Saturday, May 29, 2010

What Does It Cost to Provide Electronic Public Access to Court Records?

US Courts have long faced a dilemma. Public access to proceedings is essential to a well-functioning democracy. On the other hand, providing public access requires expenditure of funds. Charging for access works against public access. Traditionally, these costs have been considered to be part of the general operating cost of courts, and there have been no additional fees for public access. The cost of the courthouse, the public gallery, and the bailiff are included. The administrative cost that the clerks incur in providing free public inspection of records is also covered, although the clerk may collect fees for filing actions or making physical copies.

I have been trying to understand how these practices have been translated into the networked digital era by exploring PACER, the US Courts' system for "Public Access to Court Electronic Records." Digital technologies have a way of pushing the cost of information dissemination toward zero, but as I observed in a recent working paper, this does not appear to be the trajectory of public access fees. Congress has provided a statutory limitation that states that the "Judicial Conference may, only to the extent necessary, prescribe reasonable fees... to reimburse expenses incurred in providing these services." In short, you can only charge for public access services if those fees are used to, at most, cover the operating expenses for those same services. What's more, in an accompanying conference report, Congress noted that it "...intends to encourage the Judicial Conference to move... to a fee structure in which this information is freely available to the greatest extent possible."

As described below, the Judiciary's financial reports appear to tell a different story: In the past several years, the Judicial Conference has consistently expanded the scope of its expenditures of public access fees such that the vast majority is now spent on other services.

The Judiciary Financial Plans

The first source for my analysis is the Judiciary's annual set of Financial Plans, submitted to Congress after their funds for that year have been appropriated. These are not made publicly available, but I have obtained the relevant excerpts from 2007, 2009, (appended to my working paper) and 2010 (here). I haven't yet obtained the 2008 Plans, so for two data points from that year I have to estimate based on averages for the prior and following years. You can download my Excel spreadsheet that combines the top-level data and drives the chart below (note the comments in the spreadsheet for details on how the numbers were derived).

EPA (Electronic Public Access) funds are collected solely via PACER fees, and are expended on a variety of programs. One of these expenditures is the PACER program itself, but many other expenditures are not. This includes things like "courtroom technology", "telecommunications", and "CM/ECF" (the electronic filing system). I described some of these in my working paper, but after I published that I had the opportunity to ask a panel made up of staff members from the Administrative Office of the US Courts and federal judges how these fees were used. At the 7th Conference on Privacy and Public Access to Court Records, the Hon. William E. Smith from the United States District Court for the District of Rhode Island explained that PACER fees:

"...also go to funding courtroom technology improvements, and I think the amount of investment in courtroom technology in '09 was around 25 million dollars. [...] Every juror has their own flatscreen monitors. We just went through a big upgrade in my courthouse, my courtroom, and one of the things we've done is large flatscreen monitors which will now -- and this is a very historic courtroom so it has to be done in accommodating the historic nature of the courthouse and the courtroom -- we have flatscreen monitors now which will enable the people sitting in the gallery to see these animations that are displayed so they're not leaning over trying to watch it on the counsel table monitor. As well as audio enhancements. In these big courtrooms with 30, 40 foot ceilings where audio gets lost we spent a lot of money on audio so the people could hear what's going on. We just put in new audio so that people -- I'd never heard of this before -- but it actually embeds the speakers inside of the benches in the back of the courtroom and inside counsel tables so that the wood benches actually perform as amplifiers. So now the back of the courtroom can really hear what's going on. This all ties together and it's funded through these fees."

Clearly, the costs of expensive multimedia systems for courtrooms are not part of the expenses incurred in providing PACER. The 2007 Judiciary Financial Plans delineate between EPA (PACER) and non-EPA programs, illustrating the substantial discrepancy in funds generated by the PACER program and the funds spent on PACER. As described in my working paper, the Courts can point to no statutory justification for spending PACER fees on these non-EPA programs. As of 2009, the Financial Plans no longer separate EPA and non-EPA expenses, but it is easy to reconstruct these totals based on the individual breakouts included in the plans. By doing this, I generated the following graph:

Income is in green, which consists of either direct collections or carryover from the previous year. Expenditures are in red. As you can see, according to the courts, the cost of running PACER has grown only slowly over time, whereas other services have grown dramatically. The carryforward peaked in 2008 at $44.5m, around the time that the courts decided to start spending more aggressively on non-PACER programs. Specifically, in March 2007, the Information Technology Committee of the Judicial Conference observed that, "In recent years, significant unobligated balances have accumulated," and proposed to, "expand use of Electronic Public Access funds for IT efforts, such as applicable network, courtroom technology and jury management requirements. The IT Committee did not support any reduction to the fee at this time." In 2010, expenditures on non-EPA services will actually exceed EPA revenues. As of 2011, the courts plan to have spent out most of the carryforward they had accumulated.

In their defense, the courts argue that all of the programs on which they spent PACER funds are somehow generally related to electronic public access. The current PACER site notes that PACER fees are "used to finance other expenses related to electronic public access to the courts in areas such as courtroom technology and the Bankruptcy Noticing Center." Nevertheless, the fact remains that many of those do not represent "expenses incurred in providing [the charged for] services." Programs like CM/ECF or Telecommunications represent, at best, ancillary programs. However, most if not all of their expenses would exist regardless of the PACER program. What's more, parties have always had to pay filing fees for certain actions, and although CM/ECF has saved them time and money compared to the days of couriers, public access fees are instead paying for the entirety of the system's development. Likewise, the Telecommunications program extends far beyond anything required to support PACER, and would be necessary regardless of any EPA-related use. Bankruptcy Noticing ($9.7m planned for 2010) is a free service that creditors use to monitor incoming bankruptcy claims.

Long Range IT Plan for the Judiciary

My second source for a big-picture perspective on IT spending by the courts is the annual "Long Range Plan for Information Technology in the Federal Judiciary." The 2010 version is available from the US Courts website, but the link to the 2009 version was broken in the recent upgrade of the site (which was, ironically, intended to make information more easily accessible). Fortunately, I have it. [Update: the 2010 link is now also broken, but I posted a copy here.]

The Long Range Plan covers IT financing of the entire Judiciary, and as such it describes far more than just EPA (PACER) fees. That being said, there is a fascinating shift from the 2009 Plan to the 2010 Plan. Each year, the Judiciary forecasts costs for many different IT-related program areas. We can therefore compare the projections for FY 2010 that are found in the 2009 Plan with the FY 2010 projections found in the 2010 Plan. Four of these program areas immediately pop out in such a comparison: Electronic Public Access Program, Court Allotments, Court Administration and Case Management, and Telecommunications. You can see the changes from one year to the next reflected in the chart below:
Program Costs
FY 2010
in 2009 LRP
FY 2010
in 2010 LRP
% ChangeChange
Electronic Public Access Program (PACER)
Court Allotments$143.9m$102.7m-28.63%-$41.2m
Court Administration and Case Management$22.1m$2.6m-88.24%-$19.5m

Somehow, the projected costs of the Electronic Public Access program in 2010 grew by about 300% between 2009 and 2010. The cost of Court Allotments, Court Administration and Case Management, and Telecommunications shrank by an equivalent amount. It is hard to imagine that the actual plans of the Judiciary changed so dramatically from one year to the next. Rather, it seems far more likely that they simply decided to change their accounting practices to portray a cost for the EPA system commensurate with the amount they are collecting.

What Should PACER Cost To Run?

The FY2010 Financial Plan represents the lowest estimate from the Judiciary that I can find for current PACER costs, listing "Public Access Services and Applications" at $21.9m. But is that a reasonable number for what PACER should cost to run? Even if the Judicial Conference believes so, there are several reasons why it could be run far more efficiently:

PACER is run on a highly inefficient decentralized infrastructure
Every court runs its own instance of PACER software, requiring its own hardware, network connection, and support staff. This means that, between district, bankruptcy, and circuit courts, these resources are duplicated approximately 200 times. I have heard various theories for why this is the case, including the notion that control of records has been traditionally delegated to local jurisdictions. It may also be true that at the time PACER was first deployed this was the only technical and operational way to implement it. However, a modern system administrator would never choose to implement a system that exhibited these inefficiencies. Fortunately, the Administrative Office of the Courts already controls the whole network and a first step of physical (if not logical) centralization should be fairly straightforward.

PACER costs include maintaining a staff in San Antonio, TX to answer phones
Although the average PACER user may not be aware of it, there is a full-time staff at the PACER Service Center just waiting to answer their various PACER-related questions (In 2009 this included 135,000 help desk calls, and almost 30,000 support emails). This service helps to overcome some of the more confusing usability barriers of the current PACER system, because these people will walk users through the process. This service is funded out of basic PACER access fees, which are based on per-page access rather than phone calls to the support staff.

PACER costs ironically include overhead from fee collection itself
Every quarter, PACER staff must prepare and physically mail bills to all PACER users that have incurred a billable level of fees. They must deal with all of the administrative overhead of managing these collections, including chasing down delinquent debtors and prosecuting them, if necessary. This portion of costs is a self-fulfilling prophecy.

PACER costs include expenses from upgrading the user interface, when third-parties could do a better job for free
The courts could publish all PACER data in bulk-downloadable format with relative ease and at a low cost. In this scenario, it is very likely that third parties would make the data more easily accessible in a variety of formats, at no cost to the courts. This general principle is laid out by my colleagues in a paper entitled "Government Data and the Invisible Hand."

If providing electronic public access can be grounded in free bulk access, the costs might well be manageable even within a no-fee system. The courts might also find it easier to avoid straying from their statutorily constrained requirement to, "only to the extent necessary, prescribe reasonable fees... to reimburse expenses incurred in providing these services."

[An advance copy of this post was sent to the Administrative Office of the Courts, which declined to provide comments, corrections, or additional documentation.]

Thursday, May 27, 2010

Summary of "The FCC’s Authority Over Broadband Access"

[Update: the video and transcripts for Panel 1 and Panel 2 are now posted]

Today I attended the "The FCC’s Authority Over Broadband Access" event in DC. These DC policy events tend to have more talking points than I'm willing to tolerate, but today's event was both balanced and substantial.

[A disclaimer, this post is fairly hastily written and assumes some deeper background knowledge of some of the terms. I recommend Harold Feld's recent blog post if you find unfamiliar language or concepts in here. Susan Crawford also has good stream-of-discussion notes of the first panel.]

Session 1: The History and Context of the Debate

The first panel was mostly a history of how we got here, hitting many of the points I explored in my post on Freedom to Tinker yesterday, "Regulating and Not Regulating the Internet". The panelists were certainly well qualified. John Nakahata is a former FCC Chief of Staff who lived through many of the relevant policy decisions in the late 1990s. Jessica Rosenworcel is the current Senior Counsel for the Senate Commerce Committee (which has authority over the FCC). John Windhausen is one of her predecessors, former Senior Counsel for the committee, who lived through the 1996 Telecommunications Act.

Nakahata reminded the audience that in the mid-1990s, the administration had proposed a new Title VII to the Communications Act that would have explicitly set policy for broadband. However, this approach never progressed and instead the 1996 Act simply codified a version of the Computer Inquiries "basic"/"enhanced" services distinction in the form of "information"/"telecommunications" services... without explicit reference to the internet. This is a layered model, in opposition to the silo-like approach of the Act's overall structure. He also observed that the 1998 "Stevens Report" (to which Brand X refers extensively) noted that if placing broadband under Title II was too onerous, the FCC could forbear from much of it (essentially what Chairman Genachowski is proposing today -- Nakahata seemed to think this was ironic, but it seems more like evidence of good research and historical consistency on the part of the Charman's office). He also noted that since that time there have been changes in the market -- including the elimination of mandatory unbundling or line-sharing -- that alter the broader policy calculus (presumably toward greater regulatory intervention). Finally, he observed that although there may be limited precedent for defining a separate telecommunications portion of an integrated service, pursing this approach on a large scale would be a revolutionary rather than evolutionary development.

John Windhausen confirmed Nakahta's telling of the 1996 Act's "information"/"telecommunication" service legacy in the Computer Inquiries. He laid out two basic principles which he thought had guided regulation of communications going all the way back to common law: 1) Common Carriage and a non-disciminatory duty to serve of public transport providers and 2) The principle of not regulating the communications that are transported. We are having today a variation of the age-old discussion of where to draw the line between the two. He also noted that some of the ambiguity of the 1996 Act was intentional. Congress sought to defer to an expert agency on the details. He emphasized that his reading of Brand X was that it quite clearly concluded that the categorization of broadband services was up to the determination of the FCC under the Chevron doctrine. He also indicated that the Commission had already demonstrated the possibility of classifying a portion of internet service as a telecommunications service when it issued the wireline broadband order and noted that ISPs could still voluntarily offer the service as a common carrier.

Jessica Rosenworcel didn't speak to the issues in as much depth as her fellow panelists, but such reticence to make public pronouncements is to be expected from an actively employed congressional counsel. That being said, she identified some high-level themes. First, technology changes quickly, making it difficult for regulators let alone legislators to keep up. Second, she observed that in the past ten years there appears to have been an effort to (understandably and logically) treat like services alike, regardless of the different technologies used to provide those services. However, this approach is perennially made difficult because of the "siloed" structure of the Act.

Session 2: The Third Way - What Happens Next?

The second panel was a bit more rough-and-tumble as it addressed the current debate over broadband reclassification. First up was Jim Speta, a Northwestern law professor who has long argued for a more antitrust-like approach to communications law. Second was Susan Crawford, a Cardozo law professor who recently did a stint advising the Obama Administration on technology issues. Third was Yochai Benkler, the primary investigator on the Berkman Center's Next Generation Connectivity Report for the FCC (which I contributed to). Finally, there was Jon Nuechterlein, an attorney for the broadband companies (representing only his personal views), who observed that he was the only practicing lawyer on the panel. Nuechterlein is one of the more fun people to critique, because he is so very smart and often so very wrong. You can see my stream of consciousness thoughts about Jon's position in my tweets.

As the panel began, I predicted: Speta: "antitrust!", Crawford: "infrastructure!", Benkler: "innovation!", Nuechterlein: "determinacy!". I meant that Jim would continue to push his antitrust-oriented view of Communications Act reform, Susan would remind us that broadband is general-purpose infrastructure and not just another market, that Yochai would argue that oversight of the broadband market is essential to all sorts of innovation, and that Jon would emphasize that any attempts to reclassify broadband would result in drawn-out court battles that would cast a pall of indeterminacy over the market, chill investment, and slow growth. I was not disappointed.

However, the panel focused far more on the intricacies of whether reclassification would hold up in court, what arguments would hold the day, and whether there were respectable policy justifications for these arguments. I love that stuff, so I can't complain. Speta started by observing that he thought it likely that reclassification would survive, given the Chevron-based deference articulated in Brand X. However, he was not sure that this was a good policy outcome. In particular, he said that he does not agree with the portion of the Schlick memo that claims that forbearance is difficult to reverse. Thus, he's worried about subsequent regulatory overreach. His prescription is a new governance structure based entirely on analysis of whether one firm has unreasonably foreclosed business of another. Susan agreed that reclassification should prevail, and made her typically well-articulated case for precedentially and empirically grounded government oversight of public communications carriers. She also observed that the reclassification position being advanced is that only the access portion of internet services be classified as a Title II service. Yochai began by observing the across-the-board agreement on the legal viability of reclassification, and provided some comparison points from around the world where defer-to-the-market approaches failed (such as New Zealand).

Then came Nuecterlein. His comments might serve as a clue on what might be included in a petition for reconsideration on the FCC's ultimate reclassification order. I was expecting something new and challenging, but frankly I was disappointed. He of course didn't think that reclassification would survive legal challenge (given that he'd be on the side arguing against it). His argument followed the predictable pattern: There is no "telecommunications" component of internet access, and in any case the service offered to end users is integrated with an information service component which is mutually exclusive with any telecommunications component (thus transforming the service into merely an information service). His policy argument was that if reclassification succeeded, many higher-layer services (such as web applications) would be poisoned with overbearing regulation because they inevitably would be classed as telecommunications services as well (I critiqued his first stab at this argument over here). What's more, according to Jon, other portions of the internet might not be classified as telecommunications services and thus bad things could be done there with impunity. I suppose this final point is somewhat new, but the others are rather standard fare and don't really account for the counterpoints already in circulation. The success of his legal arguments will likely hinge on a series of finely focused distinctions in the web of service definitions found in the 1996 Act. He referred briefly to Harold Feld's recent fantasy FCC predictions, but thus far I am far more persuaded by Harold's position than Jon's (especially in the deference-rich environment of Chevron). You can get a sense for my opinion on a few of his specific points in my tweets.

In any case, I'll do a follow-up post on Freedom to Tinker outlining my fantasy FCC reclassification predictions. Sorry about the inevitable typos and harried prose above, but I've got some real baseball to attend to -- I'm about to get off the train to Citi Field to watch the Phillies beat the Mets.