Quote from: JoelKatz on June 17, 2017, 03:59:16 PM

Ripple's design does not require history for the system to make forward progress. Storing historical data is always optional.

In the very early days of Ripple, there were only three servers running. All were validators and all were on Amazon's EC2 platform. They were essentially identically configured.

Early versions of rippled did not check disk free space and would continue running even if the disk got full. In addition, they didn't save the ledger header into the node store. All three validators ran out of disk space at substantially the same time.

I took snapshots of the databases of all three validators and attempted to recover the missing ledgers. As I recall, I was able to recover several thousand ledgers and got stuck at 32,570.

All of this occurred before the network was open to the public, before XRP was traded, and before XRP had any value at all.

After the network was opened to the public, several volunteers offered to assist with the recovery process and we made the snapshots from all three servers public. No additional ledgers were recovered.

So those change in number of coins aren't related to the consensus rule?

By the way, i just read this post in reddit, beautiful: https://www.xrpchat.com/topic/6362-the-reason-why-xrp-will-be-adopted/