UC OSPO Network

UC Open Source as Institutional Infrastructure

Tim Dennis · UCLA OSPO Lead, UC OSPO Network · IASSIST 2026 · Slides: tinyurl.com/iassist2026 · Handout: tinyurl.com/iassist2026-handout

From Dissertation to Global Infrastructure
Open Source Foundation
1977
BSD Unix UC Berkeley
Permissive licensing that made open source possible. MIT, Apache, and BSD licenses now govern billions of lines of production software. history →
1973
INGRES / PostgreSQL UC Berkeley
Stonebraker and Wong repurposed a geography grant to build INGRES after reading Codd's relational model papers, releasing it freely. The open code seeded Sybase and, indirectly, Microsoft SQL Server. Stonebraker's follow-on POSTGRES (1986) pioneered object-relational databases; open-source volunteers forked it in 1995 to create PostgreSQL, the world's most popular independent open-source database. postgresql.org →
Universal Infrastructure
1986
tz database UCLA (Paul Eggert, since 2005)
Every OS clock on the planet depends on this. Eggert used 19th-century almanacs to reconstruct pre-1883 timezone history. A 2011 IP lawsuit nearly took it offline; IANA now has stewardship. iana.org/time-zones →
Open Science
2000
UCSC Genome Browser UC Santa Cruz
Jim Kent assembled the human genome on 50 borrowed PCs, finishing three days before Celera Genomics and its supercomputer. The genome stayed in the public domain. genome.ucsc.edu →
Scientific Computing & Domain Science
2001
Jupyter UC Berkeley
Fernando Pérez built the first IPython in an afternoon procrastinating on his physics dissertation. Carol Willing, a core Steering Council member and Python Core Developer, shared the 2017 ACM Software System Award for Jupyter's global influence. Pérez is now faculty director of BIDS and co-PI on the UC OSPO Network. jupyter.org →
2001
EEGLAB UC San Diego
The world's premier open-source environment for electrophysiological signal processing. Over 350,000 downloads; the 2004 foundational paper has 14,400+ citations, among the highest academic footprints of any scientific software. eeglab →
2003
Cytoscape UC San Diego
The global standard for network biology: visualizing disease pathways, protein interactions, and molecular networks. Over 300,000 downloads per year; foundational to cancer and therapeutic research worldwide. cytoscape.org →
Infrastructure at Scale
2004
Ceph UC Santa Cruz
Sage Weil’s PhD dissertation on exabyte-scale distributed storage. Merged into the Linux kernel (2010); commercialization proceeds funded CROSS at UCSC, the first UC OSPO and seed of this network. ceph.io →
2009
Apache Spark UC Berkeley AMPLab
Matei Zaharia built it so a classmate could beat the Netflix Prize with faster distributed ML. Holden Karau (PMC member, core committer) scaled its Python and core engines. Donated to Apache before commercializing; Databricks now valued at $62B+. spark.apache.org →
2010
RISC-V UC Berkeley
A three-month summer project to avoid paying ARM royalties. Now in 10 billion devices, 350+ member organizations in 70 countries. Foundation relocated to Switzerland in 2020 to stay neutral in US-China trade tensions. riscv.org →
2010
Named Data Networking UCLA · Lixia Zhang, Lead PI
Zhang drove tractors in rural China during the Cultural Revolution, earned her PhD at MIT, and was the only woman at the first IETF meeting (1986). She coined "middlebox," designed RSVP, and is in the Internet Hall of Fame. NDN rethinks internet architecture at the protocol level. named-data.net →
AI Infrastructure
2016
Ray UC Berkeley RISELab
Two PhD students tired of rewriting Python to scale. Now core OpenAI compute infrastructure. Donated to the Linux Foundation in 2025. ray.io →
Also from UC campuses
Apache HTTP Server (Brian Behlendorf, UCB) Processing (UCLA) UCSD Pascal ChimeraX (UCSF) OpenMM (UCSF) sourmash (UC Davis) ERPLAB (UC Davis)
The UC OSPO Network
6
UC campuses · launched April 2024
Alfred P. Sloan Foundation
$1.85M
in external funding to date
280K+
students · 25,000+ faculty served
Santa Cruz Berkeley Davis Los Angeles Santa Barbara San Diego

Lead: UCSC, first OSPO in a large state university system. Part of CURIOSS (34 academic OSPOs worldwide).

What We Found (2025 survey, n=294)
52K
institutionally affiliated repos across 10 UC campuses
58%
of UC OSS contributors are also maintainers, not just users but stewards

#1 challenge: time to write documentation

Nationally (Tidelift 2024): 60% unpaid · 43% burnt out

Why It Matters: The Economics
$8.8T
demand-side replacement value of open source software to global enterprises

Supply-side cost to recreate the top 5,000 packages: $4.15B. If OSS disappeared and every organization built its own: $8.8 trillion. Ratio: ~2,000x. Hoffmann, Nagle & Zhou (2024) HBS WP 24-038.

Three Focus Areas
🔭
Discovery: UC Open Repository Browser (UC ORB); GitHub pipeline + system-wide survey
🌿
Sustainability: Cross-campus licensing WG; OSSPREY project health assessment; community management
📖
Education: Curriculum inventory at ucospo.net/education; Carpentries + CodeRefinery + UC content
Resources for Your Researchers
README.md template + guide
LICENSE guide with UC-approved options
CONTRIBUTING.md template + guide
CODE_OF_CONDUCT.md Contributor Covenant
CITATION.cff + Zenodo → citable DOI

All at ucospo.net/oss-resources · Sept 2025

Learn More
ucospo.net ucospo.net/education curioss.org, 34 academic OSPOs worldwide

Tim Dennis
Data Science Center, UCLA Library
tdennis@library.ucla.edu

Gomez et al. (2025) arxiv:2506.18359 · Scarlett et al. (2025) doi:10.31235/osf.io/p8bx6_v1 · Tidelift (2024) State of the Open Source Maintainer Report · Ruff (2026) youtu.be/eBriL3CDNeo IASSIST 2026