The Precedent:  Federal Circuit Corrects PTAB's Grammar-Based Claim Construction in Netflix, Inc. v. DivX, LLC

Authored by Marcel Duhamel

In this edition of The Precedent, we outline the decision in Netflix, Inc. v. DivX, LLC.

Overview 

In Netflix, Inc. v. DivX, LLC, the Federal Circuit reversed the Patent Trial and Appeal Board’s claim construction in an inter partes review and vacated its non‑obviousness decision.  The Federal Circuit held that a key phrase in a video streaming patent modifies the “encrypted portions of frames of video,” not the “encryption information” itself.  Under that construction, the prior art taught the disputed limitation so the case was remanded for further obviousness analysis.

Issues

1. Which element of the claim limitation should the phrase “within the requested portions of the selected stream of protected video” be read to modify?
2. Under the proper construction, does the asserted prior art teach that limitation?

Holdings

1. The phrase “within the requested portions of the selected stream of protected video” modifies “encrypted portions of frames of video,” not “encryption information.”
2. Under that construction, the limitation is met by the prior art relied on in Netflix’s IPR petition.

Background and Reasoning

DivX’s patent describes adaptive bitrate streaming of “protected” video (that is, encrypted video).  A server stores alternative streams of the same content at different qualities.  A client device requests specific byte ranges of a selected stream and uses “encryption information” and a set of common keys to decrypt partially encrypted video frames.

Netflix challenged all claims in an inter partes review.  The dispute on appeal centered on claim 1, and particularly limitation [l], which recites:

“locating encryption information that identifies encrypted portions of frames of video within the requested portions of the selected stream of protected video.”

The Board majority read this as requiring that the encryption information itself be located “within the requested portions” of the stream.  On that reading, encryption information stored elsewhere, such as in a top‑level index file, would not satisfy the limitation.  Applying this construction, the Board concluded Netflix had not shown the claims unpatentable.  One judge dissented.

Netflix appealed.  The Federal Circuit reviewed claim construction de novo because the analysis depended only on the claim language and intrinsic record.

The Federal Circuit began by acknowledging that limitation [l] is grammatically ambiguous in isolation.  The phrase “within the requested portions of the selected stream of protected video” could refer either to the location of the “encrypted portions of frames of video” or to the location of the “encryption information.”

It then applied a rule of grammar: the “nearest sensible referent” principle, sometimes also described as the “last antecedent rule.”  When a post‑positive modifier can sensibly apply to more than one earlier phrase, normal English usage ties the modifier to the nearest phrase, absent a good reason to do otherwise.  Both “encryption information” and “encrypted portions of frames of video” are syntactically possible referents modified by the phrase “within the requested portions of the selected stream of protected video” but “encrypted portions of frames of video” is nearer to the modifier.  That favored Netflix’s reading.

The Federal Circuit next looked at the broader context of claim 1.  Limitations [m] and [n] require decrypting the encrypted portions “identified within the located encryption information” and then playing back the decrypted frames obtained from “the requested portions of the selected stream.”  Under Netflix’s construction, the sequence is coherent.  The device locates encryption information (wherever stored), uses it to identify which portions of frames within the requested byte ranges are encrypted, decrypts those encrypted portions and plays back the frames.  Nothing in that sequence requires that the encryption information itself reside in the requested portions of the stream.

Importantly, Netflix’s construction did not render any words superfluous.  Limitation [l] still requires that the encrypted portions being identified are in the requested portions, not just anywhere in the selected stream.  That is a meaningful constraint.

The Federal Circuit then turned to the specification.  Many embodiments describe “DRM information” stored inside Matroska container files along with the video.  However, the patent also discusses embodiments that use “conventional Matroska” containers.  DivX’s own expert conceded that such containers do not support embedded encryption information as required by the claims.  In those embodiments, the encryption information must come from elsewhere, such as a top‑level index file.

Claim 1 itself recites obtaining a “top level index file” that identifies alternative streams.  The shared specification, and a related patent in the family that explicitly claims DRM information in the index, confirm that the invention contemplates encryption information stored outside the requested portions of the stream.  That intrinsic evidence undercut DivX’s narrower reading, which would confine the encryption information to the stream portions themselves.

The Federal Circuit therefore held that the Board erred in reading the location phrase as modifying “encryption information.”  It instead modifies “encrypted portions of frames of video.”  DivX did not dispute that, under Netflix’s construction, the prior art combination Netflix relied on teaches this limitation.  The Federal Circuit, therefore, reversed the Board’s construction, held that limitation [l] is taught by the prior art, vacated the non‑obviousness decision and remanded for further proceedings.

Takeaway

Grammar matters.  This decision illustrates that the Federal Circuit will treat claim construction as a matter of ordinary English when the patent record does not clearly dictate a special meaning.  When a limiting phrase could modify more than one element, the default is the nearest sensible referent.  If patentees intend to tie an element like “encryption information” to a specific storage location, they should say so expressly in the claim or specification to avoid broader constructions that make it easier for petitioners to show a limitation in the prior art.