Most players who love ESO and the Tamriel world want ESO to be a success, because we want to keep playing the game in the future. This of course is what the game publishers want because they make money from an engaged players base. So the question is why and how to implement subclassing?
It is known from marketing theory and practice that, in general, more product choices provided to a consumer lead to larger market share. This is particularly true in oligopolistic markets, of which MMO publishing is an example. One analogy that is apt is between MMO properties and auto manufacturing (an oligopoly). ESO right now has the equivalent of 7 autos available for selection by players; these are the 7 classes. One might think of there being more choices, but using our auto analogy, those other choices (gear, specific skills chosen, etc.) are like a vehicle's trimline rather than a different product. On the downside, there is some research that shows there are decreasing returns to customer retention as the products choice count increases due to choice "fatigue".
With subclassing, we go from having 7 choices (with various trimlines) to 21!3 =1330 choices (at most), assuming that the main class skill line doesn't change the function of the subclass skill lines. ZOS posted over 3000 choices which they probably calculated as 21 main skill lines (7 classes times 3 skill lines each) times 153 subclass combos (18!2) = 3,213 subclass choices. The ZOS calculation assumes that the first / main skill line chosen changes the way the subclass skill lines function, which probably won't be true. Regardless of which calculation is used, the end result is expected to increase the choices to players. This is a good thing in general; however, the change might not affect all players the same.
Regarding player types, we can think of roughly three types of players in an MMORPG: PVP-ers, end-game PVE-ers, and all others (casuals). The first two types are heavily focused on min-maxing the role they play (tank, heals, and dps). Given that, they are concerned not with choice/build diversity but with finding the one (or a few) builds that maximize their contribution to their team. The contribution depends on their roles (tank = survivability and zone control, heals = HPM, and DPS = DPM). This means that the skill lines chosen are even smaller than the calculation results above because, for example, DPS players won't choose healing skill lines is most cases. Ultimately a new meta for each role will be identified and the PVP and end-game PVE players will settle back into a new normal. The number of builds (products) they consider, their product consideration set in marketing lingo, will be different but remain small. The danger is in the details; if the implementation is too disruptive or buggy for these two player types, then they could leave the game.
The story is likely different for casuals. Because, casuals are less focused on min-maxing but rather play style / novelty, they will suddenly have a huge set of builds (products) to choose from. This is potentially beneficial to the game publisher and developer of ESO. More choices, from our marketing theory, enables the ZOS and Bethesda to capture and retain a higher MMO market share. This is good for all players in the long-run due to the game's continued success. The final game outcome will depend on the balance between any increased customer retention from casuals to lost players in PVP and end-game PVE. One danger is choice fatigue for these casual players. ZOS should look to make it easy for new players to acquire and create interesting builds, perhaps through use of a user-friendly build designer directly in the game -- don't wait for the player community to create this as an addon. (This might also be adopted by the other player types if it's implemented well.)
The success of the new subclass system depends critically on avoiding loss of some players while increasing share in others. On a personal note, I find it odd that the devs are considering requiring 2 skill points for each subclass skill. The 2 point requirement puts an artificial and meaningless barrier to new players getting engaged. In contrast it has almost no impact on existing players, whom likely have plenty of skill points available to buy the subclass skills anyways. Inasmuch as engaging and retaining the new casuals will drive part of the financial success of the new system, the devs should be finding ways to make it easy for new players to begin use the subclasses as soon as possible, rather than creating barriers to use.