|The subject of this article was removed from World of Warcraft in patch 5.0.4.
Talent trees were the implementation of talents that existed prior to patch 5.0.4. Active and passive talents were arranged in three separate 'trees', with access to progressively "higher" or "deeper" tiers of talents granted by spending sufficient talent points in that tree. Talent trees were the predecessors to class specializations, predating the introduction of actual specializations by many years, with a player's choice of talents often determining which combat role they were fit to perform. Talent trees acted to limit access to talents by the character's level; by where they had chosen to place their previous talent points; and later by specialization.
Talent trees created a focused, limiting relationship with talents, with some of the most powerful talents placed deep within their trees, meaning only players who chose to spend the vast majority of their points in that tree were able to acquire them. Conversely, some talents that were useful to players of all kinds were made available on the lower tiers. This can be seen to have forced players to focus on their chosen priorities, with every choice of talent affecting all future options, and every unlocked tier in one tree meaning a lost tier of talents in another. Some talents also required the choice of specific talents earlier in the tree, and most talents had multiple ranks, with players often having to spend 5 points in a talent to gain its total effect. The majority of all talents were passive, with the few active talents in each tree usually powerful and desirable abilities.
Because talent choice was therefore cumulative and limiting, the term 'talent build' or simply 'build' was used to describe the total combination and emphasis of a character's talent choices (and to a lesser degree other enhancements). A player's spread of points across their trees was expressed numerically, such as (10/0/61) for a tanking warrior specced deep into the Protection tree, or (0/37/34) for a PvP druid balancing healing with survival. Players could choose to spec deep into a certain tree, focusing on improving their abilities in that particular area; or to balance that focus with varying amounts of talents from other trees, perhaps gaining crucial secondary-role abilities and avoiding over-specializing. Some players chose to create true hybrid builds, such as mages combining Frost and Fire talents to create an Elementalist build. With a maximum of 71 talent points in Wrath of the Lich King and more than 80 talents for each class, the many ranks available for most talents presented some classes with more than 230 possible places to spend their points, and myriad ways of building their character's spec. Placement of talent points within each tree's 11 tiers was a substantially complex matter, with numerous online calculators and recommended builds, and constant debate over the best combinations.
A common approach was for players to spend enough points in a tree to reach the top talent, usually an extremely powerful ability that was therefore only available to players who had specialized fully in that tree, before spending their remaining points in the often more desirable talents of another tree. Many talents were considered to be 'filler' talents, granting little benefit but being placed in such a position that taking them was required in order to gain access to higher tiers or other particularly desirable talents. Choice of talents was often focused upon access to certain key abilities, with builds designed to hit these talents while spending as little as possible on less desirable options. Once these key points were decided upon, choice of build often became a matter of which of the less desirable options were preferred (or at least which were the least useless), since a certain number of them had to be chosen.
With the introduction of specializations in Patch 4.0.1, the talent trees were for the first time locked, with initially only the tree corresponding to a character's chosen specialization being available for selection. The trees were also simplified, with many talents removed or reduced in number of ranks, and the total number of talent points reduced to 41. By spending 31 points in their 'primary tree' players were able to unlock their other trees, and could then spend their remaining 10 points as they wished. However, because of the progressive nature of the trees, this limited players to relatively insignificant talents from their 'secondary trees', with players having enough points to access the first tier of both of their secondary trees, but the second tier of only one of their secondary trees. This was intended, serving to allow Blizzard to focus on creating three separate and succinct specs for each class, with the aim of improving balance and deepening and refining the character and playstyle of each spec, and was also argued by some to make the previously overly-complex talent trees far easier for new players to get to grips with.
With Mists of Pandaria talent trees were removed entirely, with the number of talents vastly reduced and placed in six tiers of three talents each. Access to talents was no longer limited by specialization, although since many talents had been changed into specialization-specific abilities, arguably this was of less significance than it might have seemed. In effect, talent trees live on in the form of the active and passive abilities unique to characters of a given spec, the only difference being that the player need not spend points to acquire them.
While the direction of progression on talent trees was perfectly clear - beginning at the top and progressing "deeper" into the tree - many have remarked on the possibility for confusion, with higher tiers in fact displayed progressively lower in the display, and the talents at the top of the tree being found at the bottom of the screen. Since the form of the 'talent tree' analogy is in fact the reverse of that found in real trees, it has been jokingly suggested that "talent roots" may actually be a more accurate term.