It is not per se bad to publish papers that build on one another. In contrast, a researcher who is an expert in a given topic should be encouraged to improve his methods.
The crucial point is if the contribution of every publication is high enough to justify a new paper. This is, of course, quite subjective, but the same rule ("the contribution has to be high enough") applies to every publication and every reviewer should be able to make an appropriate judgment. So what is so special about the salami slicing technique that it should be considered in particular?
In the case of salami slicing, the pitfall for a reviewer is that he is making the judgment based on the whole contribution by the author built up over several publications and not only on the new contribution. Far fetched example: Imagine someone has developed a vaccine for HIV and describes it in a first paper. Then in a second paper he describes the same method, but also describes how to add raspberry flavor. He might use the words "We developed the first reliable HIV vaccine and it tastes like raspberries". A reviewer reading this might get the impression that the contribution of that paper is the development of a raspberry flavored HIV serum (that would surely be worth for publication), while in fact, it is only about adding raspberry flavor to a given serum (not very impressive).
My rule of thumb to avoid this pitfall is to ask the following question. Imagine the second paper would have been written by another, previously completely unknown researcher and the actual author would review it, would he accept this paper?
If you expect the review sounds like "These researchers do nothing than adding raspberry flavor to the serum previously developed by us", then the second paper is a salami slice and should be rejected.