I think it has to do with rigged tests and tests that do not represent daily carry wear. I also suspect as people say here bad batches of low quality saphire compared to the S0 Saphire glass used initially.
Here’s Consumer reports failing to scratching the Series 3
https://www.consumer-reports.org/video/view/electronics/news/5595713574001/apple-watch-series-3-final-test-results/
Kind of hard for Jerry Rig guy to rig a test when he’s just using a Mohl hardness pick. Just odd nothing lines up.