Philosophy can be defined or described in accordance with how it has been practiced in the Western tradition. Alternatively, the word 'philosophy' can be defined simply in terms of dictionary definitions or even according to its etymology.
The latter approach isn't very helpful. At least not from a metaphilosophical perspective. Similarly, saying that philosophy, for example, was simply the “study of all examples of knowledge” (at least for some ancient Greeks) isn't going to get us very far either. For a start, it simply begs the question: What is knowledge?
And neither does etymology help us. Namely, philo = love; phia = wisdom. One problem with taking the etymology of the word 'philosophy' seriously (as Martin Heidegger did) is that it seems to be the case that philosophy should be all about the self – or about the “lover of wisdom”. In other words, “how to live well”, “how to live the good life”, “how to be fulfilled and happy", etc. Clearly all this has only been a small aspect of Western philosophy; and, perhaps, a big aspect of various religions. It can even be classed as self-centered. (In certain strands of existentialism, it's mainly about living a sincere life – sincere to one's genuine self.)
Here questions abound. Why should philosophy be all about how to live one's life? Why should I live the good life rather than the bad life?
Some have explicitly said that “philosophy is committed to self-knowledge”. There's some truth in this in that Socrates famously said “know thyself”. Though was that really about the self or was it more about the self's relation to knowledge about the world/reality generally? In other words, if one knows oneself (therefore one also knows where one's going wrong – intellectually), then one will have a better philosophical grip on the world or reality.
We can also answer the question “What is philosophy?” by asking a similar questions about the sub-branches of philosophy. For example, we can ask: What is metaphysics? Here too we can become all etymological and say that the Greek word meta-physika literally means "what comes after physics". That's not very helpful either. (What's meant by “after” or “meta”?) So let's forget dictionary or etymological definitions and go with the following.
According to Wikipedia, metaphysics is “the study of existence, causation, God, logic, forms and other abstract objects”. So why isn't metaphysics the study of cups or cats? Can one study "existence" in the abstract? The point here is that we can't help but be metaphilosophical (or simply philosophical) in pursuit of an answer to the question “What is metaphysics?”.
What is Philosophy?
Every statement on what philosophy is - or what it should be - will elicit the question: Why do you believe that philosophy is x? The philosophical opponent can easily tell the original philosopher his own view on what he thinks philosophy is - or what he thinks it should be. If that occurs (which it often does), then what happens next? How is the what-is-philosophy question settled when rival views are on the market place? Surely the opposing positions on philosophy will be debated; though I doubt that the debate will be settled. And I also doubt that they can be settled by taking various metaphilosophical positions on the what-is-philosophy question.
Let's take a couple of examples.
One philosopher can say that philosophy is about “finding the fundamental nature of everything”. Why should a philosopher do that? And doesn't this stance on philosophy simply assume that there is a fundamental nature of things taken individually or a fundamental nature of “everything”? What if there are no such fundamentals? And, even if there are, why should a philosopher see them as important? (Though classing something as “fundamental” sort of gives the game away.)
Alternatively, a philosopher may say that philosophy is about (or should be about) intellectual unification. Specifically, unifying the insights from other disciplines; particularly science and philosophy itself. Another philosopher may say that such a position is impossible. He may also dd that science itself is a discipline which simply doesn't require philosophy. (Many scientists - particularly biologists - have said this.) Indeed such a philosopher may say that philosophy itself should incorporate science and its findings. Thus - from such a place - it would be very difficult to take a useful (or genuine) metaphilosophical position on science.
The Question: “What is Philosophy?”
Some of statements and arguments from philosophers on the nature of philosophy seem well-trodden. That is, they're simply of the traditional “What is philosophy?” variety.
For example, a philosopher can say that “philosophy doesn't rely on faith or revelation”. Instead it relies on “reason” or on (in 21st century pretentious terms) “cognitive criticality”. Nonetheless, such a position of faithlessness or lack of revelation doesn't automatically make philosophy a science either. Some philosophers might have said that there's no need to rely on observations or experiments in philosophy. A contemporary philosopher, on the other hand, may say that sometimes philosophers indulge in thought experiments (as the the Wykeham Professor of Logic at the University of Oxford, Timothy Williamson, does in his The Philosophy of Philosophy) which are very like the thought experiments engaged in by scientists (Williamson cites Galileo). Nonetheless, they're still not physical experiments as they're commonly understood in science. In addition, observations may be said to be prerequisites for just about any kind of philosophy. And technically it can also be said that observations (or at least “a posteriori reasoning”) can defeat seemingly a priori claims or statements.
Bertrand Russell on Philosophy
Bertrand Russell seems to have believed that when it comes to the definition of the word 'philosophy' (or to a description of the practice of philosophy), one can't help but be metaphilosophical. (Of course Russell never used the word 'metaphilosophy'; or even the words 'the philosophy of philosophy'.) In his The Wisdom of the West, Russell wrote:
“Definitions may be given in this way of any field where a body of definite knowledge exists. But philosophy cannot be so defined. Any definition is controversial and already embodies a philosophic attitude. The only way to find out what philosophy is, is to do philosophy."
Surely it can said that a definition of the word 'science' won't be equally as problematic as the word 'philosophy'. In addition, one will need to take a philosophical stance on what science is (if not on the word 'science' itself). Similarly, would all scientists agree on such a definition? Thus it can't be the case that simply because the word 'philosophy' is about philosophy that all definitions will be more problematic (or controversial) than definitions (or descriptions) of science.
So let's rewrite a bit of Russell's quote. Thus:
The only way to find out what science is, is to do science. (Or at least see how science is done.)
So it can be said that this controversy (or problem) is also the case with the definitions of many other words. That's unless one simply stipulates: This is how this dictionary defines the word x.
Despite saying all that, the analytic approach to philosophy, for example, certainly “embodies a philosophic attitude” and that attitude is “controversial”. The same can be said of deconstruction, phenomenology, structuralism, etc. - i.e., virtually any way of doing philosophy. Of course one would now need to distinguish positions within philosophy from positions on philosophy itself.
It's hard to grasp Russell's final sentence in the quote above. (Namely: “The only way to find out what philosophy is, is to do philosophy.") Surely there can't be such a case of (as it were) a priori philosophising. Firstly, a student of philosophy must read the books of certain philosophers and only then can he write about the things they too have written about. He may even adopt the prose style of those philosophers. Later he'll probably make a self-conscious attempt to write a certain kind of philosophy in a certain kind of way. In no way will he simply discover his own voice the first few times he writes philosophy. If he didn't do all that, then isn't it likely that he'd be doing stream-of-consciousness expressionism rather than philosophy? Unless, again, he's literally writing genuine philosophy from an a priori position; which, surely, is almost impossible. Sure, in order to “find out” if one can do philosophy one will need to “do philosophy”. And then one will discover which approach one likes. However, an original position can't come about simply as a result of doing philosophy from nowhere.
Finally, we must conclude that the question “What is philosophy?” is itself philosophical in nature. Or at least we need to indulge in philosophy in order to discover an answer to that question.