It can be used more empirically. The more matching points, the less likely a given match is down to random luck. A smudge will have relatively few useful points, and so far less reliability.
The big problem is you can either use fingerprints to identify someone to a scene, or to search for a match in a database. 1 in a million sounds impressive to a jury, and is in a 1 to 1 test. Compare it to a database of 10 million, and you will expect to get 10 matches by random chance alone.
It can be used more empirically. The more matching points, the less likely a given match is down to random luck. A smudge will have relatively few useful points, and so far less reliability.
The big problem is you can either use fingerprints to identify someone to a scene, or to search for a match in a database. 1 in a million sounds impressive to a jury, and is in a 1 to 1 test. Compare it to a database of 10 million, and you will expect to get 10 matches by random chance alone.