By early March, he and Møller had figured out another optimization trick to vastly speed up their queries of the blockchain. On one of their pine-forest walks, Møller had realized that labeling every transaction with their own chronological identifiers rather than Bitcoin’s native transaction IDs would reduce the size of the data their software needed to analyze by as much as 90 percent. That meant they could store the database of all the blockchain’s transactions entirely in a PC’s memory, rather than on its hard drive.
