OverRun, I concur with your concerns... had I been able to spare the time to attend, I would have caught up with you during the recent Perth Bash ... however, with my present job, it is now likely that you have more occasion to travel than do I ..
Some thoughts which might be of interest to those following the discussion ....
(a) there is not much of a problem associated with more or less maintaining or hitting the centreline when things are going well ... why, I can even recall the odd landing where I managed both the centreline and a smooth touchdown ...
(b) my reading of the tea leaves is that the ICAO numbers address routine geometric considerations but give inadequate consideration to the reasonably anticipated centreline deviations in emergency situations.
(c) for the takeoff failure case, the main problems are
(i) proximity at Vef to Vmcg .. at speeds reasonably well above Vmcg, the deviation is minimal for a competent pilot. However, as the failure speed reduces toward Vmcg, the deviation becomes rapidly and entertainingly significant ... tending, in the limit, to a very rapid and total loss of directional control and a walk in the grass ... in the absence of a throttle chop.
(ii) such involvement as I have had in this work suggests that the continued takeoff is more critical than the reject
(iii) the control effects significantly relate to (aft) CG, and nosewheel cornering force capability... in the testing programs, the case was taken to the certification extreme of having no steering capability to address runway surfaces with little or no usable friction (wet, very slippery, ice, etc.)
(iv) the effect of crosswind is either stabilising (downwind failure) or destabilising (upwind failure). As we generally don't get into guessing which side will have the failure, it is reasonable to consider the destabilising case. It should be kept in mind that Vmcg is determined for nil crosswind (US) or 7 kt (older UK standards).
The typical effect of crosswind on Vmcg varies from around 0.5 kt/kt (twins) to well in excess of 1 kt/kt for 4-motored beasties .... considering that we cheerfully depart in crosswinds of 20-40 kt depending on Type ... this might easily mean, during a Vmcg-limited V1 schedule, that the pilot is faced with a failure many knots below the REAL Vmcg of the day.
With the training emphasis on doing things by rote, this would easily result in a very surprised line pilot doing all the "right" things while watching the aircraft rapidly depart the runway edge into a ground loop with probable catastrophic results. This is one of the reasons why, when I was involved with training programs, I always put the trainee in this sort of situation to emphasise the need to consider the interaction of minV1 and crosswind ... the effectiveness depended on how the sim was set up .. but, in those which were set up well ... the lesson was salutory ..
(d) for the landing case, the main concerns are
(i) pilot mishandling which can only be addressed by training and concurrency requirements, assuming a basic skill level in the individual
(ii) misalignment off an instrument let down .. typically a non-precision runway approach in minimum vis situations. For a reasonably experience and competent pilot, the resulting sidestep manouvre, while interesting, need not necessarily be hazardous providing that the missed approach is exercised if the misalignment is more than minor or the profile cannot be maintained during the sidestep.
Following on from OverRun's comments on airports engineering's penny-wise, pound-foolish philosophies, I can only concur ....
(a) if nothing goes wrong during the operation, then we don't need more than a minimal pad outside the mainwheel footprint.
however ...
(b) if it all turns to custard, the more the better .... I am reminded of a colleague's report relating to his transition to egg-beater machines in respect of the initial hover training ... something along the lines of ... "I started off in a big area in a big aerodrome .. but needed the whole of Queensland before I got it under control ... "
(c) if we go down the path of not worrying too much about the infrequent extreme situation when it comes to airports engineering practice, then we might as well not bother with any training .. etc. ... and just recruit PacMan-proficient 6 year olds to fly our airline transports.