Mutation

A mutation that kills immediately is detrimental, no matter what the environment. A mutation in an unused section of DNA that does not activate that section is neutral, no matter what the environment. However, those are special cases; in the general case, you cannot tell if a mutation is beneficial or neutral or detrimental without considering the environment. By literally milions of random trials, filtered by natural selection.BTW, we have produced new species of flies in the lab. See section 5.3 of Observed Instances of Speciation.

I don't have figures to hand, but the majority of mutations are neutral and the remainder are somewhat more harmful than beneficial.

Does not follow. Calculating a meaningful change in data or information as the result of mutations is a challenge at which nobody to date has been successful (although a few have claimed success). Almost ... "mistake" is definitely the wrong word to use. It implies that there is a right way and a wrong way and the wrong way is the mistake. But there's no right or wrong, there's just better or worse equipped for reproduction and survival in the current environment (which, of course, implies that what's better in one environment can be worse in another environemnt). Mutations are not mistakes, they're just changes. You could substitute "difference": Yeah, that's reasonable.

Well, first you have ignored the effect of selection, which is a powerful filter that makes evolution definitely not a random process.Second, there are many ways in which "irreducibly complex" systems can arise through evolution. One that comes to mind is scaffolding; system A+B+C may not be irreducibly complex, but remove system B and you may still have an operating system that is now "irreducibly complex" (similar to how an "irreducibly complex" arch is built with scaffolding). Another is co-optation; a system may perform a particular function and mutations may produce a new system that is "irreducibly complex" and performs a different function.Third, it may be that the knee requires those four complex parts to exist to operate as a human knee but you are going to have to do some serious arguing to convince me that they all have to be present to function as a joint.There should be a discussion of the knee at http://www.philoonline.org/library/shanks_4_1.htm ("Behe, Biochemistry, and the Invisible Hand", Niall Shanks & Karl Joplin) but that seems to be down, perhaps forever.

Agreed. You probably didn't note that I carefully never describe the italian anti-atherosclerosis mutation as "beneficial" without a caveat; we like the idea but whether it contributes to, detracts from, or has no effect on reproductive success remains to be seen.However, "neutral" properly means "currently not subject to selective pressure" and can be detected with a high degree of confidence; just by pure statistics most mutations take place in "junk" DNA which seems as best as we can tell to have no function (and some obviously does not have any function), and therefore are neutral. A neutral mutation incoding DNA could possibly become another type in another environment.

I don't know. That's not one of my areas of expertise. Perhaps someone else can help you. Knees evolved very early, in the first tetrapods. We have lots of examples of tetrapod fossils with primitive knees. You might look up Acanthostega and Eusthenopteron and Sterropterygion.I do hope you realize that your inability to think up an intermediate stage does not mean that there cannot be an intermediate stage.You might also consider how poor a design our knees are for carrying our weight in an upright stance.

In humans, yes. In shorter-lived organisms there are some clearly beneficial mutations knonw.