NATIONAL VETERANS LEGAL SERVICES PROGRAM et al v. UNITED STATES OF AMERICA

Filing 52

MOTION for Summary Judgment as to Liability by ALLIANCE FOR JUSTICE, NATIONAL CONSUMER LAW CENTER, NATIONAL VETERANS LEGAL SERVICES PROGRAM (Attachments: #1 Declaration Declaration of Jonathan Taylor, #2 Exhibit Exhibit A, #3 Exhibit Exhibit B, #4 Exhibit Exhibit C, #5 Exhibit Exhibit D, #6 Exhibit Exhibit E, #7 Exhibit Exhibit F, #8 Exhibit Exhibit G, #9 Exhibit Exhibit H, #10 Exhibit Exhibit I, #11 Exhibit Exhibit J, #12 Exhibit Exhibit K, #13 Exhibit Exhibit L, #14 Exhibit Exhibit M, #15 Declaration Declaration of Thomas Lee and Michael Lissner, #16 Statement of Facts Plaintiffs' Statement of Undisputed Material Facts)(Gupta, Deepak)

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 1 of 18

IN THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
NATIONAL VETERANS LEGAL
SERVICES PROGRAM,
NATIONAL CONSUMER LAW
CENTER, and ALLIANCE FOR
JUSTICE, for themselves and all
others similarly situated,
Plaintiffs,

Case No. 16-745

v.
UNITED STATES OF AMERICA,
Defendant.
DECLARATION OF THOMAS LEE AND MICHAEL LISSNER
Thomas Lee and Michael Lissner hereby declare as follows:

Thomas Lee Background and Experience
1.
a

Thomas Lee is a software developer and technologist with

background

in federal government transparency issues. He

currently develops software for a large venture-backed software
company.

In

this

capacity

he

uses

cloud-based

storage

and

computation services on a daily basis and assists in cost estimation,
planning and optimization tasks concerning those services.
2.

Before taking on his current private-sector role in 2014,

Mr. Lee spent six years working at the Sunlight Foundation, serving
four of those years as the Director of Sunlight Labs, the Foundation’s
technical arm. The Sunlight Foundation is a research and advocacy
1

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 2 of 18

organization

focused

on

improving

government

transparency.

Sunlight Labs’ work focused on the modernization of government
information

technology

and

improving

the

distribution

of

government data. This work included technical project management,
budgeting, media appearances and testimony before Congress, among
other tasks.
3.

Prior to joining the Sunlight Foundation, Mr. Lee built

websites for large nonprofits, the U.S. Navy, and the offices of
individual members and committees within the U.S. Senate and
House of Representatives. Mr. Lee’s resume is attached to this
declaration.

Michael Lissner Background and Experience
4.

Michael Lissner is the executive director of Free Law

Project, a nonprofit organization established in 2013 to provide free,
public, and permanent access to primary legal materials on the
internet for educational, charitable, and scientific purposes to the
benefit of the general public and the public interest. In this capacity
he

provides

organizational

management,

publishes

advocacy

materials, responds to media inquiries, and writes software.
5.

Since 2009, Free Law Project has hosted RECAP, a free

service that makes PACER resources more widely available. After
installing a web browser extension, RECAP users automatically

2

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 3 of 18

contribute PACER documents they purchase to a central repository.
In return, when using PACER, RECAP users are notified if a
document exists in the RECAP central repository. When it does, they
may download it directly from the RECAP repository, avoiding the
need to pay PACER fees.
6.

In the course of maintaining and improving RECAP, Mr.

Lissner has become extensively familiar with PACER. During this
time RECAP’s archive of PACER documents has grown to more than
1.8 million dockets containing more than 40 million pages of PACER
documents.
7.

Mr. Lissner has conducted extensive research on the

operation and history of the PACER system. Among other topics, this
research has focused on the costs of PACER content and the history of
PACER fees. This research is available on the Free Law Project
website.1 Mr. Lissner’s resume is attached to this declaration.

Expert Assignment and Materials Reviewed
8.

We have been asked by the plaintiffs’ counsel in this case

to evaluate the reported fee revenue and costs of the PACER system
in light of our knowledge of existing information technology and
data-storage costs, our specific knowledge of the PACER system, and
our background in federal government information systems.

1

https://free.law/pacer-declaration/

3

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 4 of 18

9.

Specifically, the plaintiffs’ counsel have asked us to offer

an opinion on whether the Administrative Office of the U.S. Courts
(AO) is charging users more than the marginal cost of disseminating
records through the PACER system—in other words, to use the
language of the E-Government Act of 2002, the “expenses incurred in
providing” access to such records for which it is “necessary” to charge
a fee “for [the] services rendered.”
10.

In forming our opinion, we have reviewed the Plaintiffs’

Statement of Undisputed Material Facts and some of the materials
cited in that statement, including a spreadsheet provided to the
plaintiffs’ counsel in discovery (Taylor Decl., Ex. L) and the
Defendant’s Response to Plaintiffs’ First Set of Interrogatories (Taylor
Decl., Ex. M).
11.

We also rely upon our accumulated experience as

technologists and government transparency advocates.

Reasoning and Conclusions on Marginal Cost
12.

As we explain in detail below, it is overwhelmingly likely

that the PACER system, as operated by the Administrative Office of
the Courts (AO), collects fees far in excess of the costs associated with
providing the public access to the records it contains.
13.

The following calculations are intended to convey fair but

approximate estimates rather than precise costs.

4

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 5 of 18

14.

The marginal cost of providing access to an electronic

record consists of (a) the expenses associated with detecting and
responding to a request for the record; (b) the bandwidth fees
associated with the inbound and outbound transmissions of the
request and its response; and (c) the pro rata expense associated with
storing the records in a durable form between requests.
15.

As a point of comparison we use the published pricing of

Amazon Web Services (AWS). AWS leads the market for cloud
computing services2 and counts organizations including Netflix,
Adobe Systems, and NASA among its customers. Like most cloud
providers, AWS pricing accounts for complex considerations such as
equipment replacement, technical labor, and facilities costs. Although
the

division

is profitable, AWS prices are considered highly

competitive. AWS services are organized into regions, each of which
represents a set of data centers in close geographic and network
proximity to one another.
16.

For our evaluation, we first consider the cost of storage.

Researcher Matthew Komorowski3 and data storage firm BackBlaze4
have published storage cost time series that when combined cover the
period dating from the PACER system’s 1998 debut to the present.

2

https://www.srgresearch.com/articles/leading-cloud-providers-continue-run-awaymarket.
3
http://www.mkomo.com/cost-per-gigabyte
4
https://www.backblaze.com/blog/hard-drive-cost-per-gigabyte/

5

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 6 of 18

During this time their data shows the cost of a gigabyte of storage
falling from $65.37 to $0.028, a reduction of over 99.9%. During this
same time period PACER’s per-page fees increased 43%, from $0.07 to
$0.10.
17.

The effect of economies of scale makes it difficult to

assemble comparable time series for bandwidth and computing costs.
We are therefore unable to easily compare PACER fees’ growth rate to
the change in bandwidth and computing costs from 1998 to the
present.
18.

Fortunately, it is possible to compare recent PACER fee

revenue totals to reasonable contemporary costs for the technical
functionality necessary to perform PACER’s record retrieval function.
The AWS Simple Storage Service (S3) provides this necessary data
storage and retrieval functionality and publishes straightforward and
transparent pricing for it. S3 costs vary by region. Using the prices
published on August 27, 2017 for the “GovCloud” region, which is
designed for U.S. government users, we find storage prices of $0.039
per gigabyte5 per month for the first 50 terabytes, $0.037 per gigabyte
per month for the next 450 terabytes, and $0.0296 per gigabyte per
month for the next 500 terabytes. Retrieving an item from the

The quantity of data contained in a terabyte/gigabyte/megabyte/kilobyte varies
slightly according to which of two competing definitions is used. Our analysis
employs the definitions used by Amazon Web Services. c.f.
https://docs.aws.amazon.com/general/latest/gr/glos-chap.html
5

6

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 7 of 18

GovCloud region currently costs $0.004 per 10,000 requests, plus
data transmission at $0.01 per gigabyte.
19.

Determining how these prices might apply to PACER’s

needs requires knowledge of the PACER system’s size. We are not
aware of a current and authoritative source for this information.
Instead, we employ an estimate based on two sources from 2014: that
year’s Year-End Report on the Federal Judiciary,6 and an article
published in the International Journal for Court Administration.7 The
former states that PACER “currently contains, in aggregate, more than
one billion retrievable documents.” The latter states that the PACER
“databases contain over 47,000,000 cases and well over 600,000,000
legal documents; approximately 2,000,000 new cases and tens of
millions of new documents are entered each year.” Although the large
difference in document counts makes it unlikely that both of these
estimates are correct, they provide an order of magnitude with which
to work. For the sake of our estimate we double the larger of these
numbers and make the generous assumption that PACER now
contains two billion documents.
20.

Mr. Lissner’s custodianship of the RECAP archive allows

us to make estimates of the typical properties of PACER documents.

https://www.supremecourt.gov/publicinfo/year-end/2014year-endreport.pdf
Brinkema, J., & Greenwood, J.M. (2015). E-Filing Case Management Services in the
US Federal Courts: The Next Generation: A Case Study. International Journal for
Court Administration, 7(1). Vol. 7, No. 1, 2015.
6
7

7

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 8 of 18

21.

The

RECAP

Archive

contains

the

most-requested

documents from PACER, making them appropriate for our analysis.
22.

Mr. Lissner finds an average document size of 254

kilobytes and 9.1 pages, and therefore an average page size of 27.9
kilobytes. Assuming a PACER database size of two billion documents
and the prices recorded above, we calculate that annual storage costs
of the the PACER database on S3 would incur fees totaling
$226,041.60.
23.

This leaves the task of estimating the costs incurred by the

retrieval of documents. To do this we must estimate the total number
of requests served by PACER each year. The PACER fee revenue
reported for 2016 in the spreadsheet provided to the plaintiffs’
counsel in discovery is $146,421,679. The per-page PACER fee in 2016
was

$0.10.

Simple

arithmetic

suggests

that

approximately

1,464,216,790 pages were retrieved from PACER in 2016.
24.

This calculation does not reflect the 30 page/$3.00

per-document cap on fees built into PACER’s price structure; nor the
fact that some of the revenue comes from search results, which are
also sold by the page; nor any other undisclosed discounts.
25.

The RECAP dataset’s 9.1 page average document length

suggests that the fee cap might not represent a substantial discount to
users in practice.

8

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 9 of 18

27.

Out of an abundance of caution against underestimating

costs, we account for these inaccuracies by rounding the estimated
request count up to two billion for the following calculations.
28.

Using aforementioned S3 prices for retrieving an item

from storage, this volume of annual requests would incur $800 in
fees. An additional $558.24 in bandwidth costs would also be incurred.
This yields a total yearly estimate for storing and serving PACER’s
dataset using AWS S3’s GovCloud region of $227,399.84, or 0.16% of
PACER’s reported 2016 fee revenue.
29.

The tremendous disparity between what the judiciary

actually charges in PACER fees and what is reasonably necessary to
charge is illustrated by two alternative calculations. The first considers
what the per page fee could be if PACER was priced according to our
calculations. Including storage costs, we estimate that the per page
cost of retrieving a document from PACER could cost $0.0000006
(about one half of one ten-thousandth of a penny). The second
alternate calculation considers how many requests PACER could serve
if the fees it currently collects were used exclusively and entirely for
providing access to its records. Assuming no change in the size of the
dataset and using the storage costs calculated in association with that
size, $146,195,637.40 in fee revenue remains to cover document
requests and bandwidth. At the previously cited rates, this would

9

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 10 of 18

cover the costs associated with serving 215,271,893,258,900 requests,
or approximately 1,825 pages per day for every person in the United
States.

Reasoning and Conclusions on Reasonableness of Costs
30.

We offer the preceding analysis with three caveats. First,

at the time of PACER’s design and implementation, cloud computing
services were not widely available and the cost savings associated with
their scale could not be achieved. It is therefore reasonable to assume
that PACER’s costs could be artificially high due to the time in which
it

was

built,

although

effective

ongoing

maintenance

and

modernization should attenuate this effect. Second, although the
Administrative Office of the Courts could directly use the Amazon
Web Services we discuss, it would not be uncommon or unreasonable
to purchase those services through a reseller who increases their price
by some amount. Third, it is important to note that as outside analysts
with limited information, we cannot anticipate or account for all of
the costs that could conceivably be associated with access to PACER
records.
31.

But it is noteworthy that PACER fees increased during a

period of rapidly declining costs in the information technology sector.
Even after taking the preceding caveats into account, we are unable to
offer a reasonable explanation for how PACER’s marginal cost for

10

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 11 of 18

serving a record could be many orders of magnitude greater than the
contemporary cost of performing this function.
32.

It is overwhelmingly likely that the PACER system, as

administered by the AO, collects fees far in excess of the costs
associated with providing the public access to the records it contains.
33.

We declare under penalty of perjury that the foregoing is

true and correct.

Executed on August 28, 2017.

_____________________________
Thomas Lee

_____________________________
Michael Lissner

11

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 12 of 18

12

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 13 of 18

13

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 14 of 18

Thomas Lee 
understanding / making / explaining technology 
https://www.linkedin.com/in/tom-lee-a2112387/ 

EXPERIENCE 

Mapbox​ — ​Geocoding Lead 

50 Q St NE #2 
Washington, DC 20002 
[phone] 
[e-mail] 
https://github.com/sbma44 

SKILLS 

writing · team management · 

JUNE 2010 - PRESENT 

software development · data 

Guided Mapbox’s location search team through a period of fast growth 

analysis · speaking · system 

and into commercial success. Also performed a variety of legal, security 

administration · information 

and hardware tasks. 

security · embedded systems 

- Oversaw growth of geocoding business from 1% to 21% of revenue by 
line item, 39% to 71% by related-deal revenue. Shipped code, performed 
sales engineering, led hiring, participated in enterprise support, 
evaluated & managed compliance for licensed data. 

TECHNOLOGIES 
Expert 

- Managed federal government relations, including Congressional 

Javascript / Node.js · Python / 

lobbying & testimony, agency meetings & writing op-eds on behalf of 

Django / Flask · SQL / 

leadership. Liaised with relevant open data communities. 

PostgreSQL · bash / GNU · 

- Coordinated outside counsel during patent defense. 
- Designed and implemented royalty tracking pipeline and mobile SDK 
battery test methodology. Assisted in design of mobile telemetry 
security systems. Authored first version of security protocols for 
participation in infosec events with hostile networks. 

Docker · AWS / EC2 / ECS / 
CloudFormation / 
DynamoDB / ElastiCache / 
Kinesis / S3 · PHP / Drupal / 
Wordpress · AVR / Arduino · 
QGIS · GDAL · PostGIS · 
Mapbox 

Sunlight Foundation​ — ​CTO 
DECEMBER 2008 - JUNE 2010 

Productive 

Managed Sunlight Labs’ twenty-two person technology department 

Perl · Ruby · HTML5 · CSS 

during its prime years of influence and size. 
- Conceived, planned and executed mission-oriented technology 
projects. 

Tourist 
C · C++ · Swift/XCode · 
three.js 

- Represented Sunlight’s positions on various government transparency 
measures in Congressional testimony, speaking engagements, writing, 
and media appearances. 
- Expanded historically web dev-focused team to include political 
scientists, journalists, data analysts & mobile app developers. 
- Primary author of grants and reports for bulk of Sunlight funding. 
- Evaluated grant applications for potential funding. Managed 
relationships with peer organizations, funders and grantees. 

ORGANIZATIONS 

OpenAddresses · FLOC · 
HacDC · DCist 

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 15 of 18
EchoDitto ​— ​Sr. Software Architect 
DECEMBER 2005 - DECEMBER 2008 

Designed & implemented LAMP applications for campaigns and large 
nonprofits, primarily using the Drupal and WordPress frameworks. 
- Assisted in requirement-gathering, copy editing and writing, strategy 
brainstorming, customer interaction and visual design. 
- Developed variety of reporting mechanisms (SQL/Perl/Ruby). 
- Launched, maintained and generated bulk of content for 
developer-focused EchoDitto Labs site. 

Competitive Innovations ​— ​Software Developer 
August 2002 - DECEMBER 2005 

Created ASP.NET/Microsoft CMS-backed websites for committees and 
member offices in the U.S. House of Representatives; the U.S. Navy; 
George Washington University Law School; Miami Dade Community 
College; and the Corporate Executive Board. 
- Interviewed, evaluated, trained and participated in the management of 
junior technical staff. 
- Possessed security clearance as of December 2005. 

SELECTED CLIPS 

What Everyone Is Getting Wrong About Healthcare.gov 
Wonkblog, Washington Post 
http://www.washingtonpost.com/blogs/wonkblog/wp/2013/10/07/whateveryone-is-getting-wrong-about-healthcare-gov/ 

The Cost of Hashtag Revolution 
The American Prospect 
http://prospect.org/article/cost-hashtag-revolution 

The Deleted Tweets of Politicians Find a New Home 
Tell Me More (NPR) 
http://www.npr.org/2012/06/06/154432624/the-deleted-tweets-of-poli
ticians-find-a-new-home 

Enhancing Accountability and Increasing Financial Transparency 
U.S. Senate Committee on the Budget 
https://www.budget.senate.gov/hearings/enhancing-accountability-and
-increasing-financial-transparency 

EDUCATION 

University of Virginia ​— ​BA, Cognitive Science 
1998-2002 

Concentration in neuroscience, with work in the Levy Computational 
Neuroscience Lab. Computer Science minor. Echols Scholar. 

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 16 of 18

MICHAEL JAY LISSNER

[e-mail] • [phone] • 2121 Russell St., Suite B, Berkeley, CA 94705
EXPERIENCE

Executive Director and Lead Developer
Free Law Project

2013-Present
Emeryville, CA

Founded Free Law Project as a 501(c)(3) non-profit. My responsibilities as
founder/director include identifying and pursuing grants and contracts, handling the
marketing and accounting needs of the organization, and developing solutions for our
stakeholders.

Free Law Project has been awarded grants or contracts from Columbia University,
Georgia State University, University of Baltimore School of Law, and The John S.
and James L. Knight Foundation, and has partnered with Google, Inc. and the Center
for Internet and Technology Policy at Princeton University.
I am the lead developer for several of Free Law Project’s biggest initiatives, including:
• The first ever full-text search interface for documents from the PACER system,
containing nearly 20M records;
• The creation of the largest archive of American oral argument recordings,
consisting of nearly one million minutes of recordings;
• The development of a comprehensive database of American judges;
• The curation of 4M court opinions, which are available via a powerful search
interface, as bulk data, or via the first ever API for legal opinions;
• The creation of a web scraping infrastructure that has gathered more than 1M
documents from court websites.
This work has enabled a number of research papers, made legal research more
competitive, provided a useful resource to journalists, and helped innumerable people
to engage in the legal system.

New Product Designer/Developer
Recommind, Inc.

2012-2013
San Francisco, CA

• Worked with the new products team to design and develop new enterprise-class
products for AMLAW-50 law firms.
• Led design of new API-driven document sharing platform from initial concept
to final specification, seeking stakeholder approval from upper management,
sales, product management, and development teams. This process was guided
by the creation of paper prototypes and low fidelity wireframe diagrams,
culminating in high fidelity mock-ups and a written specification.

Solutions Developer
Recommind, Inc.

2010-2012

San Francisco, CA

• Designed and developed new features, products and processes for internal team
of technical consultants.
• Implemented distributed search systems for top international law firms.
• Collaborated with internal and external stakeholders to gather requirements and
scope work.
• Developed custom crawlers and search indexes for systems with millions of
records.

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 17 of 18

Technology Intern

Center for Democracy and Technology

Summer, 2009

San Francisco, CA

Wrote design specification and began implementation of location privacy
enhancements for the new Android operating system.

Systems Analyst and Community Researcher
Community Services Bureau

2005-2008
Contra Costa County

• Designed and implemented system for reporting educational outcomes and
program metrics to senior management.
• Researched and wrote federally-mandated annual assessment of community
needs.
• Worked with contractors to administer departmental databases and systems.
• Discovered and responsibly-disclosed security vulnerabilities in department
systems, protecting tens of thousands of child and parent records.
• Tracked and reported daily enrollment of more than 2,000 children.

EDUCATION

School of Information, UC Berkeley

2008-2010

• Masters in Information Management and Systems (MIMS), with a focus on
Internet Law and Policy and a certificate in Management of Technology from
Haas School of Business
• Theoretical coursework in information privacy, policy and economics,
intellectual property law, and technology strategy
• Technical coursework in security, networking, programming paradigms,
distributed computing, API design, and information architecture
• Taught Web Architecture summer seminar to class of twenty undergraduates
including fundamentals of networking, dynamic websites, and browsers

University of California, Berkeley Extension
• Unix/Linux fundamentals

2005-2008

• System administration programming, with focus on shell scripting and Python
• Advanced Java programming

Pitzer College, Claremont, California

2000-2004

• Bachelor of Arts in English and World Literature with a minor in Spanish
Language and Literature
• Coursework in economics, mathematics and C++ programming

PROJECTS &
RESEARCH

CourtListener.com
My capstone project at UC Berkeley and now a core initiative of Free Law Project,
CourtListener.com is an open-source legal research tool that provides daily awareness
and raw data to users via custom email alerts, Atom feeds, podcasts, a RESTful API,
and bulk data. CourtListener currently:
• Hosts the RECAP Archive, a collection of nearly 20M PACER documents;
• Has 4M Boolean-searchable opinions in its corpus;
• Has more nearly 700 days of oral argument audio;
• Has a comprehensive database of American judges;
• Receives thousands of API hits per day;
• Tracks every high court in the country, adding their opinions as they are
published.
https://www.courtlistener.com | https://github.com/freelawproject/courtlistener

Case 1:16-cv-00745-ESH Document 52-15 Filed 08/28/17 Page 18 of 18

Seal Rookery

The Free Law Project Seal Rookery is a small project to collect and distribute all
government seals in the United States. Currently, the project has more than 200
judicial seals.
https://github.com/freelawproject/seal-rookery

Selected Policy, Legal and Security Papers

• CourtListener.com: A platform for researching and staying abreast of the latest
in the law
• Wikipedia.org: Jacobsen v. Katzer, Zeran v. AOL
• The Layered FTC Approach to Online Behavioral Advertising
• Technology Revolution and the Fourth Amendment
• Transparent Panacea: Why Open Email is Fraught with Problems
• Proactive Methods for Secure Design
• Breaking reCAPTCHA
• Facebook’s Battle Sign: A Security Analysis
http://michaeljaylissner.com/projects-and-papers/

Additional Websites and Projects

michaeljaylissner.com | free.law | github.com/freelawproject

ADDITIONAL

Distance Travel
• Summer, 2013-2014: Completed south-bound thru-hike of Te Araroa Trail in
New Zealand (2,000 miles). The Te Araroa Trail is considered one of the mostchallenging long-distance trails in the world.
• Summer, 2010: Completed south-bound bike tour of California coast (1,000
miles).
• Summer, 2005: Completed north-bound thru-hike of Pacific Crest Trail from
Mexico to Canada via Sierra and Cascade mountains (2,500 miles).