|
To answer your question, there's a number of reasons. Undoubtedly some do just come down to "historical reasons" but there are many reasons why a command-line is actually the right choice. The main reason is that the scientist wants to run a very specialised analysis of her data, and so is almost certainly either running custom code, or at least code that was developed in an academic environment. Creating GUIs that are flexible enough to cover all the things somebody might do with their data is a complex and time consuming task, and does not directly produce "science". It is a lot easier to use the command line to interact with your dataset. Moreover you often don't know what you want to do in advance ("if we knew where we were going, we would't call it reasearch"). Therefore a typical user may fiddle around with the command line trying to get the most out of their data, and then string it all together in a script - so again, a command line makes a good prototyping environment. Certainly in my area (astronomy) I have seen some attempts to provide gui based intefaces to data reduction, and even mocked up a few myself; but the effort to reward ratio isn't worth it - especially since you are dealing with people who can, in general, follow instructions. Also, do not underestimate the higher effiency of the commandline - after all, there is a good reason we all use it. Even if there was a system administration gui out there, would most sysadmins use it? It's actually faster to type. I think the "appliance" model of computing (eg. iPhone/iPad) is perfectly valid for a good percentage of the population at least for some of the time, but it won't render "classic" computers obsolete. I also have to say that personally, I have generally been dissatisfied with attempts to disguise or flatten the filesystem by, for example, using tags instead. I actually think pretty damn hierarchically, so I often organise my material as IF it was a filesystem, when the application will allow me to do so - maybe it's just me? |