Physics PhD here, I would extend your comment that 90% of CS papers are garbage to include engineering and physics as well. The vast majority of those are what people refer to as "status updates" where long term projects provide minor updates that don't really contribute anything to the general knowledge but exist solely to advertise their work and provide a bullet point in their quarterly report to the funding agency. This is also added to the fact that most PhD programs require a candidate to publish at least 3 articles in peer-reviewed journals, where some programs even discriminate based on the impact factor, further entrenching the paid-for journals as open access are not ranked as highly. I personally really like the IEEE journals as they are cheap ("free" with IEEE membership to a given society) and reasonable high quality (exceptions as always of course). Years ago I looked at trying to get my department a subscription to the major Elsevier journal in our industry and was quoted almost $6,000/yr, mostly for historic papers published back in the 50's, 60's and 70's where authors were long since dead and the utility mostly on filling in gaps in modern theory. While neither bandwidth nor website design infrastructure are free, high access costs are unnecessary and just equate to greed IMHO.