This is not true for all fields. Computer science (specifically machine learning) is possibly the most lucrative area to work in right now and comparatively few people publish in journals. Computer vision has been this way for a long time even with journals like IEEE PAMI.
Our equivalent is highly competitive conferences that have sloppy/inconsistent peer review and (as a result of the format) limited submissions. You have to submit a full paper and acceptance rates are low. Researchers plan and structure entire projects around deadlines for big venues. Inconsistent peer review is somewhat made up for by community validation (most of the time, since publishing code is strongly correlated with citations). Really big papers often turn up as preprints first because speed is important (if you wait 6 months to go through review, someone will beat you).
Or you have fields like astrophysics where everyone cross-publishes in Arxiv when they submit to a journal.
And elsevier has massive profit margins. It earns way more then needed to pay emplyees.
The reason HN dislikes then is that HN has many members who work or worked in academia.