"For the Glory 1.3 Preview #1: The AI" FTG1.3的AI

624604727

Good evening, and happy Presidents Day to the U.S! Speaking of the U.S., we've improved its setup in the 1773 and 1795 scenarios and its associated events quite a bit. But that's not what I'm here to talk about.

Today I want to focus on changes to the AI, and particularly the strategic AI. In the past, the AI's strategy has been bad enough that it needed lots of cheats to survive. That has changed.

For example, you've seen the AI sending missionaries before, but only on occasion, even if it was a rich country with many wrong-religion provinces. That's because the AI had no capability for deciding when to send missionaries, so the EU2 code worked around it by occasionally cheating the AI some money and forcing it to send a missionary. That situation has been remedied; the AI is now quite capable of handling its own missionaries, so the cheat is gone.

AI colonization is also good enough that it is no longer allowed to send six colonists at once to the same province.

You'll also see the AI building more large buildings (manufactories, shipyards, conscription centers, and forts). This is because it can now plan longer term and actually take inflation if it will be beneficial. Since it now doesn't always mint like crazy every time it's in a war, overall inflation levels are actually a little down; it also helps that the AI knows governors are often worth inflation too.

Another long-term investment the AI can now track is technology. What do you do when land tech gets to level 6 or so? You probably rush it so you can reach 9 first and get guns before everyone else. Well, now the AI can look ahead and see which technology is nearing a major boost, and prioritize accordingly.

Here is the raw changelog for the AI:



- Peace AI will now take distance into account when deciding whether to annex or vassalize.
- Peace AI is now a little more likely to force-convert when possible.
- AI should not build quite so many transports.
- AI should now know that it needs to build a fleet if it has two provinces separated by water.
- Fixed bug causing AI to sometimes get stuck when it is trying to invade a landlocked neighbor.
- Large AI countries should be less likely to accept force-vassalization.
- AI countries can no longer resist peace if they have navies but no armies.
- Peace AI is now more likely to demand religious conversion if possible, especially if it is the emperor or the defender of the faith.
- Peace AI will now just take whatever it can if its enemy is completely occupied by a third party, instead of waiting indefinitely for a government collapse.
- Peace AI is now less likely to accept poor isolated provinces in peace deals.
- Peace AI will now accept isolated provinces that are cores if it has a port.
- AI will now attempt to start improving colonies as soon as they reach city status.
- AI will now prioritize its own COTs more, especially those with empty slots.
- AI will no longer attempt to build manufactories if the corresponding techs are maxed out.
- AI can now look ahead to see whether it should rush a particular tech.
- AI is now less likely to build tax collectors in wrong-religion and wrong-culture provinces.
- AI is now much more likely to build governors when it has high inflation.
- AI is now less likely to build shipyards unless it is colonizing and doesn't have one yet.
- AI now attempts to build a higher ratio of cavalry when it is in plains.
- AI should now be a little smarter at deciding whether to accept and repay loans from other countries.
- AI vassals will no longer grant military access to countries at war with their overlord.
- Peace AI now tries not to demand any of its core provinces which are occupied by allies.
- AI now considers canceling vassalization even at good relations after a long enough time.
- AI should now try to restart its colonization efforts after it makes peace, even without a save/reload cycle.
- AI should now try to put its fleets into port when they are not in use.
- AI will now keep troops and ships on half pay when possible.
- AI should be a little smarter about minting.
- AI should now prioritize naval tech a little more when in colonization mode.
- AI should now actively try to clear out pirates.
- AI should be a little smarter about assaulting.
- AI should be better at moving troops to the point of attack.
- AI should dogpile sieges less.
- AI will now prioritize getting military access through countries that border both it and its target.
- Removed AI cheat allowing multiple colonists to be sent to a province at the same time.
- Added a small culture factor and increased the weight of country size when the AI is considering whether to accept vassalization or annexation.
- AI will now be slightly more happy to declare war if it's in a good position.
- AI will now be somewhat more warmongering if it is already over its force limit.
- AI will now prioritize land tech a little more if it's far behind the leaders.
- Tweaked AI build preferences in favor of governors over chief judges.
- Streamlined AI choice of manufactories and prioritized them a little higher.
- AI is now more aggressive towards pagans if it has a CB.
- AI is now capable of sending missionaries without cheating.
- Tweaked AI investment slider logic again.
- AI should no longer be eager to blockade rebels in another country's province.
- AI will no longer build troops over its support limit while at peace.
- AI is now less likely to change religion if it has recently changed (e.g. by event).
- Removed AI bonus for asking other AIs to become vassals.
- Larger AI countries can now consider more than one construction (building, fortress, missionary) at the same time, meaning they are less likely to get "stuck".
- Landlocked AIs will no longer build naval equipment manufactories.
- Rich AIs are now more likely to consider building shipyards and conscription centers.
- Reduced AI war exhaustion bonuses.
- Large AI countries should now be a bit better about building upgrades.
- Small coastal AI countries will now limit their fleet size based on wealth, rather than just on number of provinces (looking at you, Venice).
- AI should now be better about building up its forces before a war instead of during it.
- Tweaked AI manufactory location and prioritization strategy a bit.
- AI will no longer mint so blindly during wars.
- AI now knows how to mint for important projects.
- AI is now more likely to offer peace if it has multiple revolts.
- Large AI countries are now more likely to build manufactories.
- AI is now more likely to fortify unfortified cities, possibly even taking inflation if necessary.
- AI will no longer help allies against rebels if those rebels are likely to defect to it.

猫猪蜥

留着口水的看不懂

cattell

翻译不难，但在上班的嘈杂环境里翻太累了，真吵
今年要陪老婆考英语，权当练手

cattell

晚上好，并祝美利坚合众国总统节快乐！说到合众国，我们已经改进了他在1773和1795剧本中的设置以及诸多的相关事件。不过这不是我今天要说的重点。
今天我们要谈的焦点是AI的改进，尤其是AI的战略规划。过去AI的战略性简直糟透了以至于必须做大量的弊才能活下去。这种破事现在不会再有了。
例如，你以前大概很少见过AI派出传教士，就算它是一个富国并且有很多的异教省份。那是因为AI没有相应的能力来决定何时派出传教士，因此EU2原版的设计是每隔一段时间就给AI一些火星钱并且强迫它派1个传教士。这种状况现已改观；AI现在有能力调配它拥有的传教士，所以作弊的事一去不复返了。
AI的殖民战略也得到了充足的改进，这样也不用（像过去一样）允许AI同时派6个殖民者到同一个省。
你们还将看到AI建设许多大型建筑（制造厂，船坞，征召中心以及要塞）。这是因为AI现在能够制定长远计划，并且会采取铸币策略，如果它预期将会由此获利。此外，AI不会每次战争都疯狂的铸币，因此夸张的通货膨胀出现的次数实际上会减少；AI还会懂得为了造总督而拉通胀是非常值得的。
另外一个AI能够长远规划的的投资项目是科技。当你的陆军科技到达6级左右的时候你会怎么办？你很可能会全力冲击9级以便尽可能第一个拿上枪杆子。呵呵，现在AI也会朝前看，并且估计哪个科技类已经接近一个重要的节点，并因此优先发展。
以下是一个粗略的AI改进日志：

要达成和平的AI会将距离远近列入吞并还是附庸时的考虑因素。
要达成和平的AI在条件允许时会略微倾向于强制转换（信仰）。
AI不再会建造超多的运输船。
AI现在知道他需要建造一支舰队，如果他有两个省份被水域隔开。
修正了当AI试图入侵一个内陆邻国时可能“犯傻”的BUG。
AI控制的大国更不容易被强制附庸了。
AI在只有海军而陆军已经全灭的情况下不再会拒绝和平提议。
要达成和平的AI在条件允许时更加倾向于要求宗教转换，尤其当它是皇帝或信仰守护者时。
要达成和平的AI在敌国被第三国完全占领的情况下会知道能拿多少算多少，而不是无限期的傻等直到政府崩溃。
要达成和平的AI现在更不愿在和平提议里接受一块贫穷的飞地。
要达成和平的AI现在会接受一块有核心的沿海飞地。
AI现在会在殖民地成为城市之后发展他的建设。
AI现在会优先拓展自己拥有的COT中的贸易，尤其是那些有空位的。
AI不再会试图建设制造厂，除非相应科技已经达到。
AI现在会判断他是否需要冲某个重要科技。
AI现在不大会在异教且异文化地区派遣收税官
AI在高通胀的情况下会更愿意派遣总督
AI现在不太会建造船坞，除非他正在殖民，而且一个船坞也没有
如果地处平原，AI会建设更高比例的骑兵部队。
AI在决定是否从他国接受或偿还贷款时会显得稍微聪明点。
AI附庸不再给予那些同宗主国作战的敌国通行权。
要达成和平的AI不再会要求那些被盟国占领的自己的核心地。
即使处在良好的关系下，如果经历了足够长久的时间，AI也会考虑取消附庸关系。
AI现在在达成和平之后会自动重新进入殖民模式，甚至连SL都不用
AI现在懂得平时没用的时候把舰队停在港口里。
AI会把陆海军的维持费拉在一半，只要情况允许。
AI在铸币上会稍微聪明点。
AI在殖民模式时会略微注重海军科技。
AI会消灭海盗。
AI在进攻上会稍微聪明点。
AI在指挥部队到战场的过程中会稍微聪明点。
AI围城的部队会稍微控制数量。
AI现在懂得向那些通往敌国的邻国借道。
移除了AI可以向同一个省在同一时刻派出复数殖民者的作弊。
在AI考虑吞并还是还是接受附庸时，增加了些微文化因素的考量以及提高了国家大小（在AI比较中）的权重。
AI现在更乐意在国力不错的时候宣战。
AI在超过陆军维持上限之后会变得更加好战。
AI在陆军科技落后许多的情况下会着重这一行科技。
调整了AI的建造优先序列，使其更注重总督而非主审法官。
优化了AI对于工厂种类的选择并使其略微优先建造。
AI在有宣战理由的情况下对原始国家会更加凶悍。
AI现在不用作弊也能自己传教。
再次调整了AI的控制投资滑条的逻辑。
AI不再会急不可耐地要把别国的叛军封锁在境外
AI不再会在和平时期建设超过补给上限的军队。
AI会略微降低改教的概率，如果最近刚改过的话。（例如由于事件）
移除了AI在要求其他AI国家成为附庸时的外交奖励
AI大国现在能够同时考虑一个以上的发展项目（建筑，要塞，传教），这意味着他们不再那么容易“卡死”
内陆AI不再会建造海军装备厂
富裕AI现在会更加喜欢建造船坞和征召中心
降低了AI的厌战方面的奖励
AI大国在发展高级建筑上会表现更好
拥有少量出海口的AI国家现在会基于其财力规划自己的舰队规模，而不是基于其省份总数（说你呢，威尼斯）
AI现在懂得战前造兵而不是边打边造
略微调整了AI建造制造厂的选址和优先战略
AI不再会在战争期间盲目铸币
AI现在懂得为重要工程铸币
AI在国内叛乱四起的情况下会更倾向于求和
AI大国现在会更喜欢建设制造厂
AI会更倾向于为不设防城市建造要塞，如果有必要的话还会拉铸币
AI现在不再会帮助盟友消灭叛军，假如那些叛军的目的是让省份投靠他自己

twy_2000

这东西要出新版了？

猫猪蜥

cattell 发表于 2013-2-20 16:43
晚上好，并祝美利坚合众国总统节快乐！说到合众国，我们已经改进了他在1773和1795剧本中的设置以及诸多的相 ...

感谢翻译神！