@MartinEscardo @zyang and @jonmsterling said it: it's written for the reviewers. It's no longer the case that a specialist paper is guaranteed, nor even likely, to get 2 or more reviewers who have tried something like this before and will form an opinion about novelty from an expert position.

So when writing A is novel, and B is novel, and C is novel, the expert says 'yes, yes, yes', and the non expert says 'I'm not an expert, but A,B, and C seem novel' and feel comfortable accepting the submission.

@MartinEscardo @zyang @jonmsterling other mechanisms that degrade the paper, but ease refereeing:

1. Repeatedly referring back to the paper structure:

end of section A: ... we have shown X, and next we will show Y.
\section B having shown X, let's turn to Y.

(Rationale: referee is not really interested in the paper, so might lose the plot, or has a stack of papers to read between meetings, on the bus, etc and loses the bug structure.)

2. Not citing related work during the technical development. Instead, cite it in related work section.

Rationale: makes technical development seem more novel to nonexpert while an expert will likely not complain if the citation exists. Bulks up related work section and makes it seem in-depth because it is more technical.