I once designed and manufactured a bunch of automated hearing testers. Being a small company, I hired friends and acquaintances to do the assembly of the printed circuit boards.
I paid by the board because I figured that's how I would prefer to be paid in a similar situation. My thinking was that folks could come and go as they please, working any hours they wanted, and be free to take off whenever they darn well please to deal with life.
Everyone was happy, and if you were fast, you could make $30 per hour, and this was back in the late 1980s.
People could pick up some extra money when I had a big order, and not be annoyed when the project came to an end.
I, and everyone was happy with this arrangement.
But I was later reading an excellent book on management and statistical quality control, and the author, who I greatly respect, by the way, said that paying people by the "piece" was somehow demeaning, and not popular with workers. That sort of amazed me. It was the opposite of what I expected.
I can see how, if you required the workers to be present for a strict schedule, and then paid them by the piece, this might be considered mean.
But in any case, I was still surprised.
This guy, an American by the way, basically invented statistical quality control. He was called upon to go to Japan after WWII, and help rebuild their industry.
He offered his help to the US automakers, and they basically blew him off and refused to listen to his ideas.
Meanwhile, he gave a speech to some heads of Japanese industries, and those that attended called the heads of many other Japanese companies, and talked them into listening to this guy.
His ideas were accepted and widely adopted by the Japanese industrial management, and came to be, later in the '70s, what us Americans considered to be "Japanese Management Style".
And they more or less kicked our butts with it, especially in the realm of automobile manufacturing.
Several key points that have stuck with me are:
1st: Don't settle for part quality that is "within tolerance". Instead, continuously adjust production to keep the parts at the very center of the tolerance band. That way, you don't end up with error stacking that results in problems.
2nd: If an error occurs in a process or procedure, and it happens only once, that may be the fault of the worker. But if it happens more, then you need to look at your process, because you've probably made a mistake in the design of your process that is making these errors likely.
When writing software for instruments and control programs for plants, I've always paid careful attention to that second point. If you make it easy for someone to mess up, then it's your fault as the programmer or system designer when they do mess up.
Anyhow, that's quite far from the original point of this thread!
But the talk of paying based on productivity brought that all to mind.