Why Baseline JPEG Avoided Submarine Patents

When the Joint Photographic Experts Group formulated the JPEG standard in the late 1980s and early 1990s, they deliberately engineered the "baseline" format to evade submarine patents—patents intentionally kept hidden or vaguely drafted until an industry standard becomes ubiquitous, at which point the holder demands exorbitant licensing fees. By prioritizing public-domain technologies like basic Discrete Cosine Transform (DCT) and standard Huffman coding over newer, technically superior methods encumbered by proprietary claims, the working committee successfully insulated the format from legal ambushes. This deliberate architectural choice ensured that baseline JPEG remained royalty-free, paving the way for its universal adoption across the global web and consumer electronics.

During the late 20th century, standard-setting organizations faced immense legal risks. A standard could be compromised if a participant or outside firm secretly guided technical specifications toward proprietary patents that were still pending at patent offices. Once the standard was locked into hardware and software, the patent holder would "surface" the submarine patent to extract royalties from an entire industry with no easy alternative.

The most critical battle over intellectual property in JPEG centered on entropy coding. The JPEG committee evaluated two primary compression techniques: arithmetic coding and Huffman coding. From a purely technical perspective, arithmetic coding was superior; it produced file sizes roughly 5% to 10% smaller than Huffman coding at identical quality levels. However, arithmetic coding was heavily protected by an overlapping thicket of patents held by major telecommunications and computing firms, including IBM, AT&T, and Mitsubishi.

Recognizing that patent disputes could destroy the format's potential for widespread adoption, the committee split the specification into different operational modes. They created advanced profiles that allowed arithmetic coding, but explicitly restricted the mandatory "Baseline JPEG" profile to traditional, table-based Huffman coding.

Huffman coding, originally developed by David Huffman in the early 1950s, was thoroughly entrenched in the public domain. Its foundational mathematical concepts were published decades prior, establishing a dense body of prior art that made it virtually immune to new patent grants. By mandating Huffman coding for baseline compliance, the committee guaranteed that any developer, operating system, or camera manufacturer could write a compliant JPEG decoder and encoder without licensing proprietary entropy-reduction algorithms.

Similarly, the core mathematical transform of JPEG—the Discrete Cosine Transform (DCT)—was selected partly because it was published openly in 1974 by Nasir Ahmed, T. Natarajan, and K. R. Rao. Because the underlying transform mathematics were already part of academic literature, private entities could not easily claim monopolistic rights over the core compression pipeline.

The foresight of this patent-avoidance strategy proved vital in the decades that followed. When companies like Forgent Networks attempted to monetize broad digital image patents against the tech industry in the early 2000s, the baseline format’s reliance on well-documented prior art gave developers, standards bodies, and courts the technical evidence required to challenge and invalidate the claims. Without the deliberate legal insulation embedded into baseline JPEG, the open, image-rich internet would have faced severe fragmentation and stifling licensing barriers during its formative years.