If we're just talking about one isolated GOE that's out of whack, it's a mistake of some sort.
Data entry error, didn't see something that everyone else saw, forgot the rules applying to that particular element (e.g., 0 or +1 for a nice double jump in a senior ladies' SP), etc.
Or less likely the judge did see something everyone else missed. E.g., that lutz was in the corner on her end of the panel, she could see a severe edge change, the tech panel at the other end didn't have a good view so gave the skater the benefit of the doubt and didn't call e or even !, the other judges rewarded good qualities of the jump.
Judges do make mistakes and do see elements differently. That's exactly why there is a panel and not a single person making the judgments and why high and low scores are dropped for each element.
How much time (and time is money) would it take to research every discrepancy and go back afterward to change the marks and in very close cases possibly the results if a simple error were involved?
Is the public really interested in every detail of every skater's marks? Probably not -- probably only when it affects the medalists or their favorites. And they all have different favorites.
The real problem is if a judge has a pattern of out-of-whack marks. That's what needs to be tracked and followed up on.
If there are just a lot of randomly out-of-line numbers, then the judge is likely incompetent and either needs to be educated or no longer used.
If there's a pattern of consistently favoring certain skaters or certain nationality of skaters and consistently lowballing others, that would suggest cheating.
As I understand it, the system does currently statistically check for national bias patterns and flags those judges for explanations or assessments.
If two or more judges show the same patterns, that might suggest collusion and be reason to investigate. I don't know if there's any means in place for tracking this. If not, there probably should be.
Which is why I think it's more a question of bad PR than bad policy per se.
It might be interesting once to research and break down the discrepancies for the public to explain each out-of-line mark, just to show how innocuous most of them are and give insight into the judging process, the places where there's room for honest disagreement, and the places where innocent errors can all too easily creep in. Could make a good educational feature for the ISU to put out. Do it at an event that doesn't have a lot on the line and therefore little incentive for cheating, so we could have confidence that the judges identified and interviewed would be telling the truth about the reasons for their out-of-line marks .
Hey, ISU, how about doing this at Nebelhorn some year and distributing the resulting written and video report to the media?
But I don't think the public wants resources to be spent on that kind of public nitpicking for every event. Once you understand the process, most of the details are not all that interesting.
When they make a difference in who wins Olympic medals, of course, the public does care. But those are exactly the events where the judges are under the most pressure to deliver results for their federations.
It would be great if judges were allowed and even encouraged to go public in the media after a high-profile event like Worlds or Olympics to identify themselves and explain their judging. But for those judges who choose not to, should we assume that they're trying to hide from us the fact that they cheated or trying to hide from someone else the fact that they didn't?