Autoland Longitudinal Scatter Distance
I too recall this term during my L1011 time but I never came accross any claims it was a deliberate bias or such. Others have touched upon the real reasons.
Given that the aircraft does not follow the G/S all the way to touchdown; once it starts the flare, it will of course depart the G/S and is no longer following any vertical guidance and so be subject to, as others have indicated, many variables. It could of course then be subject to float with no auto method of correcting, except perhaps a little from the autothrottle if operating and then usually only for speed control. During a manual landing, the pilot could "cheat" a little if he gets concerned.
The landing performance tables/graphs for that aeroplane and every other machine with autoland that I have flown added an additonal distance to the field length required for an autoland. And this was because of the imprecise touchdown point but usually it addded about 1100 to 1300 ft to the calculations could be much more in some cases or a bit less. This being based upon the G/S angle and threshold crossing height of the G/S signal.