Really, I don't understand what you guys are complaining about? Do you really prefer creating dozens or hundreds of place names labels with photoshop, pushing each and everyone to its proper spatial position on the map, drawing each and every river, street, building and tree at its proper position, rather than using georeferenced river, road and forestry geodata and having ArcGIS place them with labels on the proper position on a map according to a predefined geographic reference system, while you just need to define the font type and size, and the look of lines and areas and choose the tree symbol?
I'm really glad there's GIS out there, that just needs to know the spatial reference of all the geodata, whatever the data may be about, and will draw each and every line as defined by its vector data representation at its proper place on the map. While at the same time, I still have ALL the options as to how exactly that line or area should look like. There's unlimited possibilities of defining your own symbols in ArcGIS, and if you don't like those, or you can't find any one that matches what you want to see, there's limitless possibilities of creating your own symbols, even draw them with a pencil, scan them and import them into ArcGIS as a symbol, if you like, to tell ArcGIS to, let's say, "visualize every object whose type attribute is tagged <ship> by this self-drawn ship graphic of mine", or "display all the water areas with my water texture, which I have drawn on canvas and scanned to my PC".
You know, no detailed filling every inch and angle with water using complex Photoshop elements at each position that should be water, no, water areas are already defined and ArcGIS can apply specific visualization to all map areas and elements that match that attribute tag automatically, you just need to decide HOW they are supposed to look like, not having to do the same thing a hundred times.
This is a lot of work and time saved, and you still maintain all your choices and possibilities...you maintain full control over the "how", and can command the "where" by simply specifying the target class of objects on your map which are already spatially referenced and automatically drawn at their proper place. This way, by classifying the "where" just once, you can visualize unlimited "where"s of the same type with just one command/step, specifying your personal "how" all along. And that's the great advantage of real mapping GIS versus just using some graphic designer that doesn't know anything about geodata, the spatial reference attached to its data set elements (geoobjects).
I never intended to create my maps in mind with some graphic designer, rather using proper mapping GIS and importing countless of my own symbols into it, which are probably the only thing I would create with photoshop etc. And I don't think that the mapping guys at National Geographic etc. would prefer doing their work any other way.
Mapping GIS is really a benefit and allows you to save your time for the main thing: Formation of concepts. Concepts of "how do I want it to look like?", "What elements do I need to place where?", "What kind of reaction do I want from my map viewers, what impression should they gain, what elements should I put emphasis on and how should I style them for that purpose"? Creation begins with concept, it IS concept formation of things that do not (yet) exist and therefore still need to be produced. Solving the problem in your head IS solving the problem, it IS the creation. Putting that creation of your mind into the physical world serves as a means of being able to perceive and enjoy the creations of your mind on a sensual level. So let putting your solution/creation into the physical world be the "worries" of a robot or machine as much as possible, the joy you gain, is the same, because the IDEA is yours, YOU have CREATED it!