, this only works if all parties involved are AIs. The human player can maintain an Alliance with an AI who is allied to their enemy, and neither Alliance breaks. Their enemy will demand to break the Alliance when they want to propose a peace treaty, but that's all, and there is no real reason for the player to ever accept that, and if they want the peace, they can propose it on their own.

So basically, the human player has a huge advantage here, as they are allowed to maintain Alliances under a condition the AI player isn't allowed to. Eventually, they can also demand their ally to break the Alliance towards the other AI player, gaining further advantage, while the AI cannot ask the other AI to break the Alliance towards the player, see 2.

I believe either this check need to be completely eliminated, allowing AI players to maintain alliances with two people who are at war simultaneously, OR the AI players should be able to make each other break the Alliance through "2." in this situation which works against the human player as well. (In which case "4." also serves no purpose as they would break the alliance through "2." instead even between only AI players!)

Additionally, this check is actually buggy. It will only work if the player having the two Alliances has the lowest player ID of the three players involved! You can see in the latest Hadriex video that Flandre is maintaining an Alliance with player 1 and 3, despite them being at war, as she is player 2.

Speaking of Break Alliance and Declare War demands....neither considers the army strength of the target, which might or might not be a good thing.
If you are a lot stronger than AI 1 but much weaker than AI 2, you can force AI 1 to break alliance or declare war on AI 2.
This means that
-You can manipulate other weak AIs to help you against the strongest player, fighting him together, which is generally beneficial to everyone in the world, except the strong wizard of course.
-But also means that a weak AI will turn against someone much stronger than themselves, which is obviously bad for them.

I think it should at the very least consider the army strength when asking for a Break Alliance, and not do so if the ally is stronger, but...if the above changes to "2." and "4." are done, this is meaningless as they would do it anyway at a random chance, pretty much making it a 50-50 for breaking their alliance with the player, or the strong AI. So it might be better to leave this unchanged (giving the player a chance to unite the world against a large threat) and go for the changes to "2." and "4." instead.


Edit :
I've decided to remove "4." because it doesn't work correctly anyway.
I've also decided to implement the army size checks for "Break Alliance" and "Declare War" proposals. It's not fair that the player's army is considered, but the target's is not.

The only remaining question :
Should the AI that holds an alliance with two other players (human or not) who are at war with each other
a., randomly break one of the alliances, and eventually proceed to join the war on one side, fighting against the other
b., do nothing at all and wait it out, why get involved?

"b" would be more beneficial to the AI, and to an extent it also helps the human player (enemies won't be able to turn your allies against you), but it's unfair in a sense as being in this situation as the human player will make the AI allies ask you to break your alliances periodically, potentially making you lose both if no action is taken, whereas AI players don't do this towards each other. On the other hand, the player can force the AI ally to break their alliance eventually, if their army strength gets high enough, without the other AI being able to do the same against the player.

"a" would be fair to all sides, but a major game changer, whichever Alliance the AI breaks will have a huge impact on the chances of the human player to win, and would make Alliances a lot less stable than they are now.

So, in short , do we want the AI to get away with maintaining these conflicting alliances simultaneously, or not?

I would like to ask what everyone thinks about Incarnation and the Chosen One currently.
It was heavily modified to be a good support unit (buffing your armies), while it still has decent melee and casting ability.
The spell is quite expensive though, and that price was set assuming the old Chosen unit, which had Magic Immunity and was an unstoppable powerhouse.

Are you satisfied with the current version? If you play Life and have the option, do you summon him/her? Would you dismiss a hero to make room if your slots are full?

Oversized Screenshot Alert!!!

(pictured: AI not winning. It fought back though, and sure didn't go down easy.)

Pandora: First Contact was released in 2013, and got an expansion, Eclipse of Nashira, in 2014. Just in time to be buried under the wave of hype surrounding Beyond Earth. It was similarly heavily inspired by Alpha Centauri, and it's not hard to see those fingerprints on its setting and backstory.

The game was released to mixed success, often being dismissed as a "shallow Civ5 copy", or another "look, I have hexagons TOO" strategy game. And in its initial release, it had numerous problems with balance and a weak AI.

But these have all been fixed. Recognizing the potential of the game (the underlying mechanics are totally solid), a german modder going by the name of 'AIL' decided he wanted to try his hand out at AI programming and was able to make sweet with the Devs and get access to the source code. He now had all the tools he needed to make the AI as smart as possible, and over the course of 2015, the AIL's AI slowly transitioned from the industry-standard incompetence to a lean, mean, killing machine. He taught it how to Fight, he taught it how to Grow, he taught it how to Harness the terrain, and He taught it how to win.

There's an interview with him here: http://explorminate.net/2016/02/08/monda...s-pandora/

A year ago, I wouldn't have recommended Pandora. It would have been a shallow experience where you can futz around aimlessly and still win. Without challenge, there can be no depth. The sad thing about Pandora's story is that its a game with hidden depth. On the surface it looks like a generic scifi civ-clone, and you could play it sort of casually and come out with a win. But that was only because the AI was bad. The core gameplay of Pandora is a solid and subtle system of resource management and economic growth. This is what attracted AIL to the project in the first place, seeing the untapped potential and an opportunity to test his AI-making skills. The AI is good now, and so to win you have to actually play the game, and that's where the fun begins.

At its core, the mechanics of Pandora are a series of simple but interlocking systems (as opposed to the complex and unrelated systems of Civ5). Population units in cities work the terrain to gather Food or Minerals, or work in the cities to make science or to convert minerals into production. Pretty simple, though you do have a two-level resource chain and the inherent dilemma of "do I focus on researching for the future or building things now?"

But there are some costs that go along with your people's work: Pollution and Housing. When your industry pollutes (and each job pollutes differently) beyond what your forests and purifiers can handle it will hurt your city's morale and penalize its production. Same too with housing: if your population grows beyond its bounds, though in general population is power, your city will suffer. You have to carefully balance your industrial growth with your population growth.

The catch is though that you're not actually in direct control of your growth. Population grows independent of food supply (neglecting starvation) and independent of space (so long as morale stays positive) and it flows and migrates from old crowded cities to new empty ones. Managing your population is like tending to a flock, sending it to greener pastures and dividing it so that it doesn't overgraze one area. Your cities can only hold a certain number of people before you have to convert productive terrain into costly maintenance structures which cut your efficiency.

You have to build more cities, but more cities don't give you more people, just more places to keep them, and when new are devoid of infrastructure and become liabilities where your people's potential is wasted without production bonuses, and a small army's worth of minerals must be funneled into them before they bloom into the mature flowers of long-term strength that you need. Pandora takes the age-old question in Civilization of whether to build Tall or Wide and forces you to walk a middle path, following a shifting crest of maximum efficiency as your flock swells.

Similarly the Combat follows a middle ground between "Stacks of Doom" and "Carpet of Doom". There is, rightfully, no limit on how many units can share a tile, but artillery deals splash damage and so a large stack is a vulnerability. You keep your units spread out in the field for safety, but your maneuverability isn't compromised by arbitrary restrictions and there is no limit to the amount of force you can bring to bear on one target. Artillery able to attack with impunity is powerful, but it isn't lethal, is vulnerable to counterattack, and most importantly: the AI has been instructed fully in its proper use and avoidance. This is not a game where you will be killing wave upon wave of suicidal fools until your opponents bleed themselves dry and crumble. The AI is capable of making calculated attacks of opportunity, and will pounce if you show weakness.

Overall the AI is good enough that new players complain that it cheats (it does not), the economic bonuses on higher difficulty levels had to be turned down, and Very Hard difficulty is nearly impossible to win.

I will undermine my point slightly by telling you that the screenshot at the top of this post is from the final victorious turn of a Very Hard game. However, I was playing that game for the purpose of getting the "I Survived Pandora and all I got was this Achievement" award, and neglected to tell AIL about a couple of exploits I was using until I had secured my advantage. Needless to say, these have been patched, some quirks in its Naval Maneuvers have been ironed out, and I don't think I would be able to win on Very-Hard difficulty a second time.

...Can you?

The excellent replacement help file, Plight, needs to be updated for v1.5. However, all the v1.5 changes were made to the stock HELP.LBX file that shipped with v1.31. How can I figure out what's changed in order to copy those changes into Plight?

I thought of a binary diff, but that's going to be megabytes and megabytes of output.

I found the original two threads where kyrub announces and then develops Insecticide. One of them is on the long-defunct Dragonsword forum, and the other is on gog.com's MoM forum. These reveal what kyrub was thinking when he created it, and how people reacted to it.

https://web.archive.org/web/201005051712...sc&start=0

http://www.gog.com/forum/master_of_magic..._wip/page1

Interesting, it's like the chess match between Kasparov and Deep Blue:

http://www.bbc.com/news/35501537

I was pondering the idea of scenario generation, where you could have a lot more variance in possible start of game setups (you or the computer might get special initial benefits, such as additional cities, additional buildings, starting units, starting heroes, starting spells, etc), but ran into a problem that the information on how things like the map tiles work is poorly documented in anything I know. I could probably get around this by running the standard launcher to generate a savegam.9, then intercepting wizards.exe with my own program, which may be the best way around it, but is there another solution?

I'm searching for a replacement for myself for PB27. Position is not totally hopeless but there are others doing better.

If you are interested feel free to jump in my thread in the PB27-subforum, page 1 is the password, log in and take a look. Reporting was sparse in the last months, so reading the thread wouldn't be all that helpful regarding the current state.

(Oh, and please leave a post in my thread that you are taking a look and your login-name, so I know who logged in.)

DaveV guessing for the wrong team