1. Science is conservative. The 'higher' you get, the more conservative you have to be.
2. Scientists in many fields do not read enough papers. They don't have the time. They grab abstracts and conclusions and read a section or two. They also might read a paper to contradict it destructively. Look at the evolution/ID debate if you don't believe me.
3. Papers are too long anyhow. Length is equated with depth, but it might more properly be equated with obfuscation.
4. As you said, counting citations is a joke.
Much more relevant might be a system where papers consisted of
1. Background was relegated to an appendix. Maths in another. Statistics to a third (if required). The briefest of introductions
2. Next an experimental discussion which set out only information necessary for the understanding the experiments, and their results.
3. Next a section on implications of experimental results.
4. All tripe about work done to be reserved for the lecturers who are marking student papers.
Each chapter would be a generous summary of the drivel currently making reading papers such a boring job.