I get to spend the day troubleshooting a problem with the ECDL testing site they use here. Just worked out (after much pain and anguish) that running a join.me session seems to stop the error, which is why the remote support have never been able to replicate it. Bloody marvellous!
Seriously, though, I have reimaged this entire room, with a brand new image, up to date & identical drivers, software and updates across the room, and somehow, SOMEHOW, the same half dozen machines still display the same intermittent and random issue. How is that even possible? Am I being trolled by motherboards here?