If they used a good hash it would be basically impossible. I'm thinking someone got sloppy and used a hash that wasn't so good, or used only part of a hash. (I've done the latter deliberately--folded the hash down to a 32-bit value accepting that a "hit" meant go to the disk and get the whole record to check. The occasional false positive cost less than the additional time spent reading the bigger hashes as I had millions that needed to be loaded into memory.)
Not impossible they do something like just name it after the time stamp and add some random digits - I’ve seen exactly this at a telco. “What about collisions?” “We’ll add another digit if it’s a problem.”
So maybe it’s now a problem. Although with decreasing MMS volume you’d think not.
Comments
The two images have the same hash? (~Impossible, I know.)
If they used a good hash it would be basically impossible. I'm thinking someone got sloppy and used a hash that wasn't so good, or used only part of a hash. (I've done the latter deliberately--folded the hash down to a 32-bit value accepting that a "hit" meant go to the disk and get the whole record to check. The occasional false positive cost less than the additional time spent reading the bigger hashes as I had millions that needed to be loaded into memory.)
Not impossible they do something like just name it after the time stamp and add some random digits - I’ve seen exactly this at a telco. “What about collisions?” “We’ll add another digit if it’s a problem.”
So maybe it’s now a problem. Although with decreasing MMS volume you’d think not.
Or someones GUID/UUID implementation is not as random as they thought.