But seriously: what is patentable in video codecs? Isn't it just math function? Give input in the domain of function and you get output.
Beside side-effects are only reading input and writing output - nothing else is side-effect. We already figured I/O almost 50 years ago - nothing novel here (the moment you get touchscreens is the moment kbd on touchscreen is created as it is obvious)
It's akin to asking "what is patentable in a medication; it's just a chemical formula, a diagram on paper".
It's a function satisfying non-trivial constraints, like: size of the output should be smaller than the size of the input [ideally, ratio is adjustable by the user]; regardless of the quality ratio, the output should be similar to the input [really, how do you define "similarity"? -- simple RMS difference won't do]; etc, etc.
Finding such a function is a rather big research endeavour, and IF it has been privately funded, then by all means, there should be away to ensure that the researcher(s) have exclusive rights to the exploitation of their results.
I agree though that the patent system is very faulty, but I am NOT for totally abolishing mechanisms for IP protection.
There are many mathematical functions that have to satisfy non-trivial constraints and none of them are patentable. Why is it so that when you change the domain of use for function from one branch of math to another (CS is branch of math too) you suddenly can patent it?
I'm NOT for totally abolishing IP protection but there are things that shouldn't be patented, especially in very innovating areas of industry (i.e. what is patented today in CS in next 5 years will be obvious thing... while protection will last for 20, thus stifling innovation)