Making PPLNS Work For Demand Response

Bitcoin mining has come a long way since the days of GPUs and basement set ups. In that time, miners have advanced in countless ways. For example, ASICs are now the standard, not GPUs. Furthermore, enterprise grade players have entered the field, opening new frontiers and bringing with them the size and institutional recognition that opens the doors to otherwise unreachable places for smaller miners. Nowadays, the mining landscape is one where grid services, curtailment strategies, and energy market participation are no longer edge cases but core strategies. As the world around it has moved forward, there’s one question we keep hearing from miners: can PPLNS adapt?

Many miners, particularly those working closely with energy providers or integrating Demand Response mechanisms, have come to view PPLNS with suspicion. They worry that it penalizes downtime and rewards only uninterrupted hashrate—a bad deal for those who routinely curtail machines to support the grid or provide other services.

This fear isn’t baseless. It traces back to a pivotal moment in the mining industry’s recent past, one that apparently sealed the deal for many on PPLNS style payouts: the fallout between RIOT and Braiins Pool.

At the time, Braiins was using the Score payout system. Designed in 2011 by Slush himself, Score was engineered to solve the problem of pool hopping—when miners would jump between pools to exploit reward systems. There’s also been a misconception that Score is a PPLNS style payment system, but as Rosenfeld’s bible on pool payout systems describes, Score and PPLNS are distinctly different payout methods. The main difference is how they account for shares, specifically, Score implemented a rolling window with exponential decay function, this effectively made the lookback window very short. On the other hand, PPLNS is a family of payout systems with various types of fixed length lookback windows.

As shown on this archived website of how Score worked, you can see that after 90 minutes your hashrate had no more presence on the pool. This means that the moment a miner starts mining, their share of rewards fairly quickly reaches the fair value of the hashrate. On the other hand, when a miner stops mining, it drops equally fast, as shown on the gif below.

This might have worked well in the era of cowboys and hackers, but it was never designed with today’s complex mining environments in mind. Certainly not with Demand Response, where miners intentionally and profitably take machines offline to stabilize energy grids or bid into ancillary markets. To Score, that kind of behavior looks no different than a pool hopper—someone attempting to cheat the system.

So when RIOT left Braiins, citing concerns about payout mechanics, it sent a shockwave through the mining world. Due to the aforementioned misconception, Score system’s flaws got unfairly projected onto a broader category of payouts, PPLNS got caught in the fray, catching a stray bullet…