Last year ScouB KVD and Sami went out at the same stage last year. With last man standing at least two of them would have progressed.
I believe that system would allow the big guns to race conservatively and only really push the limits when the later stages arrive.
A format that maximizes variance is not the most desirable. One format punishing mistakes harder than another does not make it better. Consider an extreme example: the entire TT competition is one try on one random track, with the results from that one track determining the entire ranking. This format is very high variance and would lead to many upsets, but it should be clear that this is not a desirable format for CDM.
As mentioned in the opening post, the current format is unsatisfactory for many (in particular those making the Top 16 and beyond), in part because its variance is too high. Any format used in a competitive environment has to balance between determining the relative skill levels of participants and satisfying the event's constraints (time and space being the two major ones). That second part naturally introduces variance into the competition, but in principle the format should be attempting to minimize the amount of variance in play (i.e. if player A is better than player B, then A should be ranked above B as often as possible).
Of course, CDM also has some extra constraints due to the fairly unique nature of the event. In particular, it's not a purely competitive event: we try to cater to the enjoyment of all participants, and take into account how fun a given format is to play in and also, to a lesser degree, how fun it is to spectate. This is apparent in the Belgian System, especially with the Shakedown.
I think that the LMS system retains all the qualities of the current KO system while mitigating some (but not even close to all) of its variance, as well as resolving a couple additional issues (namely the KO stages being too short, and the incentive to peek at others' results).
I would also like to point out that I am proposing these changes as someone who has directly benefited from the current format, and that I would rather compete in a format that resolves the current one's issues while retaining the core values of the TT event (cutthroat one-try competition against the field).
in my opinion, it would be fair and coherent to use the same system in playoffs than in group stage : each player get a number of points depending on his rank on the track, and half of the players are qualified depending on the number of points they earned. they would play on 2 tracks each time, except 3 tracks for the finals.
so that there would be an incensitive to get a good rank, but a bad chrono wouldn't be as awful as it is now (it's more possible to come back if you are 10 points behind than if you are 30sec behind)
I would argue that this does not resolve any of the issues with the current format. If anything, the switch from total time to AF would change the emphasis of where variance occurs from complex tracks (like GV3) to simple tracks (like GV1), where a difference of a couple tenths can cost you a ton of AF and lead to unrecoverable situations.