The Supreme Court has fundamentally changed software patents

What I find interesting in the comments section is the assertion by some of the pro-patent people that the reality of computer programming does not align with the Supreme Court's decision. The crux of the issue seems to essentially be:

Lemley: "We shouldn't allow patents on pure functionality; only on the means of achieving that functionality." He also phrases this as patenting the solution and not the problem. http://www.wired.com/2012/10/mark-lemley-functional-claiming...

Pro-patent people: "Specifying functionality is the way software is developed - just look at any software outsourcing site. And that specified functionality is what should be patentable."

I wonder what experienced software developers would say to this. At what point, if ever, should functionality (alone) be patentable? The Supreme Court seems to have gone with "nope, it isn't patentable without some new inventive step or hardware". But that means there's no clear line.

======

A nice hypothetical, quoted from elsewhere, that cuts to the center of the issues:

Assume that a person has no knowledge of computers or programming at all, but they believe that certain functionality would be profitable, for example – a function that analyzes your food eating habits and work schedule and automatically orders and pays for delivery, so your favorite dish is waiting for you when you get home. Can this person receive a patent? And I want to be very clear here – This person knows nothing about computers, he only knows what function he wants to see them achieve. His specification will read with the same level of expertise as my exposition here does. Why or why not?

This is quite analogous to much of what's been happening in the industry. Should it be patentable or not?