I'm in mathematics, and this is very much true in my field. Papers can be, and nearly always are, widely distributed on the Internet (arXiv) before formal publishing. They are widely read, cited, and built upon.

The publishing industry survives because researchers need to put "Published in Journal X" on their CVs. The peer review process also can lead to incremental improvements, and will occasionally catch major errors, but at this point everything other than the stamp of approval is secondary.