This paper discusses the concept of actively involving highly distributed loads in power system control actions. It provides an overview of system control objectives, including economic dispatch, automatic generation control, and spinning reserve. The paper reviews existing initiatives that seek to develop load control programs for power system services and discusses challenges in achieving a load control scheme that balances device-level objectives with power system-level objectives. The paper argues that direct load control is required to enable fast time-scale, predictable control opportunities, especially for ancillary services like regulation and contingency reserves. It discusses centralized, hierarchical, and distributed control architectures, their benefits and disadvantages, and implications for communications infrastructure. The paper also explores the potential of thermostatically controlled loads and plug-in electric vehicles for fully responsive load control. The paper explores the challenges of achieving fully responsive, nondisruptive load control, which requires balancing system-level and local control objectives. It defines "fully responsive" as enabling high-resolution system-level control across multiple time scales and "nondisruptive" as having an imperceptible effect on end-use performance. The paper discusses the advantages of using loads for system services, including their ability to provide spatially precise responses to contingencies, reduce grid emissions, and support the integration of intermittent renewable electricity generators. It also highlights the importance of communications infrastructure in enabling load control and the potential of advanced metering infrastructure (AMI) in this context. The paper discusses the challenges of integrating loads into the power system, including the need for reliable communication networks, the potential for response fatigue, and the importance of customer acceptance. It also explores the differences between price response and direct control, noting that direct control is more suitable for fast time-scale services like regulation and spinning reserve. The paper presents examples of uncoordinated load control, showing how lack of coordination can lead to unexpected collective behavior. It also discusses load control metrics, the choice of input signal, and the potential for distributed control strategies. The paper concludes that fully responsive load control is a promising approach for providing power system services, but requires careful consideration of system-level and local objectives, as well as the development of appropriate control strategies and communication infrastructure.This paper discusses the concept of actively involving highly distributed loads in power system control actions. It provides an overview of system control objectives, including economic dispatch, automatic generation control, and spinning reserve. The paper reviews existing initiatives that seek to develop load control programs for power system services and discusses challenges in achieving a load control scheme that balances device-level objectives with power system-level objectives. The paper argues that direct load control is required to enable fast time-scale, predictable control opportunities, especially for ancillary services like regulation and contingency reserves. It discusses centralized, hierarchical, and distributed control architectures, their benefits and disadvantages, and implications for communications infrastructure. The paper also explores the potential of thermostatically controlled loads and plug-in electric vehicles for fully responsive load control. The paper explores the challenges of achieving fully responsive, nondisruptive load control, which requires balancing system-level and local control objectives. It defines "fully responsive" as enabling high-resolution system-level control across multiple time scales and "nondisruptive" as having an imperceptible effect on end-use performance. The paper discusses the advantages of using loads for system services, including their ability to provide spatially precise responses to contingencies, reduce grid emissions, and support the integration of intermittent renewable electricity generators. It also highlights the importance of communications infrastructure in enabling load control and the potential of advanced metering infrastructure (AMI) in this context. The paper discusses the challenges of integrating loads into the power system, including the need for reliable communication networks, the potential for response fatigue, and the importance of customer acceptance. It also explores the differences between price response and direct control, noting that direct control is more suitable for fast time-scale services like regulation and spinning reserve. The paper presents examples of uncoordinated load control, showing how lack of coordination can lead to unexpected collective behavior. It also discusses load control metrics, the choice of input signal, and the potential for distributed control strategies. The paper concludes that fully responsive load control is a promising approach for providing power system services, but requires careful consideration of system-level and local objectives, as well as the development of appropriate control strategies and communication infrastructure.