Sandbox-Mode:
This time, before I buy the game, I decided to test a sandbox match at a friend who already get the pre-ordering. I am disappointed. Otherwise I also didn't expect much more, because I thought already that (sadly) there won't be as much changes on AI improvement as on game features. It's a pity, because I think in
W1 suggestions threads AI was a often guessed point for improvement, with improvement ideas in detail. By myself I had written a big document and send it to paradox through PNs with suggestions for improving AI, but I didn't see real results in W2. Bummer, I was so exited for W2. :-(
I like the new features, like the new spells, even like the "unrest" feature, and even like the city limiting system. But why I need all this, when I am be able to run through AI after turn 60, like a knife through butter?
AI is not capable to get a strong army till turn 150 and even not later. AI sacrifice strong units all game long. On size M Maps AI never get to his city limit in acceptable time. More turns, and stronger the player mage gets, but the AI mage stays on same level and sacrifice units on world exploring against monsters/beasts.
I am not playing Warlock to build a bit around, I like to use good strategy and tactic (without using cheat / exploits / bugs) to win the game. And why I should do more exploring and researching, when I am already multiple times more powerful as my opponent? When I wait longer time, it will only be more easily to overwhelm the opponents.
It make no sense to explore portal worlds! Before getting cleared the way to portals, and clear the portal, it's easier to kill the AI mages. That's really, really boring, the game has a lot of feature nobody needs to kill the AI, and the bigger problem, so much feature AI-mage can't handle, can't counter, and also can't use by it-self.
I played on map size M, Challenging, great land (setup similar Darkwing88 suggested). Only "Challenging" because I also don't like the AI to get damage bonuses on impossible, it's like cheating to me. But I didn't think on impossible it would be much harder, because there are so much bad thing AI made:
- AI sacrifices really a lot of units against monster/beasts by exploring the world. Sacrificing "trash-units" wouldn't be a problem, better sacrifice some units and get treasures as play to safely. But AI sacrifice all strong units and his lords all game long. Especially the strong beasts/monsters in the world which W2 has new, AI have very big problems, and sacrifice a lot of units even after hundred turns. And AI don't learn to avoid these strong beasts/monsters and wait to get get in their sight range until they can beat the strong beasts. AI don't understand to avoid a way on which for example a "Greater Fire Elemental" has his place. AI will sacrifice a lot of times his units on the way past the greater fire elemental place, also strong and expensive units (like mages) AI will sacrifice and loose them by one-shoot, and that's although the AI nows already the position and sight radius of these beasts/monsters, because have already seen it. Same for other strong beasts/monsters. This sacrificing is at least very, very bad for strong and unique/rare units, they must be saved. And AI have to learn how to explore and gain more room on the map, so it have to learn to combat the strong monsters/beasts on the map without sacrificing most units (wait until it has units to beat the strong beasts/monsters, then attack and kill), so AI can expand to better places in the map.
Especially the really strong units around portals and on other places where the strongest beasts/monsters are placed, the AI mages would sacrifice all off it's units all game long.
- AI don't know when it's time to begin war, because an opponent mage gets too strong. If a mage gets too strong, much stronger then other mages, AI mages have to attack him, because they have to stop the winning mage from winning the game. At the moment, if a player is stronger as a mage, the AI mages want to trade and want peace all the time. So the player gets stronger and stronger till he is able to slaughter all AI mages down without losses. In the meantime AI mages don't get stronger and stay on early game level. And that's a bit issue in diplomacy. A player would trade more with weaker mages, not with the already strongest, and try to use diplomacy to win the game.
- Too slow expansion of AI mages. Also AI mages must expand fast and work to the city limits. And then get new spells too increase city limits. For example on size M, the player can easily have full map control with full-city-limit cities plus free cities and AI still remains with a minimum on cities. AI should be as fast as a good player on Challenging and Impossible difficultly. When it has cleared land from beasts and their is room for a city, then build a settler.
- AI don't make an alliance to stop much stronger player from winning and destroying all opponent mages. This would be very important for Challenging and Impossible Difficulty. It would be much more a Challenge for the player, when AI want to win the game. To win the game, the strongest mage should be one of the the biggest opponent (whatever it is a human mage or AI mage). More bigger opponent is only a mage which attacks already. So AI should do it, like a player would do, make war in an alliance against the strongest mage, when the strongest mage gets much stronger as the other mages in the game. An other sign to make war and not peace with a mage is, when an mage begins to overwhelm other mages. Because every AI mage must know, like any player should know, that when this powerful mage overwhelm other mages, he is one of the next mages to get destroyed and the overwhelming mage will win. Only when all mages are about the same level, then it's a better strategy to make war with the nearest and weakest AI mage. And same when the AI mage self is the strongest mage in the game, then better make war with the nearest and weakest AI mage. But when one mage gets much more stronger as all other mage and begin to destroy them, every mage must know to fight against him, to have at least a little chance to win the game.
Another important benefit of this behavior would be, that mid and late game is not so boring. At the moment, when a player has destroyed one mage and has his cities, his resources and his capital, a few turns later he is strong enough to battle all at one. And one mage alone, is not a really opponent and every mage can slaughtered down one after another. But when AI behave like a player who wants to win the game, then the mages have to know that they have to act against the much bigger mage together. And as a player to have a battle with mages in alliance, is much more of a challenge.
For me that would be my major AI improvement. If AI mages make an alliance against the strong mage, when one mage gets too strong (mostly the player, but could be a AI mage), the game would be much, much, much more exited and challenging through mid and late game. And it's really realistic and good behavior, every player who wants to win would act like that.
- AI places ships in mini lakes, and too much ships at all
- AI don't know when it's better to retreat from combat and when to attack, so it sacrifices a lot of units in battle with player and beasts. Attack when their is good opportunity to win something, for example kill beasts/monsters, make war and take down some cities of other mages. Especially against Monsters/beasts, the outcome could be good calculated or at least fast estimated (player do the same). Even when AI gets attacked, for example a bunch of it's units which are cut of the rest of mages units, then it simply to calculate/estimate, if it's better to retreat and move to direction of where more units are placed. Same when beasts or player attack with units with special buffs, like Vampiric Aura. Never attack a unit with Vampiric Aura, when the unit is not doing any damage in the end, and only sacrificing a lot HP by itself. Same for immunity, sometimes it's better to retreat against a unit with big immunities and save own units back to own territory, and in mean time build something which can harm the strong unit.
- NEW: Some rules for attacking and "war beginning" in phases of "Alliance" or "Non-aggression pact" would be nice. Attacking in these states should be punished really hard. These mage (whatever player or AI-mage) which are doing that, should be flagged as enemy with high priority by any other mage on the map. Even by mages which are already in a alliance or mages which are already in "Non-aggression pact", they should quit this and begin war immediately. Because all know, this mage is a traitor and he want to kill other mages without mercy. The rule have to be supported by ingame warnings, so a message must appear ingame, and the player has to actively agree to do such an action, so that's impossible that this happens by a clicking mistake. The rules should also include something like "Also after quitting an Alliance and attack in same or next rounds (2-5), the mage is punished hard". The same for "a few turns after quitting, with a bit lighter punishment". And same for "Non-aggression pact with lighter punishment". So punishment should be related to state of mages relation actually and how the relation was last 2-5 turns before. All these rules must be supported with ingame messages, when a player want to attack before he can attack without this punishment.
- I guess that AI don't understand what mage perks they have from start-up and how they use it to build up faster
- AI don't understand what is the weakness of other mages, other mages units and monster units. It's a really important to attack units on it's weaknesses. And no big intelligence is needed for example for "focus on damage type a attacked unit has no big resistance in". And kind of "if enemy has much flying units, I need more shooting units". In most cases this is not so important, but on units which have very big resistance in most damage types. Or units which are only vulnerable at some damage types, it will be very important.
- AI don't understand only to attack, when they have at least some chance to win the combat against some units (or single units). Sometimes it's better to retread, save units and attack later. This a very important basic strategy AI need. Also for exploring, only attack beasts you see by exploring, when you can beat them. I saw AI sacrificing a lot of units by attacking really strong monsters/beasts.
- AI don't use units effectively and safely, but this would be very important for high tier units e.g. lords, temple units and every-time for the strongest units in the own army, even if this are only 2 tier units. But it's specially important for early high tier units like lords, they have to be leveled up safely. It is must have for a challenging mid- and late-game, because player gets stronger and stronger, and will have leveled units in higher turns (mid game) and high tier units which are leveled and buffed after turn 80-120 (late game). AI need also strong and leveled/buffed units, so AI mages must learn to play some units safely.
- NEW predefined unbreakable alliances: "Predefined unbreakable alliances" are not a AI-Problem, but simply a missing feature to improve Game Difficulty.The possibility to form predefined alliances even in single player, and predefine them in teams the player want to, would be great. So player can define predefined alliances for "AI-Mage-teams" and for "player-teams" and should be also possible for "AI-Player-mixed teams". When you make war with that predefined alliances, you should have war with all directly. "predefined alliance teams"-functions should find their way to single player mode in a expanded form of team generation, that would be good for the full game and also good for AI difficultly, because player can define as example 3 teams: " 1 player" vs. "2 AI mages" vs. "2 AI mages". So every war the player begin is at least with 2 AI mages in parallel. The Team generation should be totally variable. Predefined Alliances between AI and player in a team should be also possible.
Important for that: AI mages must be able to combat in an alliance with an other mage and understand to help and to strike together against attacked mage. AI mages who are in an alliance work together in a at least basically tactical way. If not, the alliance of AI mages would be useless.
At turn 60 only 1 opponent AI mage got some good units and some cities. Yeah, this mage also kills another AI mage, which was really cool situation because in W1 this happens very rarely and a positive point for AI improvement. But because I have so much time, and nobody seems to attack me, I got a really strong empire, so it was time to begin first war. And the mage who have battled the other mage and destroyed him, has sacrificed a lot of units for his win.
So I begin war with the strongest mage and at round 80-85, half of the strongest mage cities was mine. And at this point I could easily battle all mages together, because the other 2 mages were really (really!) weak, I don't know what they are doing all the time, they got nearly nothing. Battling "monster" was more kind of challenging as to attack the opponent mage. The whole game was much too easy, never feel kind of pressure from opponents. Every opponent mage want be in alliance with me, and share some spells, it was perfect to build up.
Sure I played a good tactic. Go with Agrela and so buffs population growth and food production. But on the other hand it was my first game in W2 and also first game with svarts. So a second time it would be surely more easily.
And for me the sandbox mode is the interesting one, because I want a competitive match-up, not a scripted campaign. And I doesn't' see any real improvement for AI in W2 which will counter much. At my game I had high tier units without lots of buffs, but I am very sure that AI still can't handle single "big buffed" units which have high resistance in everything and spell buffs like invisibility and vampiric aura (same for units with immunities). At the moment I think it will be as in W1, that you can overwhelm the whole map and all mages with just 1-3 units of this kind, even I haven't tested it yet.
And I am very sure AI still can't handle when player use terra-forming spells like lower land and raise land tactically. And so much other things. There are really a bunch of improvements the AI need, and in W1 forums it were discussed a lot of about it, and even some player helped out with AI improvements suggestions in detail. For me it's a shame that AI is as bad as it is.
So without testing, I think the game has still the same problems as W1.
- AI can't handle single "big buffed" units which have high resistance in everything and spell buffs like invisibility and "vampire weapon enchantment". This units need a special tactics, otherwise single units destroying all mages in every match. And even the best challenging AI Diplomacy improvements, don't work well, when the player is able to destroy all enemy units with single units. One strategy is better using of nullify, and perhaps make nullify for player and AI mages available earlier in spell tree. But even then, AI have to learn to favor to attack on damage types which resistance are low.
- Same for units with some immunities, like the ghost lord or the rabbit lord (don't know if they exist in W2). More important when the few weak resistance they have in beginning, are buffed with perks, spells and armory.
- AI can't handle terra-forming spells like lower land and raise land tactically, and same for other spells. (Simple solution would be: drop all extrem terra-forming spells out of the game. Raising land should only raise up to hills. And lower lands only till flat land. No spell should be able to generate mountains or water or Volcano, or it will hard to implement algorithm that AI can't handle these terra-forming spells.)
- most late game features in the game the player don't need to win, and the AI never will use it (same for portal worlds)
- In W1 AI can't handle invisible units. AI don't understand that all of his units see the unit, when one of the factions unit stays in a adjacent hex. And AI don't understand that sometimes it have to search for the invisible unit, if it is a ranged unit and shoot on and kill own units.
- AI need better algorithm of using unit abilities and mage spells. Unit abilities are really important. Svarts bows have crows which are very good for safely exploring and planning exploring. I am very sure AI can't use this kind of abilities. But also all other unit abilities link stunning, healing, buffing, attacking needs improvements.
- Perhaps AI can't use portal worlds as benefit, but at the moment I think, nobody (player or AI mage) will have a real benefit of portal worlds ever, because the Game is designed that it will be more easily to kill a AI mage as to battle to a portal and through the portal world. Here is a bit of general game design change needed, the portal worlds should make a real benefit to a game-play.
- AI is not able to cross water safely (solution is to play "great land" style, but also there are some little sea, perhaps make less water on great land). Because Warlock has available not only great land, but also other map styles with much more water, AI have to learn when and how to cross waters. Because player don't want to play only great land. (Or wait for land rising). Or for what purpose are the other land type implemented in the game, when AI can't handle it?
- AI has to learn the benefit of fortress, because they are strong and cheap. And to learn to have map control with more and faster special cities.
Guys, I really think its way too early to judge the AI.
Its pre release, still about 3 weeks to go until release date, and we probably wont see patches until then which actually could already include small AI fixes,specifically when talking about Exiled Mode.
Sadly I don't think so. A good AI is a bigger task as to implement some new spells or something like that. The algorithms to improve the AI related things I have listed list above, will need lots of work in programming and and even more in testing the product. If they wanted to improve the AI much, I think it would be improved much more at this time and we could talk about improvements of single AI details, like handle/counter specific spells. Or handle/counter single tactics. But at the moment we have to talk about basic AI strategies and tactics, which is not only bad, but it's not challenging good player. I don't think their will be a big AI update which improves all the AI problems (but I hope
).
Here is an other Thread about the bad AI, very critical written:
http://forum.paradoxplaza.com/forum...xiled-mode&p=17113476&viewfull=1#post17113476
A second good thread which describes how bad and boring AI is:
http://forum.paradoxplaza.com/forum...allenging-multiplayer-And-here-is-my-thoughts.
New thread also about AI:
http://forum.paradoxplaza.com/forum/showthread.php?765693-Feedback
Another Thread about bad AI, more about Diplomacy:
http://forum.paradoxplaza.com/forum/showthread.php?766611-Diplomacy-AI-and-YOU!
A bigger starting Posting about most AI issues:
http://forum.paradoxplaza.com/forum/showthread.php?766900-The-AI-Hoping-for-a-miracle-patch
AI still can't handle water/enchanted units - issues carried over from WL1
http://forum.paradoxplaza.com/forum...-enchanted-units-issues-carried-over-from-WL1
Another thread which explained how bad AI is and that there is no real improvement to W1
http://forum.paradoxplaza.com/forum/showthread.php?772304-Warlock-2-a-mediocre-game.
Another thread about AI, and when more fixes will come:
http://forum.paradoxplaza.com/forum/showthread.php?773268-Devs-what-are-your-plans-for-the-AI
Thread about AI related to Outplanes Mod, even with some AI tipps in detail.
http://forum.paradoxplaza.com/forum...shed-Impossible-Battle-for-Outplanes-feedback
Thread about making alliance with AI and watch how bad it is, and some more postings and AI discussion:
http://forum.paradoxplaza.com/forum/showthread.php?774308-AI-non-existent
Thread about new features to the game, and ingame shop, before getting rid of basic problems and AI problems. And some more posting about AI problems.
http://forum.paradoxplaza.com/forum/showthread.php?776480-Devs-are-now-taking-the-p***
My Suggestion Thread for W2 in times of W1, and at all best suggestion thread with most important things player needed more then new features:
http://forum.paradoxplaza.com/forum...-must-have&p=16314697&viewfull=1#post16314697