Peter Ladkin is a very clever man, and has a good level of understanding of aviation in addition to mathematics.
It seems to me that, although he cites the Bodensee accident, he is simply arguing about the validity of the Eurocontrol assumptions. He does not seem to be saying that RVSM with ACAS is unsafe, but rather that Eurocontrol have not proved that it is safe.
For us mere mortals it is much better to concentrate on our contribution to flight safety than to get involved in semantic disputes over the wording of a safety case. The whole subject of Human Hazard Analysis is, in any case, being reviewed by the JAR authorities anyway.
The prime thing for we pilots to understand is that TCAS is an aid to the 'see and avoid' principle of collision avoidance. This means that it in no way negates our responsibility to listen out, look out, and keep a mental plot of other known traffic. A sharp crew may well spot a possible conflict when it is 'other' traffic, their suspicion will be further aroused when the traffic becomes 'proximate', and, before it becomes a TA, they may well be questioning ATC.Thus an RA is avoided before it occurs. It is also worth pointing out that, in the event of an RA in IMC (ie we can't visually acquire and avoid the other traffic) our responsibilty is to react to the minimum necassary to avoid the conflict. TCAS RAs give you a V/S differential of 1500'/min on your original V/S. More is not better, as that is what will induce the three way conflicts that TCAS is not good at solving.
You can try the above in the sim. Set up a head on co-altitude RA. If you follow the RA by staying just in the green, and returning to your cleared altitude as soon as the TCAS display permits, the altitude loss or gain is only a couple of hundred feet. If you treat it like a GPWS or windshear pull-up (which of themselves are also two subtly differant beasts) you will climb straight into all the other traffic.
The ATC authorities only accepted TCAS on the promise that a correctly followed RA (on a non manoeuvering target) would not compromise the normal ATC altitude tolerance. If an RA occurs, the pooch is already screwed, to unscrew it requires correct, timely, but not over-reaction from both crews.
Offsetting the track is somewhat akin to a security blanket, it will only help in the extremely rare head on co-altitude case. It will not definitevely avoid the triggering of an RA either, due to the poor lateral resolution of the TCAS system. It would not have helped the Bodensee crews, who were on crossing tracks.
To end, I can do no better than to pass on the advice of some of the best brains in the Situational Awareness and Human Error business which was offered at a conference in 2001.
Prof Jim Reason - You are only there to get it right when the 10 to the power of minus 9 event occurs.
Dr Mica Endsleigh - Use Situational Awareness to be pro-active
Tom Seamster - Expert performance requires effortful practise