The original discussion was about the possibility of duplicate ids or aliases on the database, and how this could be avoided by cross matching the master record with the database. The master record is not a low quality scanned record as taken in a bank or school for example of your index finger, it is a high quality measurement of your fingerprints and other biometrics. What you are going on about is identity verification which poses considerable problems in practice and is a separate issue (although related) to the quality of the database.
You have misunderstood the statistics quoted from the Biometric Enrolment Trial. Those verifications were not performed by asking people to have their biometrics scanned on second-rate scanners in remote locations. They were taken using the same scanners that had been used to enrol them on the system, in the same room, under the same conditions. The verifications that failed under those conditions could not be expected to work under any other. If those "verifications" had been attempted second enrolments by those people, they would not have been detected as duplicates.