CourtListener expenses are less than $100k/year for providing storage and indexing of the PACER documents they have [1] (the Internet Archive also keeps a copy as mentioned in my above comment). There is room for the court's costs to decline substantially, and those cost control pressures on the judiciary's technology budget are what I'm interested in seeing. They are asking for far more than what they need from their user fees. I am not against reasonable user fees (I said "freely" above, and that's my faux pax), but very much against gouging because of politicians and bureaucrats.

For funsies, let's assume the entire PACER corpus is 10TB (edit: appears to be closer to 250-300 TB). Let's also assume that you want all of it OCRd, indexed, and available over a web interface and an API. Let's also assume a staff of ~5 technologists to oversee such a system. What do you think a reasonable annual budget would be for such a system? I would argue no more than $2-3 million. Let's Encrypt (non-profit for public certs) has a budget of $3.6 million and employees 13 people (as of 2019).

Perhaps the judiciary should just pay Free Law Project to host PACER, they’ve already proven they can provide a superior product cost effectively.

[1] https://projects.propublica.org/nonprofits/display_990/46334... (Free Law Project, 2018 990-EZ, Line 17, total expenses)