SpaceFarmers.io Wiki

One of the unique features of the Proof of Space and Time (PoST) consensus algorithm is that is it allows to cheaply precompute hashes, rather than computing them on the fly. This allows for a much lower energy footprint compared to other secure Nakamoto chains like Bitcoin or Litecoin that use Proof of Work (PoW). This low-energy feature is colloquially referred to as the “re-usable bingo card”; A rather large table of truly random numbers stored on a drive (HDD/SSD) that responds to the networks call to “who can give me the number closest to this number in a given timeframe?”.

Although pre-computing the hashes is generally the most cost-effective way to respond to these network challenges, from a technical perspective it is still entirely possible to compute this bingo-card on the fly. This would require a supercomputer to brute-force the plot filter in order to create a plot that passes for each challenge. Effectively, it is a trade-off between using the electricity to compute on the fly over and over again, versus computing those hashes once, and storing them on a storage medium for multiple years with a very low energy cost on look-up. Specifically, the cost of creating and storing 512 plots over a drive's lifetime + operating cost < cost of creating a plot on the fly every challenge over a drive's lifetime. The former would be in the realm of $750 dollars, whereas the computer required to create such tables within 30 seconds alone would cost substantially more, currently.

Important determinants for the economic viability of PoST over PoW are the size of the table (k-size) and the plot filter. The k-size determines the drivespace required to save the table, based on the following approximation: 0.78005* k * 2^k = size in bytes (see Chia's PoS paper). i.e. If you reduce the network's required minimum k-size by 1 (e.g. from k-32 to k-31), then the file size shrinks by a factor of ~2, and the computer required to compute on the fly needs to cost only <$375 instead of <$750. A similar effect would occur if the plot filter where to be adjusted from 512 to 256.

note that Chia Network is working on releasing plotformat 2.0; a newplotter that produces >20% improvement in plot compression then what Chia's current plotter produces.

It is important to note that the space-time trade-off does not compromise the security of the network. It only allows to “fake” some space; i.e. allocate a different type of resource to reach the same goal of responding to the challenge (and nothing more: it only makes a plot that passes the filter but CANNOT create a plot that will win every block).