I have some heartburn with Baysian approaches, because they require assumptions that should vary if you want to proof a theory. (A statistical approach can wash-out an important observable.)

Still, it is encouraging to learn such a treatment of the data is not able to conclude the supernova population is evolving.

__________________
jwj

It's ok not to know. We should try harder to find out.

Jerry, you have a few years to sharpen up your 'radical sceptic' critique of the combined results of supernova science this past half-century or more, and even describe - quantitatively, perhaps even with simulations - the holes in the LSST's announced plans for supernova science (use the 137.11 Olivier et al. link; the 137.10 one doesn't seem to work).

I for one would be curious to learn of any significant quantitative¹ holes in a program that, if implemented as planned (and combined with an adequate spectroscopic follow-up, of course), should permit extremely robust testing of a great many hypotheses concerning supernovae ...

¹ Please, no more word-salad handwaving.

I can't open any pdf files right now (help!?); so I can't see if they are addressing my biggest concern. In manual searches for supernovae type Ia candidates, events that are 1) observed at high redshift and 2) have light curves that are too small relative to local events AFTER the time dilation factor is figured in are not included in follow-up observations (See Adam Riess, 2004 http://xxx.lanl.gov/PS_cache/astro-p.../0402512v2.pdf).

Since you like numbers, in current standard candle studies, if a supernova Ia-type event is found at a redshift of z=1, the lightcurve must last at least ~ten days before losing one magnitude (in blue) to be considered 'in family' with local supernovae. But if the time dilation correction is wrong, and that is what I am contending, a light curve of only five days at a redshift distance of z=1 after redshift corrections is equivalent to a local lightcurve that lasts ten days.

If the same criteria is used in automated searches, they will be artificially cutting events that are very like the local sample (if and only if the redshift time dilation correction is incorrect.) So an automated survey that does not allow for a major flaw in established principles may unwittingly produce a very highly skewed sample, cut at the 'short' end, where the best nullifying evidence of the Wilson hypothesis should be found.

Again, I cannot open ANY PDF files, so I can't tell if Kuznetsova et al have addressed this issue.

__________________
jwj

It's ok not to know. We should try harder to find out.