Have you seen previous years' spread?
For instance, 2018's looked like
this.
Here's
2017.
Here's
2016.
And here's
2015.
Looking at them on a comparison basis, even ignoring the slightly different format each year, there are greater differences in the bell curve than I would have expected to see. Also interesting to see that last year the average score was 112, and previous years it has been 100. That seems odd, although it was the first year with GL.
It may well just be a statistical coincidence given how much variation there seems to be in the "shape" each year. That said, given the controversy over the VR paper and the increased likelihood of appeals, I can see how a theory could develop that they're reducing the amount of children in the likely-to-appeal range. I can't see how they could do that though - children score what they score. I'm no mathematician but while you can adjust the cut-off point for qualification, you can't change the shape, surely? If fewer children have scored in that range (assuming it is true) then it can only be a coincidence. Happy to be corrected though.