A very significant element of the Human Interface is the menus -- since Menus are used so frequently, it can be argued that they are THE most important single element in a User Interface -- they are the basic way of issuing "verbs" (commands) to a System.
There is certainly a lot of personal preference involved -- but most of individual bias (preference) is based on what users are accustomed to. Just because you are used to a particular sort of pain, even to the point of thinking it is "natural", does not mean that it is right -- no matter which platform you are accustomed to. This type of reasoning lead to all the DOS/Unix people, ridiculing the GUI for the first few years. Try to be objective when reading the different aspects of menuing that I factored in, and read my reasoning. If you have any other major issues (that I may have forgot), then please email them to me for consideration.
Read How Menus work to get all the low-down on Menus (and the nomenclature that I use). These are the basics, but with quite a bit of detail.
Once you understand the basics of Menus, the next issue is how they are implemented differently on each System. Read Who does What to understand how the menuing Systems differ (and my impressions of each, and why).
Whose Menus are better?
Each of the topics is a link to a sub article that explains my reasoning, and analyzes some of the differences (in more detail than above). There is a lot of information in each section to help people understand User Interface (and menuing), and my rationale. Overall, I think this gives a pretty objective ratings for the different menu implementations.
One of the most important parts of Menus is their TARGETING (how easy they are to "hit") -- the first 3 items (distance, size and position) are all parts of targeting, and are very inter-related (and so are a single article).
There is an interesting way to prove which is better -- get a few different people together (who each know different systems), then call out menu items and watch the results. See who takes longer to find the items, and how they approach the problems. I've done some things like this, a few different times, and it is interesting and enlightening. Past studies (though more broad) have shown the Mac users to be more productive -- which is no surprise if you know the others. Of course an easier way is to just watch users (without them knowing) -- watch windows users overshoot Menus, or give a Mac user a multi-button mouse and see what he does with it, watch a new user explore BeOS or NeXT and see how long it takes them to figure things out.
Unfortunately because NeXT's menuing system is so unique, I think it has died in the marketplace (and will probably not be used again). Sad, because it was a good way to do things (better than what BeOS and Windows use). But unless "different" is far better, then there is no reason for it -- and NeXT's Menus just aren't better enough (if at all) to justify being as radically different than what people are used to. I don't think Apple will use them, and I don't think that Microsoft will change their System -- so this idea may die out. I think BeOS could improve their OS immensely by just going the NeXT menu route, or throwing out the menubar completely and doing it all with contextual Menus, or a drop-down menubar.
The only opinion that really matters is yours. Use the various systems and decide which is best for you. That is what matters TO YOU. I wrote this because I have dealt with a lot of people that want to debate which is better (over all), or they make claims that their UI is better (over all). For those people, we can do some definitive measurements of what is good, and explain exactly why, and prove to them that they are either being too broad in their terms, or that they are wrong. Hopefully in the process, I can teach people a little more about User Interface, and tradeoffs in engineering choices (or at least for menuing in particular).
Special thanks to Leon von Stauber for helping me out with some corrections on mistakes I had made on the NeXT menu's.