There are quite a few things to address and some of the topics were a little mixed together. The explanations are also rather long-winded so I am going to try and break it up into topical sections rather than direct conversational replies.
DATA MINING
Using the words "real behavior" was probably a mistake, I should have just used the word behavior. Data mining is ridiculously useful for many reasons, but for the sake of brevity it does 2 very important things. One is that it is a census rather than a survey, that is to say it gives you data about every member of the population rather than just a piece of it. Secondly, the information you get is "close to the metal" and therefore generally free of interpretation and selection bias at the point of collection.
I never said community, player interaction, and many other kinds of feedback weren't important. They are very important, but so is data mining. Developers have access to more information because we not only have the same access to publicly available information (forums, websites) but we also have access to other kinds such as the aforementioned data mining, company conducted information gathering, developer feedback, any private communication the public is not privy to, and other avenues.
All that together gives a much more complete picture the state of a game. Even when you compile all of the information the way forward still isn't clear cut, there are various interpretations of it. But it is important to remember that the more information you have the more likely you are to make the correct one.
Data mining is useful for all games, no matter how big or small, no matter their intended goal. The goal is irrelevant to the information you can gather, because the information you gather informs you about how close or far you are from your goal.
The first Vanguard Beta were an epic disaster. People simply didn't play. Mere days after a huge invite I usually couldn't find a full group of players to test my content. The sheer apathy surrounding it was astounding, it wasn't so much that people didn't like it but that most people cared so little about the game they didn't even bother reporting that they didn't like it, they just stopped playing never to return. I of course don't remember the exact numbers, but it was shocking how little time people played, how little they did, and how infrequently they came back... all this is information you get from data mining and it is indicative of a serious problem you might not (although in this specific case you did) learn from just accessing publicly available information.
GOAL SETTING
The important thing to remember about goals is that they are not mechanics or implementations. Goals are more abstract, they deal with feeling and types of decision making not on what is used to achieve these goals. It is a mistake to confuse an implementation with a goal, which is something I see all the damn time not only here but even in professional situations. It is alright though, discussing implementations is generally more involved and more fun than discussing goals, but when analyzing another game it is important to keep the difference in mind.
Here is what I mean using the Vanguard chain/counter system we are discussing as an example. The goal is to create combat that primarily engages the "executive" style decision making found in the pre-frontal cortex rather than say the type of "reactionary" decision making used in shooters that primarily involves the Occipital lobe and motor cortex. This isn't to say that nearly all games don't use both, but I am referring to general concentration for the sake of simplicity.
There can be many, probably an infinite number, of specific implementations that would fulfill this goal. Further constraints such as cost, technical feasibility, UI, etc all narrow the choices down significantly but it is important to note that there are multiple implementations. It is also important to remember that some implementations may not achieve what they set out to.
This is one such case. The game, explicitly, told you what you should do next removing all "executive" decision making completely which is a total failure. Now, given enough time, money, and technical adjustment it was certainly possible to make something that had chains/counters and executive decision making. It was chosen instead to move to an implicit "combo" system rather than an explicit one because that reintroduced at least some decision-making while also being feasible within the constraints we had. Was it the absolute optimal solution? Unlikely. Was it a massive improvement over what we had? Absolutely.
This is the reason developers should discuss goals instead of implementations until they are sufficiently tested to reasonably assume they are final. It is also why developers can be "evasive" when giving answers. They don't want to end up in a situation where they have a real-world implementation compared against everyone's vague, imagined personal version of it because actual implementations rarely measure up. However, keep in mind the people handling the public face are the game are rarely, if ever, realistic professional designers.
ON NICHES
I love niche games. I support them and I think developers should make games for themselves and an audience that it appeals to. But there is something everyone has to keep in mind when talking about a "niche" game, and that is it needs to at least have a reasonable chance of paying for itself. If a "niche" is very small but the game that serves that niche would be very expensive to make then no games will get made for that niche. I tell you this for your own benefit, so that you can manage your expectations.
Now, don't get me wrong, there are ways out of that paradox. Technological improvements, design refinements, and so on... but the game you get would probably fulfill the niche in goal not in implementation, which is why it is so important to recognize the differences.
Great success usually happens when a game is budgeted for a niche, but is so compelling that it expands well beyond those confines. This is what happened with WoW when it was released, it served the EQ niche but for whatever reasons blew the doors wide open. It in essence became its own "genre" that no one could compete with no matter how much money they spent.
