How does this work, exactly?

> Where, exactly, is the line that must not be crossed to remain free of software patent responsibilities? And what, exactly, happens when that line is crossed by different entities?

Make sure you don't go any further than running the software?

By pretty much EVERY statute or common law in pretty much EVERY jurisdiction, "pure" software is not "eligible subject matter". There are plenty of patent lawyers who would beg to differ, but equally for every such patent lawyer there 10, 20 sleight-of-hand tricks to confuse hardware and software, and not one argument that actually works as applied to pure software.

Proof? Well, could you get a patent on your high school exam paper? You'd get laughed out of court!

Go back 40, 50 years, pretty much ALL programs would start their life on paper looking like a high school exam paper. Not eligible subject matter. They were then mathematically transformed into computer files - again not eligibile subject matter. Next they were typically fed through a compiler - not eligible subject matter. You get the gist - mathematical transformation is not eligible subject matter!

And finally, at the end of it all, they get fed through an ALU. An Arithmetic Logic Unit - the brains of what we now call a CPU. That's logic, that's maths, that's NOT ELIGIBLE SUBJECT MATTER.

Finally we come to print our results on a printer. Using a printer to print (aka "do what it was designed to do") is not eligible subject matter! And that printout could be a hard copy of the source or binary of our program - we've come full circle, and there is NO ELIGIBLE SUBJECT MATTER ANYWHERE.

Likewise, using a Visual Display Unit to display the program's output is not eligible subject matter.

Or making a motor motor. Etc etc.

Of course, there's plenty of eligible subject matter inside the printer. Or the VDU. Or the motor. But that's technology, not maths. Software is pure maths all the way round the circle, and maths is not accepted as eligible subject matter.

Personally, I divide this world (collectively called philosophy) into three distinct areas. Mathematics is the building of internally self-consistent models. Software lives entirely in this world. Science consists comparing those models with the way the real world behaves, in an attempt to understand it. And technology consists of manipulating the real world so that it mimics the model you want it to, not the model it would choose if you left it to its own devices.

Patents belong very firmly in the technology camp.

The problem is the classic one of people mis-classifying stuff. Computer Science is a complete mis-nomer - it should be the study of how computers physically work, and nothing to do with the logic behind them. Physics is classed as Science, when in reality most of it is either Mathematics (Theoretical Physics) or Technology (Experimental Physics). A large chunk of what I thoroughly enjoyed studying as Chemistry is actually theoretical physics (atomic chemistry, the study of electron levels and chemical bonds).

In real life people are far too ready to define things as belonging on "their" side of the fence, rather than see that actually their "garden" contains several fences. Which sadly also means they miss a lot of other people working in the same garden because they've misclassified their fences.

Cheers,
Wol