Yes Neil that's the offending item, but make no mistake, I have very low expectations of this machine, so not quite void you're quite imagining! 
Anyway, from one end of the table to the other the deflection currently is 12.5mm. I could work round this by setting the fence at the matching angle as per the video above, but unfortunately this only goes on somewhere between almost straight, straight, and almost straight on the opposite side. It cannot be locked down on an angle like that as it just pulls itself towards straight. I'd have to add a wedge shape that corresponds to the deviation of the blade in order to make it useful, and bolt it onto the fence. Or possibly see if I can modify it to be able to go at any angle. Not sure it'd be very easy though. One end of it (opposite the lock handle) could be given a radius where it pulls against the table but the handle end has a piece half and inch or so each side that are cast, and ensure it sits more or less square. I think to get round it I'd have to get rid of those and add a central curved clamping point somehow.
Another issue with it is that it's just plain sh*t! The fence once locked down is locked in place, but not properly secure against the table so it can rock/tilt, making it mostly a waste of space. I've tried modifying it to correct this but so far it hasn't worked. I might add some sort of shims beneath it to try and prevent it.
Here's the culprit and it's 'deviant' factor though:
