functionality of -b option alone?

I am trying to understand what the -b option does when it is not paired with the -c option. I am working with a 4,640,668 bp long alignment.

When I run snp-sites on the alignment without either the -b or -c options, I get a resulting alignment of 1,733 sites. My understanding is that these are all of the variant sites in the full alignment, regardless of whether or not there is missing data (N, -, or ?) in some samples.
When I run snp-sites -c, I get 944 variant sites (ACGT-only sites), which implies there are 789 variant sites with missing data in at least one sample (1,733 - 944 = 789).
When I run snp-sites -cb, I get an alignment of 2,903,621 bp. My understanding is that this is the ACGT-only sites plus the monomorphic sites (944 + 2,902,677 = 2,903,621).
Based on the above logic, I assumed that running snp-sites -b would give me all of the 1,733 variant sites (both ACGT-only and those with missing data) plus the monomorphic sites (1,733 + 2,902,677 = 2,904,410).
However, when I do run snp-sites -b I get the complete 4,640,668 bp alignment.

Am I missing something about what the -b option is doing? Any help would be much appreciated. Thank you!

sanger-pathogens / snp-sites