Short answers, pulled from the story.
Demand response is a change in the power consumption of electric utility customers to better match the demand for power with the supply. Rather than adjusting how much electricity generators produce, it adjusts how much consumers draw from the grid, typically in response to price signals or direct utility requests.
Ahmad Faruqui of the Brattle Group estimated that a 5% reduction in US peak electricity demand could produce approximately $35 billion in cost savings over a 20-year period. Two Carnegie Mellon studies from 2006 found that an approximately 10% reduction in peak demand could yield system savings of between $8 and $28 billion.
FERC Order No. 745, issued in March 2011, required compensation for providers of economic demand response participating in wholesale power markets. Critics including Professor William W. Hogan at Harvard University's Kennedy School argued it overcompensated providers and functioned as a buyer's cartel. The DC Circuit vacated it on the 23rd of May 2014, but the Supreme Court reinstated FERC's authority in a 6-2 decision on the 25th of January 2016.
In Ontario between August and September 2006, wholesale prices paid to producers ranged from a peak of $318 per megawatt-hour to a minimum of negative $3.10 per megawatt-hour. A negative price meant producers were charged to supply electricity, and consumers on real-time pricing may have received a rebate for consuming power during that period.
Energy efficiency means using less power to perform the same task on a continuous basis whenever that task is performed. Demand response means adjusting or shifting consumption specifically in response to grid conditions or price signals, without necessarily reducing total consumption over time.
Industrial customers including aluminum smelters and data centers are active participants. Alcoa's Warrick Operation participates in MISO as a qualified demand response resource, and Trimet Aluminium uses its smelter as a short-term nega-battery. Some data centers migrate computing loads between geographically separated sites to participate in demand response while maintaining redundancy.