If it is really an alternate implementation, and not just a minor variation on an existing design, then it is a different process. To be a valid patent, you have to get at the same result using something fundamentally different from existing designs.
In some fields, if you cannot patent a process, you get no protection for your work. This is especially relevant for, say, chemistry - because chemistry is all about process. You do This with stuff A under condition Q, then you raise the pressure and do That, and suddenly, you have Unobtanium! If I were to design a device and patent it, it wuold only be a device that enacts the process - the process is the interesting, valuable bit, not the machinery that enacts it.
I agree the US patent system has issues. I just don't agree that this is one of them.
I think also, a proper implementation of the non-obvious requirement would probably dismiss most of your issues with process patents.
As I have a patent, I agree. Somebody with skin in the game has a different view of things than folks who don't.
Software is a process, so the algorithm is the thing to protect. The implementation can usually be done in any programming language, so protecting the exact code I used (which falls under copyright) isn't sufficient.
If people can get past the Amazon One-Click fiasco, there are a ton of important and valid patents on RAID technology, encryption, etc. The processes these describe took time to figure out and give product advantage to the inventor if their competitor is barred from reverse engineering and copying the process so their product works the same way.
The non-obvious part is really the tricky part of the problem. What is non-obvious changes before and after you've seen the proposed solution.
Tesla's tale of how he got funding for the AC motor is by telling the tale of how Columbus got funding for a return trip to the Americas. His investors said there was no point in returning to the Americas. He told his investors he could solve the riddle of balancing an egg on end, and if they agreed to his bet, they'd fund him. Columbus then demonstrated balancing the egg on end by breaking the egg on end.
The investors cried "no fair, that was too simple." Columbus explained that it was only easy after the work was put in to figure it out. Nobody else had come up with the solution.
Tesla in telling his parable to his investor, then secured his own funding by improving Columbus's balanced egg. he balanced a metal egg on end by means of an alternating magnetic field under the egg, which was the underlying technology behind the AC motor.
In short, what is obvious or non-obvious needs to be weighed carefully. The general test I favor is whether an engineer in the same field would have solved the problem the same way. Amazon's One-Click would have failed because any engineer would have done it that way if their security analyst had told them not to. WotC's Tapping had plenty of prior opportunity to be invented, but they were the ones to come up with it. So it is only Obvious In Hindsight.