Complexified octonions have led to a dramatic progress in the understanding of TGD. One cannot however avoid
a radical question about fundamentals.
 The basic structure at M^{8} side consists of complexified octonions. The metric tensor for the complexified inner product for complexified octonions (no complex conjugation with respect to i for the vectors in the inner product) can be taken to have any signature (ε_{1},...,ε_{8}), ε_{i}=+/ 1. By allowing some coordinates to be real and some coordinates imaginary one obtains effectively any signature from say purely Euclidian signature. What matters is that the restriction of complexified metric to the allowed subspace is real. These subspaces are linear Lagrangian manifolds for Kähler form representing the commuting imaginary unit i. There is analogy with wave mechanics. Why M^{8} actually M^{4}  should be so special real section? Why not some other signature?
 The first observation is that the CP_{2} point labelling tangent space is independent of the signature so that the problem reduces to the question why M^{4} rather than some other signature (ε_{1},..,ε_{4}). The intersection of real subspaces with different signatures and same origin (t,r)=0 is the common subspace with the same signature. For instance, for (1,1,1,1) and (1,1,1,1) this subspace is 3D t=0 plane sharing with CD the lower tips of CD. For (1,1,1,1) and (1,1,1,1) the situation is same. For (1,1,1,1) and (1,1,1,1) z=0 holds in the intersection having as common with the lower boundary of CD the boundary of 3D lightcone. One obtains in a similar manner boundaries of 2D and 1D lightcones as intersections.
 What about CDs in various signatures? For a fully Euclidian signature the counterparts for the interiors of CDs reduce to 4D intervals t∈ [0,T] and their exteriors and thus the spacetime varieties representing incoming particles reduce to pairs of points (t,r)=(0,0) and (t,r)= (T,0): it does not make sense to speak about external particles. For other signatures the external particles correspond to 4D surfaces and dynamics makes sense. The CDs associated with the real sectors intersect at boundaries of lower dimensional CDs: these lowerdimensional boundaries are analogous to subspaces of Big Bang (BB) and Big Crunch (BC).
 I have not found any good argument for selecting M^{4}=M^{1,3} as a unique signature. Should one allow also other real sections? Could the quantum numbers be transferred between sectors of different signature at BB and BC? The counterpart of Lorentz group acting as a symmetry group depends on signature and would change in the transfer. Conservation laws should be satisfied in this kind of process if it is possible. For instance, in the leakage from M^{4}=M^{1,3} to Mi,j, say M^{2,2}, the intersection would be M^{1,2}. Momentum components for which signature changes, should vanish if this is true. Angular momentum quantization axis normal to the plane is defined by two axis with the same signature. If the signatures of these axes are preserved, angular momentum projection in this direction should be conserved. The amplitude for the transfer would involve integral over either boundary component of the lowerdimensional CD.
Final question: Could the leakage between signatures be detected as disappearance of matter for CDs in elementary particle scales or lab scales?
See the article Does M^{8}H duality reduce classical TGD to octonionic algebraic geometry?: part II.
