In their ongoing coverage of IBM Edge Dave Vellante and Stu Miniman speak with Jeff Jonas, IBM Fellow and Chief Scientist. The three discuss the success and applicability of his G2 invention, why data errors are sometimes useful and the future of IBM technological innovation.

G2 is designed to integrate very diverse datasets and was recently used in voter registration technology in the last election. Given that people are so mobile, states want to ensure that if you move from one state to another you are not represented on two different voter rolls. Typically, states had been just using name and date of birth to confirm someone’s identity to guard against duplicates. However, these two characteristics did not generate quality output. Given that some states could access DMV data, they could obtain social security numbers, which allows for a more high-quality predictions. G2 was used to compile and assess data to help states increase the quality of the election roll and provide better sense of who has voted. Jonas notes this example represents, “Just the tip of the iceberg in what G2 is going to do.”

G2 can be summarized as a platform that utilizes data  to find data – and determine how found data relates to existent data. Or, as Jonas describes it as “putting a puzzle piece into a puzzle.”

Jonas also notes that errors in data can be useful. He explains that natural variability, as with spelling errors may enhance predictability. When an error is detected, platforms can correct themselves.

Speaking to the future of IBM, Jonas is very optimistic about Watson and Flash and is “excited to see IBM make some big bets on that.” Jonas is not shy about his love of Flash and says he loves to see all systems run on Flash because it enhances velocity.