Second, classic desktops are generalist desktops. They make few assumptions about the expertise of users or how they work -- much less about how they could work better. Instead, they provide a set of basic tools, and leave users to work how they see fit.
By contrast, both Unity and GNOME were designed with the assumption that they could improve how users worked. GNOME 3, for example, is supposed to make it easy "for users to focus on their current task and reduces distraction and interruption."
That sounds reasonable in theory. However, in practice it translates into opening most windows full-size, removing customization from the desktop, and reserving the panel entirely for system purposes, all of which makes multi-tasking vastly more difficult.
The result is a desktop that is much more efficient than a classical desktop -- but only for those whose habits happen to match those that GNOME 3 enforces. For others, it is considerably less efficient. Instead of learning a new way to work, users are likely to feel themselves working against the desktop to a much greater degree than with a classical desktop.
While a classical desktop is unlikely to be optimized for any given habits, it has the advantage of almost never being completely unsuited for any habit. In KDE's case, the interface will even provide an alternative menu or system settings dialogue window for users who dislike the default.
The difference is that of a special tool versus a Swiss army knife. The specialized tool only works well when you use it for the purpose it is intended, while the Swiss army knife, as clumsy as it can be, works well-enough for all purposes.
However, by far the most important aspect of classical desktops is that they allow extensive customization. A couple of months ago, when I asked experienced users why they chose KDE, almost all of them mentioned customization. I have often heard the same from those working from other GUIs. The situation may be different on Windows or the Mac, but Linux users apparently have a long habit of preferring to do things their way – which is as strong on the desktop as it is at the command prompt.
This preference is not exactly a revelation. Yet one of the goals of alternative interfaces -- at least originally -- has been to reduce the amount of possible configuration, subordinating user preferences to a vision of branding, or as GNOME describes it, "a consistent and recognizable visual identity." Similarly, users' preferences for widgets and desktop icons were removed or reduced in number.
Perhaps the decision to reduce customization may have spared a few new users option anxiety -- the sense of being overwhelmed by too many choices. Yet whether option anxiety is a major issue on classical desktops seems questionable, since new users can always stay with the defaults until they feel ready to explore. But, as might have been predicted, the reaction to the reduction of options were extensions and utilities that replaced much of the missing configurability. In fact, today extensions make GNOME one of the most customizable Linux desktops available.
The classical desktop is not a particularly ergonomic interface. As the designers of GNOME and Unity observed independently, it can often be improved upon for a given sense of assumptions.
Yet, at the same time, the classical desktop has advantages that should not be thrown away without serious thought. Well adapted to its niche of traditional computing, good enough for most purposes, and highly configurable for personal tastes, the classical desktop still retains an edge over alternatives.
True, a classical desktop would hardly work as a phone interface. Yet if it is rarely ideal, it still gives readers more of what they want than anything that has been designed so far to take its place.