History YMMV / CrusaderKingsIII

** Any buildings that serve primarily to add levies, as well as the entire levies side of vassal obligations. Levies have an extremely low combat value of 10 damage and 10 toughness with no pursuit or screening. While this is almost competitive with some of the less-powerful men-at-arms like skirmishers at their base level, men-at-arms are subject to a wide variety of buffs from buildings, traits, traditions, holy sites, and other modifiers, while levies are subject to few or no buffs anywhere in the game. About all they're good for is swarming tribals and carpet sieging and employing them even for these tasks generally requires a fair amount of micromanagement to keep the stacks appropriately sized to not exceed supply limits. Buildings that provide large numbers of these are giving little military value and make a lot of work and wind up often used only for whatever man-at-arms bonus they give. Levies are also the only part of a vassal's military they send, whether it's a Clan vassal who loves you or a Feudal vassal with maximum contractual obligations, making their military obligations arguably more of a hindrance than a help. This is further made worse by the fact that AI only checks the raw number of troops, so it sees levies and levy-increasing buildings as the best possible choice, neglecting the economic side of things (particularly egregious when AI builds levy-increasing buildings in farmland and flood plains, instead of manoral estates) or enhancing its men-at-arms, making majority of AI armies into {{Paper Tiger}}s.

** Any buildings that serve primarily to add levies, as well as the entire levies side of vassal obligations. Levies have an extremely low combat value of 10 damage and 10 toughness with no pursuit or screening. While this is almost competitive with some of the less-powerful men-at-arms like skirmishers at their base level, men-at-arms are subject to a wide variety of buffs from buildings, traits, traditions, holy sites, and other modifiers, while levies are subject to few or no buffs anywhere in the game. About all they're good for is swarming tribals and carpet sieging and employing them even for these tasks generally requires a fair amount of micromanagement to keep the stacks appropriately sized to not exceed supply limits. Buildings that provide large numbers of these are giving little military value and make a lot of work and wind up often used only for whatever man-at-arms bonus they give. Levies are also the only part of a vassal's military they send, whether it's a Clan vassal who loves you or a Feudal vassal with maximum contractual obligations, making their military obligations arguably more of a hindrance than a help.

** ''II'' had some infuriating mechanics when it comes to your council. Whatever task the councilors were doing was instantly made void and stopped at their death. For example, if you were proselytising in a province for a decade and was almost done, but your chaplain died, it removed all progress and you had to start all over again. Same with fabricating claims. [=CK3=] makes those tasks separate from the people doing them, so upon the death or departure of a councillor, their current task is simply paused until a successor is appointed. Additionally, their tasks are deterministic, rather than MTTH based. Your chaplain ''will'' eventually convert whatever county they're tasked with so long as they're making non-zero progress, and it will take a fixed amount of time, rather than having a certain chance to succeed every month. On top of that, tasks had a cooldown in ''II'', meaning that once sending councilors for a job, even if they've finished it in a single month, you couldn't send them to other task, ''because''. Since it was seen as obtuse (not to mention annoying in Iron Man mode when you mis-clicked), the cooldown was scrapped.

** Fervor, which replaces Moral Authority from ''II''. Unlike Moral Authority, which went up when anyone of your faith won a holy war and down when they lost in what was admittedly a potential UnstableEquilibrium, Fervor goes down on winning '' or'' losing as the attacker in a holy war and increases on losing as the defender in a holy war (winning as the defender in a holy war has no effect on one's own Fervor). As Fervor goes down, outbreaks of heresy become more common and the realm priest's county conversion task becomes less effective. Additionally, passive Fervor growth is negatively modified by the size of a faith. When compounded by the Sinful Priest event (which also naturally fires more often for faiths that have more land and therefore more priests) and the lack of other easily available [=CBs=] for feudal rulers, this commonly results in Catholicism (and to a lesser extent, Ash'ari Sunni Islam) imploding into heresy as its massive size stymies Fervor growth and its numerous rulers burn away its Fervor on numerous individual holy wars with further hits from sinful priests. Also, with little ability to religiously stabilize their realm after a Crusade (a Crusade leads to a ''60'' point Fervor shift between winner and loser), Crusader states have become infamous for swiftly converting to the local religion (or simply being overthrown by religious rebels), leading to several successive crusades for the exact same kingdom to depose the previous winners.

** Due to weights of various innovations, in the Tribal Era the AI is far more eager to focus on the military ones than civilian. This leads to an entire CycleOfHurting: the AI never has the necessary innovations to develop its land, it takes longer to be able to build cities or temples, and if a particular county has empty holdings while a single city and temple already built, it will be filled with castles, since that's all the AI can build. As a result, AI-controlled realms end up being underdeveloped and with a terrible economic base, which has a lasting result on the rest of the game. Gets extremely annoying if the AI mismanages land you plan to conquer, because you won't be able to do much about it once you take control. Buildings you're unsatisfied with can be torn down and replaced, but holdings cannot.

** City holdings went from the most profitable to the most useless type in the game. Unlike previous entries to the series, in [=CK3=] cities specialise in ''development'' now (which is a scaling modifier to a county's income), rather than raw money income. On top of that, cities have fixed, non-negotiable taxation, which has a relatively low baseline and can only be increased by small handful of innovations and lifestyle perks. Furthermore, since the economy was reworked, each holding type generates the exact same amount of money from economic buildings, meaning cities provide the same amount of money as temples when it comes to income. For comparison, castle holdings can be directly controlled by any given ruler regardless of government type (providing 100% of produced taxes and levies) and when given to nobles, they have a negotiable feudal contract, meaning heavier taxation is possible. Temple holdings, depending on religion, can either be held directly like castles, or provide a ''substantial'' 50% tax and 100% levies to the liege if he's on friendly terms with the archbishop - all while also giving piety to whoever controls them. Cities end up being best ignored, beyond the mandatory "1 of each type" in counties that can have more than two holdings.

** There are exactly four innovations to bee-line for in each era: new development cap, new succession law, new economic buildings and increased building slot. The exact order might vary depending on your situation, but those four are the most important and everything else is an afterthought. And if your culture has some specific, more convenient succession law from the start, then that removes the need for researching succession law innovations entirely until Late Medieval's primogeniture (which is no sooner than 1210s).

** ''II'' had some infuriating mechanics when it comes to your council. Whatever task the councilors were doing was instantly made void and stopped at their death. For example, if you were proselytising in a province for a decade and was almost done, but your chaplain died, it removed all progress and you had to start all over again. Same with fabricating claims. [=CK3=] makes those tasks separate from the people doing them, so upon the death or departure of a councillor, their current task is simply paused until a successor is appointed. Additionally, their tasks are deterministic, rather than MTTH based. Your chaplain ''will'' eventually convert whatever county they're tasked with so long as they're making non-zero progress, and it will take a fixed amount of time, rather than having a certain chance to succeed every month.

*** A Reddit user looking into the matter discovered that the game is actually coded so that discovering a love affair can retroactively change a child's parentage from its actual parent to the lover, which is fairly immersion breaking for players wanting to roleplay a family tree. It's especially jarring because said event ignores IncompatibleOrientation - even homosexual or asexual women can be retroactively revealed to have had an affair with another man [[MamasBabyPapasMaybe and passed off a child born of that affair as the player's]]. This bug was mostly fixed by patch 1.1.

** And God forbid you are at the mercy of an AI Cultural Head deciding which innovation is currently picked for active research or if said cultural head has awful Learning - even the best developed land won't help when there is only a 10% chance of any research happening.

** ''II'' had some infuriating mechanics when it comes to your council. Whatever task the councilors were doing was instantly made void and stopped at their death. For example, if you were proselytising in a province for a decade and was almost done, but your chaplain died, it removed all progress and you had to start all over again. Same with fabricating claims. [=CK3=] makes those tasks separate from the people doing them, so upon the death or departure of a councillor, their current task is simply paused until a successor is appointed.