Service FAQ: how to correctly answer the question?

Good day, dear Habraspolzovateli!

Some time ago we began to develop a system of questions and answers (two little-known companies have already created by analogy: mail.ru answers or Google answers ). And we, among other things, set the task to get rid of all the shortcomings and problems that we identified in existing similar systems.

One of these problems is the “foruminess” of receiving answers, that is: the whole discussion of the issue is similar to the topic on the forum. One list includes the answers from different users, and additional clarifications addressed to the author of the question, and generally any messages indirectly related to the original question. This leads to “littering” of the answers, which complicates perception and reduces the quality of the service.

Alternatively, we can offer the following scheme. The information entered by users is divided into three types:
')

1. Answers to the initial question (main and main type)
2. Clarifying questions to the author of the initial question (and answers to them)
3. Messages that are not directly related to the question, but which often arise when discussing a topic (chat).

In addition, each of these types has its own natural features. For example, answers are rated by users, clarifying questions form an additional context of the question; in the chat you can discuss indirectly related to the subject of things, for example, some specific nuances or related issues.

We were faced with a question: which response structure to choose?

If we take a more complex data structure, then we get rid of littering the answers, we get more structured information, which allows us to index the material more correctly.

However, this complicates the form of the answer. Instead of one field and the "answer" button, we have three such blocks.

The situation is illustrated by these screenshots:

"Simple" approach:

"Difficult" approach:

Does it make the introduction of these additional entities more difficult for the user (clarifying questions and chat)?

Perhaps, with a non-capped eye, are improvements in usability seen?