Indeed, that's precisely what I meant by frequency regulation via demand control and batteries.
The problem I was thinking about was demand harmonics.
It's not unreasonable to envision a scenario where a large number of independent devices, acting in their own self interest, end up oscillating between on and off together. E.g. by using the same logic.
It seems like you need something approaching a parallelism primative to coordinate behavior. E.g. bid two steps in advance, find out if your bid was accepted one step in advance, then act.
Or is bidding in energy markets already complex enough to avoid that outcome?
Frequency regulation is, however, a relatively small amount of power, and the grid itself needing to run at a particular frequency is more the result of loads that rely on the grid frequency for timing (i.e. A/C electric motors in fans).
In the future the grid frequency could be allowed to vary much more if loads become resilient to frequency variations.
The bigger (in terms of total energy) issue is shifting load to match supply, or vice versa. That can be achieved with appliances that store energy in one form or another. One of the most interesting examples of this is generating ice with wind energy at night to be used for cooling during the day.
Instead of spinning up a natgas turbine, why not let my water heater or car charger take a break as a part of the market?
There are some grid level implementations of this (install our thermostat that we can remotely turn off to get a 5% rebate), but these seem to work around the primary problem: we pay an average price for electricity, not the actual price.
And if it all works based on grid frequency, the system could even work offline: water heater sees frequency drop and takes a random break between 1 and 60 minutes.
Or variable-frequency drive AC drops 10% in intensity, or increases by x% when frequency is getting a little high.