Every few years captures the excitement of the community

Some community members were curious about what pruning actually is and how it is achieved. Very generally, pruning is the practice of removing blocks from the database, under the assumption that the block structure is not needed for future verification. More specifically, at any point in time there exists a pruning block which is the oldest block kept in the database, along with its UTXO set. Assuming that the UTXO set is valid, one could verify all the blocks above the pruning block (the validity of the pruning block itself follows from consensus).

Pruning is a simple idea which is terribly hard to implement. For one, how does one even design a system where it is guaranteed that old enough blocks are never needed for verification of new blocks? More importantly, how does one implement this idea without opening the hatch for splitting attacks -- attacks which abuse the fact that some nodes might be slightly lagging behind others to create blocks which only a portion of the network will find valid. A simple invalidation threshold rule like "don't accept blocks which point ye deep" can trivially be subject to such deadly attacks, where a single block can split the network in half forever!

When approaching the challenge of designing a pruning protocol, we ran into several barriers. Every new approach we came up with turned out to be subject to more subtle forms of attack. At some point I was almost convinced that this task is mathematically impossible.

But sure enough, after much heated debate and many rushes back to the planning board, we have managed to come up with an efficiently implementable and provably secure pruning scheme for your enjoyment.

For anyone who wants to deep dive into the theory of our pruning mechanism, start here:
https://research.kas.pa/t/some-of-the-intuition-behind-the-design-of-the-invalidation-rules-for-pruning/95

The community is budding. Main-net has been online for half a month, and the computing power of the whole network has been close to 4G/s. At present, only CPU can dig kaspa, and kaspa is not listed on any exchange. Don't be cheated!

I am a community supporter. If anyone is interested in kaspa, please move to
https://bitcointalksearch.org/topic/m.58538650

Topic: Every few years captures the excitement of the community (Read 192 times)