A well-written blog account of a current controversy in human genetics, by Joe Pickrell: "Questioning the evidence for non-canonical RNA editing in humans".
The observation that I personally found most convincing is displayed in the plot at the beginning of this post. What I’m showing is that mismatches to the genome at RDD sites occur almost exclusively at the ends of sequencing reads. All three technical comments include this observation. Importantly, Lin/Piskol et al. take this analysis one step further. They show (in their Figure 2) that this effect is driven by the fact that mismatches to the genome at RDD sites tend to occur at the beginning of sequencing reads that go in the opposite direction of transcription (this effect is masked in my plot).
It's a bad sign when 90% of your observations may result from sequencing errors. That's something we spend a lot of time trying to understand and work around in the archaic human genomes. We frequently find that, while the genetics ought to follow a mathematical model perfectly well, the sequence data are noisy in ways that interfere substantially with our predictions.
It's the same thing as using bad wiring in a neutrino experiment, really. If you know about it, you can work around it. Otherwise, it's liable to mislead you.