At my company (non-tech), all leadership compensation is paid based on percentages hit over the next five years.
Your first year, you’re only getting 20% off the position’s salary, and 80% goes to your predecessors. When you retire, 20% of your salary got the previous four years is based on how well the company does the year after you leave.
It strongly incentivizes healthy transitions and long-term plans.
The study of reinforcement learning, I've found, is incredibly insightful as a way to really force us to operationalize reward functions. I've benefitted from the process of reasoning and exploring RL reward functions.
Sure, I also value game theory, sociology, and psychology as well. People are complex, with culture, identities, imperfect reasoning, and biochemical influences.
That said, one key benefit of RL is the hands-on experimental aspect. You get to play God and see what happens. Surprise triggers cognitive dissonance and drives further questions. In contrast, some academic disciplines become rather enamored with their own theories, to the detriment of operationalizing them.
Your first year, you’re only getting 20% off the position’s salary, and 80% goes to your predecessors. When you retire, 20% of your salary got the previous four years is based on how well the company does the year after you leave.
It strongly incentivizes healthy transitions and long-term plans.