An autonomic computing system is automatically maintained in a steady state. The system has parameters, each of which has one or more threshold. The system may further have influencers, adjustment of which affects the parameters. A value of a first variable specifying how the parameters or influencers are changed once a decision has been made to change the parameters or influencers is received. A value of a second variable specifying how the system and the parameters are affected responsive to a change in the parameters is also received. Where influencers are present, a value of a third variable specifying whether the influencers are distinct or common is also received. One or more of the parameters or influencers are adjusted, based on the values of the first, second, and/or third variables, to return the parameters to more-normal (e.g., median) values.