Trying to look at this from a high-level non-TD user perspective... In Soft, whether it was intentional design or just fell out of the toolset that evolved, any geometrical object (not using the word in the programming sense) has, through centers, pivots, reference planes, neutral poses, constraint compensation and child compensation, effectively got a complete and fairly intuitive rig already embedded. The ability to toggle in and out of various manipulation modes easily is very important in terms of usability. The combination of the tabbed PPGs and operator stack allow people to drill down to the atomic level if necessary, but don't tend to interfere with the user experience. The fact remarked on above, that all the tools continue to work in all modes, is also key.
In Maya, it seems to me that an underlying assumption was made that everything should be REALLY atomic, and that users should do everything by assembling things into mostly single-purpose, single-mode rigs or tools. I guess this makes sense in a pipelined, scripted environment, especially when you have TDs to build and deploy a stable toolset, but it's hell on Earth for a sole-practitioner/generalist who is expected to quickly produce results from scratch. I gather that you *could* devise a rig, using locators and constraints and whatnot, that would emulate the flexibility of all the Softimage transform controls, and use a script to apply it to any object that gets a transform node, but unless you also spend considerable time to make a custom UI for it, the usability would be awful. Not to mention the mess you'd have in the node editor or outliner. To me, this goes along with the silly amount of clicking on things you have to do in Maya -- it's all very logical and sensible from a really granular standpoint, but scales poorly (in the sense that repetitive tasks become REALLY repetitive) and neglects the difference between things that need to be done frequently and things that are rarely needed. And I guess the tendency toward single-purpose single-mode tools and operations is also related to this, with the often-disastrous effect of being unable to make even a simple change without recreating the whole history of operations. Again, if you do everything with a script, that might not be a big deal -- you edit the script to make the change, then run the whole thing again. But if you are trying to keep everything "live" and editable, and don't have scripting skills, or a TD who does, Maya is intrinsically more limited. I'm resigned to the prospect of dusting off scripting skills I haven't exercised in nearly 20 years, but I'm not happy about what is basically a regression in the state of the art. And I know that scripting and coding are hugely powerful tools -- there are things that you can, or should, ONLY do with them -- but I am not nearly as facile with those tools as the ones in the Softimage workflow/toolset. Besides -- it's wasteful and inane to use a sophisticated, powerful tool to do a simple frequent task. Just because I could use a 6-axis CNC milling machine to drill a hole in a board doesn't mean it's a better tool for the job than my old Makita driver drill. Yes, if I needed to make several dozen holes, all precisely sized and spaced, at different angles, in several different boards meant to fit together at a later stage, that CNC machine would be handy -- but 99.99% of the time, all I need is that one simple hole, right NOW, and I don't want to take the workpiece off the jobsite back to the machine shop.