To put it in simplest terms, similar shapes fill the same hole, and target similar locations. A good example is the way mutations occur during replication.Why can't chemistry be consistent?
These are your 5 base pairs (Uracil replaces Thymine in RNA)
As DNA is being rebuilt by the various enzymes responsible for guiding each new base pair into proper position along the template strand during replication, the chemicals don't have a conscious ability to select from, for instance, adenine instead of guanine. Most of the time they'll get it correct, but lots of times they get it wrong. Other enzymes then have to come in and correct those errors by excising the incorrect base pairs and then rebuilding news ones. This process, as well, is imperfect. This is how mutations come about, primarily, and is pretty much the entire basis of why natural selection works. As these mutations are formed and then passed on to off spring, some increase survival and some decrease survival. Many lead to fetal inviability.
As long as purely chemicals are doing the job, without some sort of device that we've designed to examine and determine if the molecule being selected for is the exact proper one, there will probably always be some error built into the system.
That's why its good that the primary amino acids for life can also be formed from several different combinations of base pairings.
Glycine, for instance, can be formed by alleles made up of combinations of GGU, GGC, GGA, or GGG. Phnylalanine can be made up of UUU or UUC. Remember, again, that Uracil is replaced by Thymine in the DNA. So those would be GGT, GGC, GGA, GGG, and TTT, TTC respectively on the actual DNA strand.
The CRISPR-Cas9 system utilizes specially designed cocktails of enzymes to attempt to cut the DNA at a precise location, and insert new genes into that location, before "resealing" the DNA strand. These cocktails of enzymes are the same enzymes we see performing these operations in nature, mostly synthesized, and therefore suffer from the same imprecision.
For instance, when the Chinese researchers were modifying embryos using this technique, one of the biggest flaws was the failure rate, and the head scientist at the time was talking about how he would basically have to go back to the drawing board and redesign the cocktails of enzymes used to target the locations he wanted to insert these genes into, because the failure rate was so high.