This is a combination of the ideas in Paul Graham's fantastic essay at http://www.paulgraham.com/philosophy.html and my extrapolations, based on thinking about the essay and discussions with a few brilliant people.
- Up to and including Aristotle, philosophers were working with brand new ideas (or ideas that were new to them). So there was a lot of progress.
- At some point, philosophy encountered the problem that words are not precise (alternatively, our brains are not equipped to understand the universe at a non-superficial level). If you ask questions you actually need answers to, like "is this action just?", you first have to define justice. And then you go deeper and deeper into the rabbit hole with no useful answers, just more questions.
- To avoid this problem, philosophers started focusing on even more abstract ideas, such as "what is causality".
- During the early enlightenment, Aristotle and Plato were so revered that nobody had the confidence to break down their work into simpler ideas. Instead people assumed the concepts they were talking about were the important ones to focus on. But we had not even invented tools to solve more basic questions, so it was trying to solve calculus without having learned arithmetic.
- Up to the point of Wittgenstein, many smart people realized this problem, but did not try to change it. Instead they quietly chose other fields to work in. This is why the most brilliant people in Greek times worked on philosophical problems but now few of the smartest people become philosophers.
- Wittgenstein made enough of a fuss about this problem that philosophers had to face it. From that point on, philosophy has been in even more disarray, and has become even more academic. Nobody knows how to build on what we have so far, which is why its hard to think of someone who has recently contributed anything great to philosophy.
- Instead of trying to answer the most abstract and general questions, we should recognize the limitations of our brains. We can try to find the answers that are useful and generalize them as far as our current tools allow. The test of usefulness is that if someone first learned this today, they would change how they behave. Some of these will not be as impressive as discussing causality, but we can find real general answers and build on these over time. Some examples that come to mind are Mandelbrot's ideas on the fractal nature of the world, or the recent work people have done on complex systems. There are ideas in these that are general enough that they move beyond science into philosophy.
- We have to recognize that it is much harder to make rock solid building blocks in philosophy than in math, because of the imprecise nature of the questions. But it is worth it. A lot of progress could still be made because most of the useful work in it has only been done over a few hundred years.