Many Mathematica functions can use several methods to accomplish the same task. Sometimes the documentation has references to the papers these methods are based on, sometimes it doesn't. I found that when I contact Wolfram support for more information on the methods, they will usually send me the references.
We have some questions asking for the same kind of information:
This question is closed with the reason that people here can't answer it.
I would like to suggest not to close such questions in the future.
While it is likely that we can't answer it, Wolfram support can, and likely will. The OP might eventually end up contacting Wolfram about it, and will receive the information. By closing the question, we prevent the OP from being able to share this information in an answer.
Furthermore, it is possible that some of our members have already asked WRI support for this information, so it is not inconceivable that some users can answer.
I just do not want to miss an opportunity to have such questions publicly answered.
Instead of closing these questions, I'd suggest to the OP to ask Wolfram support and share their response on M.SE.
@belisarius is asking in the comments:
What limit do you propose? Is the "What is the internal implementation of WhateverSysFun?" question always going to be allowed? We could get a lot of un-anwserable questions!
I propose restricting to questions that can likely be answered with an article reference. This is likely the case for most questions along "How does
Method -> "xyz" work?"
Something like "How does
Integrate work?" is too general, as I'd assume these "meta-functions" use many different methods internally. (I know that there's actually quite a bit of info about Integrate here.)