I have a few questions for discussion:

- Figure 1. b: It would be interesting to see how does this Bianchi template looks in more neutral coordinates (say with axis to the pole). As far as I can gather these models produce a spirally looking pattern and my question is whether such pattern could result as an artifact of the scanning strategy. If this is the case, the the "telescope" would add this kind of template and then, of course, you will get all those cold-spot detections, assymmetries, etc. without them really be there;
- The WMAP best fit theoretical squadropole of 869 [tex]\mu {\rm K}^2[/tex]: surely this is the running spectral index value? And the n_run was there to fit the low quadrupole (among other things), so it would make more sense to compare it to the pure LCDM of 1150 or so [tex]\mu {\rm K}^2[/tex]. BUt yes, anything higher than 300 is OK with normal cosmology.
- This model brings 6 or 7 new params, which is quite a lot. Looking in pixel space, after subtracting template, by how much does the probability that what remains is a realisation of the pure LCDM increases? I am slightly confused whether this kind of analysis is neccessary, if we can just get rid of all NG detections by introducing 5 extra params? How does one go and quantify the improvement of fit in such cases...
- Finally, and most importantly: it seems that we cannot really get considerably higher statistical significance from CMB alone. If the universe is indeed described by such models, what other observational signatures could one get? Could you discard such models on the basis of LSS considerations?