I clarified my question, by editing the title. Using the comments and links below, I have answered it myself:
The answer to my question is: The similarity is between 98.4%-99.9%, depending on which type of differences you include. The numbers relate to writing my and the other person's DNA as a long string, and compare differences (change of letter, missing/inserted letter). The most important differences seem to be SNPs (=difference in a letter in a gene-coding sequence) and account for 0.1% of difference between our 2 DNA's:
- http://bionumbers.hms.harvard.edu/bionumber.aspx?&id=110248
- http://book.bionumbers.org/how-genetically-similar-are-two-random-people/
I have striked out the original question text, because it includes some motivation and interpretation, which seems false.
edit: It was suggested that the question is a duplicate to "Same" DNA vs genes but this might seem not to be the case, as this question just concerns the used metric (compare answer below).
99.9% vs 88%:
2 person's DNA is 99.9% the same (which goes down to 99% in some sources), e.g. http://book.bionumbers.org/how-genetically-similar-are-two-random-people/
Does this number include differences in copy-number variations and maybe other ?
On the other hand https://www.nature.com/scitable/content/global-variation-in-copy-number-in-the-13571 says that there may be up to 12% of some other differences (I am not sure what those differences mean).
Question:
I have been pouring through the internet and books, but have not yet understood to which metric this 99.9% number refers to? (I am a pure mathematician)
My hypothesis what it means:
If we look at a 1000 base-pair long snippet of the DNA of person A, then we will find this same snippet somewhere also in person B's DNA, with just one difference in one of the 1000 base pairs.
Now, if the latter is true, then this means that my DNA may still be significantly different than my neighbours, e.g. by containing a different number of repetitions in the non-coding DNA.
Ultimately, my DNA could be only, say, 2% similar instead of 99.9% (depending on the number of different repetitions), if we used as a metric the number of single base changes (including single deletions/insertions), we need to do to go from the DNA of person A to the DNA of person B (assuming we wrote down each other's DNA in one long string). Question 2: In fact, I would be quite interested to know how much the DNA of two people is similar using this last metric?