There's this notion of sleek and minimalism perpetuated by the teenagers of the internet world. We've all encountered them. The Arch users. And I'm directing that to a specific group - the old-guard and the distro are lovely. But this is a thing where a lot of things associated with youth - lack of experience, haughtiness, a strict adherence to some pre-conceived dogmatism perpetuated by their peers - that leads to this divide between the people who like messing around with these things for the fun of it and the productivity of it, and the people who use these things to stroke their own egos.
Forgive me for the elitist undertone but when in Rome...
Contrary to popular belief, it's not very hard to "set-up" a "minimal" setup - the original intent and purpose though, is to be productive on it, to introduce a workflow that flows well and configuration options that accompany this. That's what tools are, and that's why there's a never-ending endeavour to find and create better tools.
But then comes these people, springing from the brain-child of modern vanity, a byproduct of the "ricing" gamers. They create their own approach on how setups should be (which is fine, software is what the individual make and use out of it). Then they spread their half-baked ideals to more and more of their like-minded kin, saturating the community, until the notion becomes adamant and absolute to a newcomer.
First you should ask yourself - why do you want a minimal setup? What purpose does it fulfill for you that cannot be done with some other approach. And that's not meant to be rhetorical, it something that you should really ask before already deciding. There are many untruths or half-truths to what the *nix world is about, perpetuated from the same means as I mentioned up there earlier.
And if you do decide that "minimal" is what you want, it then comes down to just this - their notion and approach to minimalism is not the only notion and approach to minimalism. You might discover that it's the right one for you, but until you've been exposed to all of the choices, you'll never know otherwise. Tiling managers are nice, but it shouldn't be used because someone tells you that it's sleek and lightweight and all the cool kids are using it - that's something you should be able to decide for yourself without getting caught in all of that pressure to conform to that mindset.
The world "minimal" itself has also been butchered and mishandled in the software world for quite some time. Minimalism regarding featuresets? Regarding the writing style that the author used? SLOC? Approach? Logic? Until you decide for yourself what this means to you - that never-ending search for the ideal setup won't get any shorter.
And more importantly - is it practical? Once you've defined these things for yourself - you must realize whether or not this approach is sustainable towards productivity. Because ultimately that's what you'll be using a computer for. Having fun with experimental things has its pros, and can often lead to finding things that help with productivity as well, but it'd be wise if you can discern between one or the other.
To end off the preaching section of this - when you've achieve a level of self-realization where you understand your needs, understand what you want, and more importantly understand why you might want it, and what influences you to do so, you become all the better for it. And it needn't be through experience that one achieves this - a fresh, critical, but open mindset can do the same.
[rule]Let's line up the choices though. My own take.
XFCE
- very traditional layout. A panel with a menu (if you want)
- a good level of componetal customization. No panel? Why not? Want to replace the default window manager (XFWM - which is rather nice) with something else? Don't see why not. The components are more modular than one might expect. For example I was able to strip out all the power management (and upower along with it) quite easily. And they work quite well by themselves, although some parts were not intented to be. The base is composed of xfce-session (the bread and butter out of what you would want from a DE, otherwise it's no different from a WM), and the panel. If you're using Thunar the xfce-desktop stuff might be integral, but if you're using otherwise, you can strip that off and use some other utilities to substitute in the functionality.
- layout is customizable to an extent - i.e. you can add your applets and move things around.
- the features are kept simple and to what you might expect from a DE. No overload of options if you have OCD and simply need to configure everything.
LXDE
- There's a divide between DEs and WMs. Essentially the former is what ties all of the desktop components - the panel, the application (i.e. xdg ongoings so that your menu knows what the heck's going on and where exactly your applications are, and so that programs can call and interact with one another with more ease), the look-and-feel integration, and things in between. WMs manage windows. It's quite a different paradigm when you realize that the basic ongoings of a desktop, and of a system in general, can be as sparse and cobbled together as one would like. So in that, one realizes that perhaps a graphical server and something to manage the windows on there is all they will ever need - let's not let some other people dictate how they should manage my environment!
- Why is this relevant to LXDE (I knew you'd be anxious waiting for this)? It adhers to a very sparse notion of desktop integrations, environments. It is essentially Openbox, with a few apps to manage the desktop, plus lxde-session for some very basic means for integration. The look and feel is as integrated as the GTK theme you choose. So it's a bit pointless to point out the functionalities. Even the tools that come with it are so completely modular, any sense of them being mandatory is forgone. Expect that from LXDE. It becomes what you make of it. That middle road between those who don't quite feel comfortable without integration, and those who want to dictate how they want their desktops to be managed, because the other full-fledged tools don't do it the way they want (not to say that they don't do it well enough, just not the way they want)
KDE
- Used to be the one. The de-facto. What one would simply expect from a DE. And then plasma came. But don't let the popularity-things (which is not as bad as people make it out to be anymore) influence your choice. With a project this large, the only qualm to a lack of popularity goes out through the window.
- It is a full-fledged DE, so it works well for those who don't feel comfortable without integration. But the customization aspect, brings users who are not really uncomfortable, but requires a workflow that only this level of integration can achieve.
- The customization aspect. Tons of options you can change. And this extends beyond the basic System Settings app. Applications targeted and designed for KDE have this included as well. It's a philosophy, simple as that. If I want to adjust every movable pixel and behaviour...well it doesn't quite go to that extent.
- Eye candy. Better yet, customizable eye-candy that can be turned off if you try hard enough (the difficulty in changing some "default" behaviours is what often leads to people going to more WM-pastures)
- A lot of indexing of files. At some point in time the people there decided to implement a semantic desktop - a desktop that is aware of the state of function of everything within it. Which is admirable and definitely something that pushes advancement for new technologies. The bad thing is that they've made it into the default behaviour before it was fully-finished. The implementation, at its current state, works, but is not truly semantic. So now people have something not-fully-finished pushed down into their throats, and it can be unsavoury. Don't get me wrong, I think it's a great thing to have around and play with, but it should really be something optional and seperate - I'm not keenly aware of the ongoings in the KDE project so my best guess would be something to do with manpower? It's quite a bit leap. I might be off-base about it as well. But the fact stands - it takes a bit of effort to get rid of the components supplementing the indexing - nepomuk and akondai. But it can be done. Problem is that some of the offical KDE programs like Kmail and the likes, bakes this stuff right into themselves as well, so you'd have to take some time in compiling it all out. Sometimes it simply can't be done. I'd recommend Gentoo-variants if you're being anal about this and need a standardized and sorted solution.
- I'm not even going to talk about the look-and-feel. You can change it to however you'd like. Rather easily. Of course there might be some exaggeration but the core truth of the statement is there.
GNOME
- Minimalism through less feature sets on the surface. This is an approach that works well for some, and not for others. Don't be swayed from the hate or the hype. Decide for yourself. Unlike trying out headphones, it takes markedly less effort to do this with software.
- This can be perceived as the standard for many things in *nix world right now. Whether people like it or not. Red Hat has chosen this as their main squeeze for distribution. Not because of any staunch beliefs, just because this is currently the biggest project and the corporate logic makes sense. I'm not riling on anything. But they've pushed a lot of development for and into integration with gnome. A lot of key applications require Gnome dependencies, and they have been for quite some time. So there is an inherent ease-of-adaptation when your core stuff has all the standard behaviours worked out just for this DE - although it doesn't always demonstrate itself practically, I get the feeling. GTK3 - that's pretty GNOME-y right now. VTE. Etc, etc, etc.
- As mentioned in the first point, less features and options to change around. Of course you'll find an onslaught hidden in config files which can be accessed centrally through dconf. But on the surface, the developers have this vision and they don't really want you doing anything about it. But just like how they should probably take a less pragmatic approach to developing, so should the user when decide - it's the functions, not the people, that you should be judging when choosing. You can argue that what the developers are doing dictates the future, and that if you are choosing this, the experience might not be as sustainable and good in the future. They've taken a lot of backlash, and you see less bad directions, but then again they still exist. Playing a bit of devil's advocate here.
Awesome/E17/Blah.
- They exist. If you want I'll talk about them. Don't count them out of the running.
I'm gonna install Mint with xfce until I get used to linux, then go to arch, and then probably funtoo! 
