The Harvard researchers didn't use gene therapy to lengthen the telomeres. They engineered a knock-in allele encoding a 4-hydroxytamoxifen (4-OHT)-inducible telomerase reverse transcriptase-Estrogen Receptor (TERT-ER) under transcriptional control of the endogenous TERT promoter. Basically, the mice had short telomeres and the researchers could reactivate telomerase by administering 4-OHT. That's genetic engineering, not gene therapy in adult mice.

Furthermore, the Harvard researchers showed the reversal of artificially-induced aging, but not an increase in lifespan. The researchers in this study demonstrated an increase in lifespan in normal mice.

The Harvard study showed that improving telomerase activity could reverse or slow aging, but it didn't show how to actually accomplish this in normal, adult organisms. That's what the researchers in this study have done, at least in mice.

No, the Harvard researchers didn't do the same thing. They genetically engineered mice to have short telomeres, inducing faster aging, and then reversed the process by reactivating telomerase. The mice didn't actually live longer than normal. By contrast, the researchers in this study used a single application of gene therapy to extend the lifespan of normal mice, and they did so using techniques that have already been used in humans to treat other conditions.

That would mean that, as a practical matter, no patent that included a software component would ever be infringed. The all elements rule requires each and every element of a claim to be present in order to find infringement. Thus, a patent on a new, useful, and nonobvious machine that happened to use a microcontroller would be worthless because it would never be infringed. A competitor would simply sell the machine and the software separately. The machine by itself would not infringe the patent, and selling and installing the software would not infringe the patent under your proposed rule. Trying to patent the machine without the software installed would not work either, since without the software the machine would not function and thus would fail the utility requirement.

This is an example of why, even if Congress were prepared to ban software patents, it would not be very easy to do so. Until there is a workable definition of "software patent," calls to ban software patents are equivalent to saying things like "we should stop crime." It may be a laudable goal, but without a workable plan for accomplishing that goal, it's not very helpful.

Okay, great. Now turn that into a definition of "software patent" that is neither under- nor overinclusive, can be easily and unambiguously applied, and cannot easily be gamed.

Actually, you could. Consider a simple rubber curing machine that has only one control: a thermostat. One coud patent that machine and one could patent a method of curing rubber using the machine, which gave various temperatures and hold times that produced a good product.

Sure there is. Such a program would be highly useful to the designers of rubber curing machines, as it would allow them to test new designs without having to build physical prototypes. For example, the Boeing 777 was the first aircraft designed entirely on computer, which was apparently a great improvement over the old design process. Why wouldn't we want to encourage such innovation?

Actually, Jobs was famous (some might say infamous) for involving himself in the tiniest details of user interface design and software features. See, e.g., this article: "The design process behind OS X 10.0 and Steve Jobs’ remarkable level of involvement".

Even number 5, which is strictly a patent on a computer program for modeling the rubber curing process and thus involves nothing physical, apart from the computer itself? What about a patent on a computer program for simulating the behavior of objects made of rubber (e.g. in a 3d modeling program)?

How is that different? The software is just a piece of the process for displaying a cursor. What's so different about a cursor on a computer screen and a piece of cured rubber? They're both useful, physical applications of a mathematical idea.

Then allow me to be more clear: I mean programs in the context of software patents, which is the context in which we are discussing the issue. A software patent does not claim an algorithm by itself (i.e. the claim does not begin with "An algorithm, comprising ..."). It claims either a computer programmed in a particular way or a computer-readable medium on which the program is stored or a method for programming a computer or some other connection to a physical artifact. In the context of software patents, a program is tied to a physical device in a way that makes it both useful and concrete.

By "ponder" I meant to consider a program or mathematical algorithm in one's mind. The point is that one can "do math" in a way that is not covered by a software patent that claims a useful application of the same mathematical idea. This demonstrates that such a software patent is not a patent on the underlying mathematics.

But software patents claim more than just a program. They claim a program running on a computer, or stored on a computer-readable medium. Consider, for example, this hypothetical claim: "A method for calculating a sum comprising a user entering two numbers into a computer followed by the computer calculating the sum of the numbers and displaying the sum on an output device." No amount of math (in the abstract sense) will cause a user to enter numbers into a computer or for the computer to display the sum. I can do math (e.g. prove theorems, use formulas to calculate results, etc) all day and never cause that to happen because it requires input from a person and output on a physical device. Neither of those are math.

You've only stated a conclusion, not given reasons for it. I gave reasons: both are embodiments of mathematical ideas that have useful applications in the physical world. The curve of an airplane wing is defined by physics, which is in turn defined by math. A patent on a wing design is, in that sense, a patent on an application of the equations that define an optimal wing. A computer program is defined by math, and a patent on software is a patent on the application of that math to achieve a useful result.

In order to ban software patents, one must first define software patents. I challenge anyone in favor of banning software patents to come up with a definition of the term that is neither under- nor overinclusive, can be easily and unambiguously applied, and cannot easily be gamed. Here are some example technologies to think about as you develop your definition:

1. A machine that cures rubber by heating and cooling it, controlled by hand.
2. A machine that cures rubber by heating and cooling it, controlled by a computer program using a new, nonobvious, and useful application of a mathematical equation.
3. A new and nonobvious kind of rubber curing machine that uses such a complex curing process that only a computer could control it, resulting in a significantly superior product.
4. A computer program for controlling the rubber curing process using a new, nonobvious, and useful algorithm that could only be carried out by a computer, resulting in a significantly superior product.
5. A computer program for modeling the rubber curing process using new, nonobvious, and useful applications of mathematical equations.

If you want to ban software patents, where do you draw the line?