In proof-of-stake, by definition, each stakeholder deposits and locks some tokens (stakes) for more chances to be selected as an author of block in a random-based election procedure.
Till here, it's clear.
But, as the system is not centralized and there is no server which would do the election procedure, each stakeholder installs a client application on their machine to do the random-based selection procedure.
The question is what happens if a stakeholder manipulates the client application, such that he could always be elected as an author of block?
In other words, who (or which entity) does the random election procedure in each round?
P.S. If the question is not still clear enough, please let me know.