Your analysis seems about right, if slight adjustments are made to your numbers...
1. The last common ancestor of bacteria and humans is surely far older than the Cambrian. Plentiful fossil evidence exists of metazoan life in the preCambrian. Lets adjust your 600 million number up to about 2 billion years.
2. The number of genes is debatable, as you pointed out, so there is no reason to quibble about 30K or 100K or 66K genes. We can use your 66K number - if we are wrong, it is not by more than a factor of 2 or so.
3. It is definitely a mistake to assume that all speciation events and distinctions between organisms are the result of entirely NEW genes. IIRC, chimps and humans share almost 100 percent of their genes - there are no genes that chimps have that humans don't have, and only one gene that humans have that chimps don't have. Of course, there are small differences in many of these shared genes at a base-pair level.
4. So your statement...
"A more universal agreement among genome researchers is the number of DNA base pairs 3.1 billion in humans and 2 million in E. coli." ... is probably a more accurate gauge of the long term changes in genomes between bacteria and human. Again, some of your numbers probably need some minor tweaking, but the final munber -
"each isolated race line should have added 500,000 different base pairs" - is probably pretty close.
Your incredulity...
"Surely, this is did not happen! " ... is insufficient to prevent it from being essentially true. Perhaps you should have done some research first to find out the actual level of diversity in the human genome. Do a web search on "SNP human genome"
for example:
"...brings the total number of SNPs in the Celera SNP Reference Database to 2.8 million. The average SNP density along the human genome is one SNP per 1,250 base pairs."
from
http://www.pecorporation.com/press/prccorp091300.html
So your 500,000 number is a little low. But not bad for a beginner.
And welcome to the EvCforum.