For Anthony Giddens, what is more significant is the logical conjunction of the notion of human agency and that of power. He writes:
“Action intrinsically involves the application of ‘means’ to achieve outcomes, brought about through the direct intervention of an actor in a course of events, ‘intended action’ being a sub-class of the actor’s doing or refraining from doing; power represents the capacity of the agent to mobilize resources to constitute those ‘means.’ In this most general sense, ‘power’ refers to the transformative capacity of human action.” (New Rules)
For Giddens, the understanding of power as domination is narrower than his general conception of power as transformative capacity. Furthermore, in his view, “the relation between power and conflict is a contingent one.” He continues:
“If power and conflict frequently go together, it is not because the one logically implies the other, but because power is linked to the pursuance of interests, and people’s interests may fail to coincide…Power is a feature of every form of human interaction, division of interest is not.” (New Rules).
For Giddens, the reproduction of a domination structure during interaction does not necessarily imply that there exists a conflict of interests among the interacting agents. There are contexts when the interests of a dominant person (or group) and a dependent person (or group) do coincide, and although such contexts are perhaps less common than not, they can be significant enough not only for sociological analysis but also for anticipating social changes.
For Giddens, “power relations are relations of autonomy and dependence, but even the most autonomous agent is in some degree dependent, and the most dependent actor or party in a relationship retains some autonomy” (Central Problems). All forms of dependence make some resources available to subordinates for influencing the activities of their superiors. Giddens calls this notion the dialectic of control in systems of interaction.
Giddens’ conception of the dialectic of control recognizes that power relations are not constant-sum games. Social power is not a fixed quantity that gets divided between the interacting parties so much so that when one party becomes more powerful the other party becomes weaker correspondingly.
The dialectic of control is open to a wide range of possible effects: (1) both interacting parties become more powerful, though usually to different degrees; (2) one party becomes more powerful while the other becomes weaker; or (3) both parties becomes weaker.