I think you're exactly right Ami. While I think playing without aTMA is the way to go, the reality is that players need it, either because they're new or because they need to spend the time doing other things. Those players deserve to have aTMA, and the rest of us deserve a game in with aTMA can be used without it cheating.
As for 10-20% error...this concept is OK with me, I think that's about how aTMA worked in Sub Command... I would prefer an aTMA that started out sloppy but became refinded based on the data provided, that is it gets better with time, gets better with DEMON data, gets better with more legs, gets better on lag LOS, gets worse on lead LOS...etc. Maybe that's asking for too much, but I think they already have that now, or at least they've claimed that they have implemented such a program...its just that the way that program is confined to the real data prevents us from ever seeing it in all its splendor.... Kinda like the hard limits on sensors keeping the rather nice acoustic model from being noticed until those limits are removed...
