Not sure strength in numbers is a good thing if your talking about lemmings. It's like science being settled by consensus and not accurate data.
A perfect rocker is something that can't be improved upon. Technology will always find improvements, but math is math. No one ever came up with a math equation that changed the outcome of natural law. They just discovered a formula to explain the outcome. It just shows how insignificant we are in the universe.
With todays technology, a "perfect" rocker is one that transfers the cam information to the valve with the least amount of wasted motion or losses. You call this ratio change, but it is not. It is simply the result of angularity introduced into the system through arc motions. No matter where you check the rocker in it's rotation around it's axis, the lengths remain the same. (Have you watched the video I linked to yet?) A perfect transfer of cam lobe information wouldn't use a rocker arm at all. Overhead "bucket" type cams come to mind.
Using your logic, a small block Mopar, which does not have a pushrod that follows the trajectory of the lifter (it's on a different angle), the loss of lobe lift at the end of the pushrod would mean that the lobe is getting smaller, therefore, less lift. That would be utter nonsense, but somehow, because you see losses at the valve, all of a sudden the rocker arm is losing ratio, and the fulcrum length is getting longer or shorter. Utter and complete nonsense. It is called Trigonometry. You have to understand what the "whole" system is doing before you can attempt to make a better rocker, or make a claim that someone who does understand the system is wrong. Back up through the system to see where the losses actually originated, instead of automatically assuming the ratio changed. This is where Smokey went wrong. He observed a loss at the valve, and then noticed the fulcrum sweep which led him to conclude there was a ratio change. He never investigated beyond the fulcrum for the origin of the losses. Had he done that, he would have found that pushrod/cup angularity, from sweep and rocker design, was reducing the input, and thereby, reducing the output as well.
Ok, you keep beating this old nag. Just shoot her and put her out of her misery. There is no valve side ratio or pushrod side ratio. But, because you insist, here is a test question for you.
Henry has a Jesel NASCAR SB2 rocker arm that has a .545" roller tip and has a 1.620" fulcrum length and a .575" offset. What is the ratio of this rocker? (For informational purposes only. Specifications are random, and do not necessarily reflect an actual Jesel SB2 rocker arm.)
If you want something to really make your head hurt; (this may not seem relevant, but it really is)
Three salesman travel to New York for a large conference. When they try to find a hotel room they find they are all booked. Finally they stumble across a budget hotel that has one room left, so they decide to share it. The young man at the counter tells them the room is $30, so they each pay their $10 and go up to the room.
Meanwhile the hotel manager comes back from a lunch break, and the young man tells him he rented the last room to three salesman. The manager asked how much he charged and the young man replied $30. No, no says the manager, that room is only $25. Take $5 up to the room and return it to the salesmen.
On his way to the room, the young man is thinking, "how am I going to divide $5 among three salesmen"? "I know", he thinks. "I'll give them each one dollar and pocket the other two." So, that's what he did.
So, the question is, If initially the three salesman each paid $10, for a total of $30, and they each got back $1, then they each paid $9. 9x3=27, so they paid $27 for the room. The young man pocketed $2. $27+$2=$29. If they initially paid $30, then what happened to the missing dollar?