The Free PACER Bill Will Save Money (Despite the CBO Score)
The Congressional Budget Office finally provided a "score" for the Free PACER bill. A CBO "score" is the way Congress figures out how much a bill costs and it's an important part of the process. The Free PACER bill can't advance without it.
It's great to finally have this score, and the bottom line is that the CBO estimates that the deficit will go up by only $77M over ten years. That's basically nothing, but the reality is even better.
The score vastly overestimates the cost of building and maintaining the system. A few numbers from the score are:
- $180,000,000 to make the new system
- $82,000,000 cloud software/hardware costs
We run CourtListener on the Amazon Cloud. CourtListener gets millions of hits and hosts millions of PACER documents.Based on our experience running CourtListener, we can say that the estimates from the CBO are an obscene amount of money and are wrong.
First, according to the CBO, these numbers pay for 12 software teams of 8-10 people each to build the new PACER system. That's over 100 developers! By contrast, we have four, and it's honestly a challenge to imagine how more than 100 people could develop a single website of this level of complexity. Talk to any VC. This is a vast sum of money. We can probably take 80% off this number.
Second, the score budgets 40 people to do maintenance. That's obscene and would be about $8M/year for three years for a total of $24M. Simple websites like PACER do not need 40 people to maintain them. Even ten is probably a lot. We can probably take 75% off this number.
Finally, the CBO says the new system will cost $82M in cloud costs. There's no way to put this mildly. That's wrong. That's just not how much things cost. We can probably take 90% off this number.
Taken together, these three errors make a huge difference. Let's do some math.
Original score says the bill will add $77M to the deficit.
Less 80% of $180M ($144M) to build the new system = $67M deficit reduction. We're in the green.
Less 75% of $24M ($18M) for maintenance = $85M deficit reduction.
Less 90% of $82M ($73.8M) for cloud costs = $158.8M deficit reduction.
That's pretty fantastic!
The CBO also notes that the Tax Court redid their system at $3M/year over three years, and says that PACER would be "significantly more expensive b/c of larger scale and complexity." First, scaling software is cheap. Second, the complexity isn't a 20× factor. This should have been a clue that their numbers were wrong.
Ultimately, if we take a step back, this is actually pretty intuitive. There's a horrible system that's 20 years old and costs a fortune to run. A new system will cost much less and reduce the federal deficit, even with usage fees removed. Congress should do this just to save money.
On top of this, even if the CBO score is right, Congress should pass the Open Courts Act because PACER is an anti-democratic, national security nightmare. Even if the original estimate is correct and everything above is wrong (it's not) — even if it's true that it'll cost $77M over ten years — that's an absolute bargain to fix the PACER system so it's secure and accessible.
(This blog post is a lightly edited version of a Twitter thread.)