However, we also don't really know how many mutations are necessary for cancer to arise, and one of the first things to happen may be either immortalization through mechanisms including telomerase activity (or the tumor could arise in stem cells where this activity is already present) or (more often) a mutation that in turns leads to an increased mutation rate! Regardless, it's probably not that many mutations, and don't underestimate just how large a number of cells there are in the body (on the order of 10^13 - 10^14)!
In humans, telomeres limit cells to ~50 divisions, which is probably related to how DNA replication is only 99.9998% accurate. After that many divisions, the genome is 0.001% different from when it started, which is one error per 10,000 base pairs, or an error in 1/3 of all genes. This is in addition to the slow rate of spontaneous mutations you accumulate over your lifetime.
Where did you get your numbers? Human DNA replication (in normal cells with no damage) is 99.99999999% accurate (i.e. about 1 mutation per 10^-10 base pairs). Please do not mod parent informative for this misinformative post!
While the parent is correct that 24 SNPs is sufficient in a given population, in practice it's probably hard to choose 24 SNPs that cover ALL populations in the world well (since a SNP with a high minor allele frequency in, say, Europeans, may not have a high minor allele frequency in Asians, or Indians, or Australians...