Reasoning:
1) software is easy to build, so 5 years is more than enough
2) to claim any damages the patent needs to be validated by the market first
Makes sense?
"A patent to compress video by an average of 8x over current state of the art algorithms.
Inventor seeks 240 months of protection for disclosure of the algorithm.
[ ] Worth it [ ] Not worth it"
If we set the threshold at X thousand, or Y % of those in the sample, then the inventor can get the patent protection for 240 months. This could be done as the very first step. So if value-less inventions are applied for protection, they could be screened out without draining resources.
Alternatively, you could do something more like a market value approach, and ask folks how many months of protection the invention would be worth, and take an average of the most commonly suggested amounts (in terms of months).
The point is, there are much better alternatives than a fixed N years of protection for any invention of a particular category. Not all inventions are equally valuable, so it doesn't make sense to price them by category any more than it does to price them (in years of protection) the way they are now (everything's equal to 20 years of protection).
So, rather than wait for the market to validate the patent after it's been granted (which wastes USPTO resources), let's validate it's value ahead of time. If no one thinks it's worth much, the inventor may abandon the claim. Or the USPTO can reject it outright as not meeting the utility criterion since the estimated value (by society) is too low.