In management, the level up to which someone would raise in the organizational ladder is governed by the Peter Principle, named after the Canadian psychologist Lawrence Peter. Consider a team of dozen employees out of which Alice is the most competent. So Alice gets promoted to the next job level. Now given the fact that Alice was the best among her contemporaries does not mean that she would be the best even in the current job - which is, say, supervising those who earlier worked with Alice. Peter's Principle says that all new members in an organization climb up the ladder until they reach the maximum incompetence.
Common sense tells that in any organization, the person who performs the best in a job has to be promoted to the higher level. But most of the time, the kind of job involved in one or two level above would be radically different from the current job responsibilities. So the person who does best in one job may fail big time in the other. For example, Michael Jordan was able to accomplish great things as a player of Chicago Bulls. But he did not achieve the same amount of success as the coach of Washington Wizards. Other better examples could be seen in your day to day lives.
Two solutions proposed by Peter for this are: (i)increase the pay for the best performing employee rather than immediately promoting them. As a result, some lower level employee may earn as much as their boss. This also requires a psychological change in boss' side. (ii) train the potential candidates for the job and promote on the basis of their performance in the training program. The success of this depends upon how good one can devise that training program. Personally, I think both of these can be applied concurrently.
But a recent paper by Alessandro Pluchino et al published in a journal of computer science and game theory proposed two dangerous solutions. Pluchino and his team has performed an agent-based simulation of a hierarchical organization in which the promotions are offered on the basis of performance at a given level, and the job responsibility differs at each level. Based on these results, this paper offers two unconventional solutions. First solution is to promote the most competent and the least competent alternatively. Second solution is to promote people in random without considering the competence at the current level.
These techniques might have worked wonders among the agents. But humans are complex entities. The psychological effect of these solutions cannot be overlooked. First if the person is to be promoted in random, then there is no incentive to be competent. This effect may spread through the entire organization making it a zombie. Instead if the most competent and the least competent are promoted alternatively, this gives an incentive for those who are below average to become intentionally and glaringly incompetent. Most of the time, the failure of the theoretical model in a recurring system can be ascribed to its failure to consider the practical implications of its first round implementation on the subsequent rounds.
Even Hari Seldon made a reservation that future cannot be predicted and announced, because of the human nature to act according to the result of the prediction and there by falsifying it. Unless this human nature has to be considered in these kind of simulations, they would never become practical.