Quote -  Good info, and definitely food for thought.  Please forgive me for waxing philosophical here a bit.  My approach to evaluating software (3D or not) is based on my theory of knowledge:

Met cognition: how to learn anything

1. What are my basic facts (nouns)?

2. What are my basic actions (verbs)?

3. In what order, with what tweaks, do I perform these (grammar, declinations)?

 Works for a foreign language, chemistry, physics, math, 3d software,…

Now for 3D software the question is: given that time is limited, what are the basic things I (as a rookie amateur with aspirations) want to checkout?

For me, for modeling:  (a) Make a cube.  Add and take away polygons.  Round some edges.  Extrude and bevel.  (b) Make a cylinder.  Do the same as the cube.  Then round off the body of the cylinder and make a gear (Silo3D was really disappointing here).

I still have to come up with a set of basic things for those operations beyond the modeling.  Access to systematic training materials and ease of use count.

Any suggestions for what basics to learn and try out to do a (relatively quick) software evaluation?

 

User Interface. That's really the simplest answer. How easy is the user interface of your chosen software navigated. By you, not by others. Because some people say Blender is great, while others won't touch it because they find the UI impossible. And that's true for just about any software. Some will love it for this and this, while others will hate it for exactly the same reasons. 

Most any modeling software is going to include the basic functions - move, scale, rotate, translate, extrude, etc. From there, it's all about how quickly you can access those functions, and that requires being familiar with the software and knowing what commands to look for. If you have to dig through a bunch of menus and sub menus to find the extrude command, and you can't place a button for it within easy reach of the mouse or pen, or if it takes a third hand and a degree in yoga to hit all the macro keys, I'd say move on to something else. 

But without first having a solid grasp of what it is you're trying to do, or the steps involved in doing it, I don't think you'll be able to evaluate any software to a satisfactory level. 

On the same note, I can't tell you that you should be making your gear from a box or a cylinder because one modeler will use a box and the next will use a cylinder, and yet a third will use a flat plane. It's all about developing your own methods, and that comes with learning your software.

Your goal is to make a gear, but you aren't familiar enough with modeling techniques to know how to make that gear in each piece of software you're trying to evaluate.

It's not because Silo doesn't have the functions of making the gear, or that a gear made in Silo would look any different than a gear made in Blender or Wings or Maya, it's just that you're unfimilar with the steps involved.

Every software title is going to have a somewhat different approach to modeling than the next title. That doesn't mean the model consists of anything other than vertices and edges, it just means that the software used to generate those vertices and edges functions a bit differently than the next.  The final product should still be the same, and is based almost entirely on the skills of the user and his/her comprehension of the software they're using, not the functionality of the software itself and definitely not its price tag. 

So my advice would be to first learn how to make a gear, and learn it in one title, before trying to learn it in another. Once you get familiar with one piece of software, while you're learning and building your modeling techniques, you'll have a better understanding of what features to look for in other software. There's always a dozen ways to accomplish the same end result in modeling, regardless of the software being used.

Take your time to learn at least the basics of free software and get a good understanding of what goes into polygonal modeling before spending lots of money on a program that you may wind up finding isn't any better than the free titles. Art is based entirely on the skills of the artist, not the medium used to create it. whether it is traditional art or digital art, it doesn't matter. There is no "make art" button, no matter how expensive your paintbrush is. 

I think by trying out a lot of different titles early on, you will only wind up confusing yourself more in the long run. Pick a title and stick with it, until you're confident in your skills in that title and can try out other software. Every time you pick up another title, you're going to have to take some time to learn that title's interface - how much time depends on how fast you're able to learn new skills. It doesn't happen instantly. 

I've probably made this too long as it is, but maybe this will help you settle on whichever package you're most interested in. Chances are they will change many times as you get more familiar with 3D. You may love the modeling features and ease of use of one piece of software, but hate its rendering engine, so at some point you're going to want to look for another title that can import your scene files and render them to your liking, and that is going to require learning another title. 3D work is a never ending process of learning and progressing and experimenting with new things.