Data, data - you say genuine, I say véritable
Every DFDR/FDR data output set that I have ever laid eyes upon contains data points that do not fit the overall picture. Some are spurious and unavoidable due to monitoring and data collection system limitations and are rejected unless they fit a pattern under investigation. Other odd bits are known transducer noise anomalies. In a highly digital aircraft such as the Airbus or most modern Boeings, most of the raw transducer anomalies are filtered before they get to any exterior data bus. The DFDR, by virtue of its design, recording limitations, and things like the time slice paradox, does a bit of filtering itself.
Now, take what is left, and try to paint a perfectly accurate picture of exactly what was happening and when - it can't be done, however a reasonably accurate representation of the various parameters is what we have to work with, and in most cases, this is enough for forensic analysis.
On to the public release - the mere act of taking those fiddled numbers and placing them on a chart is once again a manipulation of the raw data. Subtle, and mostly innocuous, but the movement away from fidelity is incontrovertible and obvious to those who have seen the raw traces.