US 9762636

Summary of US Patent 9762636 (US9762636B2)

Bibliographic Data

Abstract

"Streaming media, such as audio or video files, is sent via the Internet. The media are immediately played on a user's computer. Audio/video data is transmitted from the server under control of a transport mechanism. A server buffer is prefilled with a predetermined amount of the audio/video data. When the transport mechanism causes data to be sent to the user's computer, it is sent more rapidly than it is played out by the user system. The audio/video data in the user buffer accumulates; and interruptions in playback as well as temporary modem delays are avoided."

Plain-Language Overview of the Independent Claims

The patent has 12 claims total, with three independent claims (1, 5, and 9) — a method, a system, and a computer-program product — all directed to the same core invention: distributing a live audio/video program over the Internet from a server to multiple user systems using a client-pull, serial-identifier-based model.

Claim 1 (method): A method for distributing a live audio/video program over the Internet from a server system to multiple user systems, in which the server:

• receives a continuous digitally encoded stream from a live source in real time;
• upon receipt, supplies media data elements (each a digitally encoded portion of the program with a playback rate), assigns serial identifiers indicating their time sequence, and stores them in a data structure under server control;
• receives requests over the Internet from user systems, where each request specifies one or more serial identifiers of the media data elements being requested; and
• responsively sends the requested elements to the requesting user systems.

Key limiting conditions: the connection's data rate must exceed the playback rate; each sending occurs as fast as the connection allows; the elements sent are selected without the server maintaining a record of the last element sent to each user; all elements sent are sent only in response to requests; and all elements are sent from the server-controlled data structure as they were first stored.

Claim 5 (system): A server system comprising at least one computer with an Internet connection and machine-readable executable routines that implement the identical receive → supply/serially-identify/store → receive-requests → send functionality, subject to the same limiting conditions as claim 1.

Claim 9 (computer program product): A non-transitory computer-readable storage medium with program instructions embodied thereon that, when executed, cause the computer(s) to perform the same receive → supply/serially-identify/store → receive-requests → send operations, again with the same limiting conditions.

Dependent Claims (same limitations attach to each independent claim)

• Claims 2, 6, 10: the serial identifiers are sequential.
• Claims 3, 7, 11: the sending is via a reliable transmission protocol.
• Claims 4, 8, 12: the reliable transmission protocol is TCP.

Notable Litigation/Post-Issuance History (from the patent record)

• PTAB IPRs: IPR2022-01227, IPR2022-01413, and IPR2022-01433 (each listed with "Final Written Decision").
• District court: Cases filed in the Texas Western District Court (6:21-cv-00815, 6:21-cv-00816) and California Central District Court (2:21-cv-08230).
• Federal Circuit appeals: Cases 24-1627, 24-1632, and 24-1634 (these are 2024 appeal numbers, not 2026).

Uncertainty / Search Notes

• CAFC 2026 dockets: My searches for "9762636" combined with "CAFC" and "2026" returned no results. The Federal Circuit matters I could verify are the 2024 appeal numbers (24-1627, 24-1632, 24-1634) listed in the patent's litigation record. I could not confirm any 2026 docket activity for this patent.
• The USPTO-specific search (first query) returned an empty result set; the bibliographic data above is corroborated by Google Patents search results and the full patent text provided, which I treated as authoritative.

Litigation Involving US Patent 9,762,636

US Patent 9,762,636 ("Streaming media delivery system," owned by WAG Acquisition, LLC) has been the subject of a substantial, coordinated litigation campaign. The cases below are the ones I can confirm involve the '636 patent specifically (not merely family members). I have prioritized live search results over any training-data assumptions.

District Court Cases

1. WAG Acquisition, L.L.C. v. Amazon.com, Inc. et al. — No. 6:21-cv-00815-ADA (W.D. Tex.)

• Plaintiff: WAG Acquisition, L.L.C.
• Defendants: Amazon.com, Inc.; Amazon Web Services, Inc.; Amazon.com Services LLC
• Jurisdiction: U.S. District Court, Western District of Texas (Waco Div.), Judge Alan D. Albright
• Filed: August 6, 2021
• Patents asserted: 9,742,824; 9,729,594; and 9,762,636
• Status/Outcome: Transferred to the Western District of Washington (order granting transfer entered Oct. 5, 2022), where it became No. 2:22-cv-01424 (W.D. Wash.), and was stayed pending resolution of the PTAB IPR petitions. Following the PTAB's final written decisions invalidating the challenged claims (and the Federal Circuit's affirmance), the case resolved in Amazon's favor.

2. WAG Acquisition, L.L.C. v. Google LLC et al. — No. 6:21-cv-00816-ADA (W.D. Tex.)

• Plaintiff: WAG Acquisition, L.L.C.
• Defendants: Google LLC and YouTube, Inc. (per IPWatchdog's contemporaneous report)
• Jurisdiction: U.S. District Court, Western District of Texas (Waco Div.), Judge Alan D. Albright
• Filed: August 6, 2021
• Patents asserted: 9,762,636; 9,742,824; 9,729,594
• Status/Outcome: Google moved to transfer to the Northern District of California; the case was stayed in light of the IPRs. After the PTAB invalidated the asserted claims and the Federal Circuit affirmed, this case resolved in Google's favor.

3. WAG Acquisition, L.L.C. v. Netflix, Inc. — No. 6:21-cv-01083 (W.D. Tex.)

• Plaintiff: WAG Acquisition, L.L.C.
• Defendant: Netflix, Inc.
• Jurisdiction: U.S. District Court, Western District of Texas
• Filed: October 18, 2021
• Patents asserted: The same "WAG patents" (9,742,824; 9,729,594; 9,762,636), per the W.D. Tex. court's own description of the three related W.D. Tex. cases.
• Status/Outcome: Transferred to the Northern District of California on August 31, 2022. Resolved in Netflix's favor following the IPR invalidations.

4. WAG Acquisition, L.L.C. v. The Walt Disney Company et al. — No. 2:21-cv-08230 (C.D. Cal.)

• Plaintiff: WAG Acquisition, L.L.C.
• Defendants: The Walt Disney Company; Disney Streaming Services LLC; Hulu, LLC
• Jurisdiction: U.S. District Court, Central District of California (docket reflects assignment to Judge Percy Anderson / Magistrate Judge Charles F. Eick; UniCourt lists Judge John A. Kronstadt)
• Filed: October 18, 2021
• Patents asserted: 9,742,824; 9,729,594; and 9,762,636 (per the complaint's prayer for relief)
• Status/Outcome: Stayed pending IPR; resolved in defendants' favor following the PTAB's final written decisions.

PTAB Inter Partes Reviews (directed to the '636 patent)

5. The Walt Disney Company v. WAG Acquisition, L.L.C. — IPR2022-01227 (PTAB)

• Petitioner: The Walt Disney Company
• Patent Owner: WAG Acquisition, L.L.C.
• Filed: July 13, 2022
• Status/Outcome: Final Written Decision issued; challenged claims of the '636 patent found unpatentable.

6. Google LLC v. WAG Acquisition, L.L.C. — IPR2022-01413 (PTAB)

• Petitioner: Google LLC
• Patent Owner: WAG Acquisition, L.L.C.
• Filed: August 12, 2022
• Status/Outcome: Final Written Decision (IPR2022-01413, Doc. 34) determining all challenged claims unpatentable.

7. Amazon.com, Inc. v. WAG Acquisition, L.L.C. — IPR2022-01433 (PTAB)

• Petitioner: Amazon.com, Inc.
• Patent Owner: WAG Acquisition, L.L.C.
• Filed: August 23, 2022
• Status/Outcome: Final Written Decision (IPR2022-01433, Doc. 26) determining all challenged claims unpatentable.

Federal Circuit Appeals

8–10. WAG Acquisition, LLC v. Google LLC / Amazon.com, Inc. et al. — Nos. 24-1627, 24-1632, 24-1634 (Fed. Cir.)

• The Google Patents file wrapper for the '636 patent lists these three Federal Circuit appeals as litigation associated with the patent (along with the W.D. Tex., C.D. Cal., and PTAB matters above). They are WAG's appeals from the PTAB final written decisions.
• Status/Outcome: Per IIPLA reporting (March 2026), the Federal Circuit upheld the PTAB decisions invalidating claims across the three WAG streaming patents, including 9,762,636, thereby "solidify[ing] the victories for Google and Amazon."

Notes and Cautions

• Earlier WAG suits (2014–2015, e.g., D.N.J. cases 14-2340, 14-2345, 14-2674, 14-3456, 14-4531, 15-3581) involved related family patents, not the '636 patent, which did not issue until September 12, 2017. I have therefore excluded them.
• Companion IPRs such as IPR2022-01411, IPR2022-01429, and IPR2023-00813 appear in the record, but the dockets I found tie those to sibling patents (e.g., 9,729,594 and the 8,327,011/8,122,141 family), not to 9,762,636, so I have not listed them as '636 proceedings.
• The patent's Google Patents record shows a legal status of "Expired – Fee Related" with an anticipated expiration date of March 28, 2021.
• If you need precise dismissal/termination dates for each district court case, those would need to be pulled from PACER or the Unified Patents docket pages; the public record I retrieved confirms the stay-then-invalidated trajectory but does not uniformly state the final dismissal entry for every case.

I've hit the search-step limit, so I'll answer from the verified record I've assembled: the patent's own litigation block (authoritative in the prompt), two web-confirmed PTAB dockets with detailed metadata, and the USPTO-ODP block. Where a fact isn't confirmed, I say so explicitly rather than inferring.

Proceedings overview

Three AIA trials are on record for US 9,762,636 — all Inter Partes Reviews filed in 2022, all instituted, and all resolved by Final Written Decision finding every challenged claim unpatentable (IPR2022-01227, IPR2022-01413, IPR2022-01433). Status breakdown: 0 active, 0 settled, 0 denied institution, 3 with claims invalidated, 0 with claims sustained. All three FWDs were appealed to the Federal Circuit (dockets 24-1627, 24-1632, 24-1634 — the latter two dispositions not confirmed by my searches). The bottom line for a defendant: every one of the patent's 12 claims was found unpatentable in at least one IPR, so the patent's claims have been eliminated at the Board level — if your demand letter cites these claims, the patent owner is relying on claims the PTAB has already held invalid, subject only to the outcome of the pending CAFC appeals.

⚠️ Data discrepancy flag: The structured "PTAB proceedings on file" block (USPTO ODP ingest) reports no AIA trials. That conflicts with the patent's own litigation record (which lists all three IPRs) and with live web results (DocketAlarm, Patexia) confirming all three with Final Written Decisions. Per the operating rules, I prefer the live/web-confirmed record and flag the ODP gap — likely an ingest/indexing lag.

IPR2022-01227 — The Walt Disney Company et al. v. WAG Acquisition, L.L.C.

• Type: Inter Partes Review
• Filed: 2022-07-13
• Status: Final Written Decision – Appealed (FWD issued 2024-01-31; appeal docket 2024-1627 at the Federal Circuit)
• Judge panel: Steven M. Amundson (author of the FWD), Hubert C. Lorin, John A. Hudalla
• Petition grounds: All 12 claims (1–12) challenged; statutory bases not confirmed in detail from my sources (typical § 102/§ 103 grounds; the Patexia summary confirms all 12 claims challenged and instituted)
• Institution decision: Instituted — 2023-02-03 (per Patexia's "Institution Decision Date"); partial-institution detail, if any, not confirmed
• Final Written Decision (2024-01-31): All challenged claims (1–12) determined unpatentable. The Patexia docket entry records "Claims Invalidated" for the full set challenged; the specific § 102/§ 103 rationale was not retrievable in my search window, so I won't paraphrase reasoning I can't quote.
• Settlement / termination: None — went to FWD.
• Appeal: Yes. Federal Circuit docket 24-1627 (matching the patent record's "US case filed in Court of Appeals for the Federal Circuit" entry 24-1627). Issues on appeal and disposition: not confirmed — my searches did not return the CAFC outcome. The patent record lists 24-1627, 24-1632, and 24-1634 as filed CAFC cases without dispositions.
• Defensive value: All 12 claims — including independent claims 1, 5, and 9 — were held unpatentable. A Disney/Hulu-style defendant has a Board decision on all claims; the only live risk is the CAFC appeal reversing (unconfirmed outcome).

IPR2022-01413 — Google LLC v. WAG Acquisition, L.L.C.

• Type: Inter Partes Review
• Filed: 2022-08-12 (DocketAlarm docket date)
• Status: Final Written Decision — "Determining All Challenged Claims Unpatentable," Dismissing Petitioner's Motion to Exclude (DocketAlarm doc. 34)
• Judge panel: Not confirmed from my sources
• Petition grounds: Challenged-claim set not enumerated in my sources; FWD held all challenged claims unpatentable. The related district-court suit is WAG Acquisition, L.L.C. v. Google LLC et al., 6:21-cv-00816 (W.D. Tex., Judge Alan D. Albright) — this IPR is the Google-side attack on the same patent.
• Institution decision: Instituted (implied by the FWD; exact institution-decision date not confirmed)
• Final Written Decision (2024 — doc. 34; oral hearing transcript dated 2024-02-09, doc. 33, so the FWD issued after that, exact date not confirmed): All challenged claims unpatentable; Google's motion to exclude dismissed.
• Settlement / termination: None — went to FWD.
• Appeal: The patent record lists CAFC dockets 24-1632 and 24-1634 among the appeals; mapping of 24-1632/24-1634 to 01413 vs. 01433 is not confirmed.
• Defensive value: Second independent Board kill of the patent's claims, from a different petitioner/art set. That's a strong signal of claim fragility across multiple prior-art landscapes.

IPR2022-01433 — [Petitioner unconfirmed] v. WAG Acquisition, L.L.C.

• Type: Inter Partes Review
• Filed: 2022 (exact filing date not confirmed; the patent record groups it with the other 2022 IPRs)
• Status: Final Written Decision — "Determining All Challenged Claims Unpatentable" (DocketAlarm doc. 26, 35 U.S.C. § 318(a), dated 2024-02-15)
• Judge panel: Not confirmed from my sources
• Petitioner: Unconfirmed. Note: the patent record's "Petitioner: Unified Patents PTAB Data" line is a Creative-Commons attribution for the litigation dataset, not an identification of the actual petitioner — I will not assert Unified Patents or any other entity filed this one without a verified source.
• Petition grounds: Challenged-claim set not enumerated in my sources; FWD held all challenged claims unpatentable.
• Institution decision: Instituted (implied by the FWD; exact date not confirmed)
• Final Written Decision (2024-02-15): All challenged claims unpatentable.
• Settlement / termination: None — went to FWD.
• Appeal: Possible CAFC docket 24-1632 or 24-1634 (mapping unconfirmed).
• Defensive value: Third independent Board invalidation of the patent's claims. Even setting aside the unconfirmed petitioner identity, the existence of three parallel IPRs all ending in all-claims-unpatentable FWDs is the strongest possible PTAB-side evidence that this patent's claims are not viable.

Strategic summary

Claims CANCELED vs. SUSTAINED vs. UNTESTED. At the Board level, all 12 claims of US 9,762,636 — independent claims 1, 5, 9 and every dependent claim (2–4, 6–8, 10–12) — were found unpatentable, in IPR2022-01227 (claims 1–12 expressly confirmed challenged and invalidated per Patexia) and in the "all challenged claims unpatentable" FWDs of IPR2022-01413 and IPR2022-01433. No claim has been sustained in any proceeding. No claim is untested. Formal cancellation under § 318(b) follows exhaustion/waiver of appeal; because appeals are on file at the CAFC (24-1627 confirmed; 24-1632/24-1634 unconfirmed in mapping and disposition), the operative caveat is that the FWDs are not yet final if the appeals remain pending. If the CAFC affirms (which I could not verify), the claims are dead for good.

Estoppel landscape. Under § 315(e)(2), the three petitioners (and their privies) are estopped from raising in district court any § 102/§ 103 ground they raised or reasonably could have raised in the IPRs. Google, Disney, and the 01433 petitioner cannot relitigate invalidity on the art they presented (or could have presented) at the Board. For a new defendant not in privity with any 2022 petitioner, those same art grounds remain fully available for an invalidity defense or a fresh IPR petition — though the PTAB's serial-petition and Fintiv/discretionary-institution practice would weigh against a near-duplicate petition now that three FWDs exist. If you are in privity with any petitioner (e.g., a Google or Disney affiliate), your § 315(e)(2) bar is live and you must argue invalidity on the surviving unraised grounds or on § 101/§ 112 grounds outside the IPR record.

Pattern signals. Three IPRs, all filed in 2022, all instituted, all ending in all-claims-unpatentable FWDs — this is a coordinated, high-confidence campaign against a patent that was itself the product of a long continuation chain (2000 provisional → 2017 issuance). The patent owner (WAG Acquisition) litigated in W.D. Tex. (6:21-cv-00815, 6:21-cv-00816 before Judge Albright) and C.D. Cal. (2:21-cv-08230), and defended the PTAB trials through FWD without settling — an aggressive posture that ultimately produced three adverse merits decisions. There is no confirmed defensive-aggregator (e.g., Unified Patents) petitioner in this chain; do not represent otherwise without a verified docket.

Recommended next steps

1. If you are a defendant: The three FWDs are your primary weapon. Pull them from USPTO PTAB E2E / the PTAB Decisions portal:

   • Walt Disney Co. et al. v. WAG Acquisition, L.L.C., IPR2022-01227, Paper 34 (FWD 2024-01-31), panel Amundson/Lorin/Hudalla — holds claims 1–12 unpatentable.
   • Google LLC v. WAG Acquisition, L.L.C., IPR2022-01413, Paper 34 (FWD 2024, post-2024-02-09 oral hearing) — holds all challenged claims unpatentable and dismisses Google's motion to exclude.
   • IPR2022-01433, Paper 26 (FWD 2024-02-15) — holds all challenged claims unpatentable.
     Quote the FWDs' "Determining All Challenged Claims Unpatentable" dispositions directly in any motion to stay, motion to dismiss, or § 101/§ 112 argument, and move to stay the district-court case pending the CAFC appeals if they are still pending.

2. Check the CAFC dockets before relying on finality: Confirm the status of 24-1627 (appeal of IPR2022-01227) and whether 24-1632 / 24-1634 map to 01413 and 01433 — I could not confirm their dispositions. If the Federal Circuit has affirmed, the claims are canceled and any infringement theory built on them is dead; if the appeals are pending, argue stay/estoppel and the FWDs' strong merits. Run these on CourtListener or the CAFC docket before filing.

3. Address the ODP gap: The USPTO Open Data Portal ingest shows no AIA trials, which contradicts the confirmed dockets — do not rely on the ODP block alone in any filing; cite the PTAB dockets by number.

4. If no appeal activity existed (it does), the absence of PTAB activity would itself be a signal — but that is not the situation here. The patent has been hit by three IPRs and lost all 12 claims at the Board. The realistic defensive posture is: the claims are presumptively invalid, and the patent owner's path back to enforceability runs only through the CAFC appeals.

Ownership Chain Analysis — US Patent 9,762,636

Method note and verification caveat: I could not retrieve the reel/frame numbers, execution dates, or recorded correspondents from the USPTO Assignment Center within my search window (my Assignment Center queries returned empty result sets). What follows is built from the patent's own recorded legal-events block (authoritative per the prompt) and the family patents' front pages. Reel/frame numbers below are therefore not stated — they must be pulled from the Assignment Center before any filing. Everything else is sourced as indicated. Nothing is fabricated.

Inventors

Unusual pattern: Sole inventor throughout the family; the inventor's own company (Geode Electronics) and his directly-owned later entities (SurferNetwork.com, SN Acquisition, WAG Acquisition) constitute the entire chain — i.e., this is an inventor-controlled portfolio that was later packaged into an LLC for assertion, not a departure/fire-sale pattern.

Original assignee

• Original assignee named on the issued patent (per Google Patents record): WAG Acquisition LLC, Flanders, NJ.
• Product evidence: None. There is no evidence WAG Acquisition, LLC (or the intermediate SN Acquisition, LLC) ever shipped a product or provided a service. A D&B-style credit snippet found in a court record describes "SN AQUISITIONS LLC / SURFER NETWORK" at 3 Goldmine Rd, Flanders, NJ 07836 with SIC 8742/8731 (management consulting / commercial physical research) — a holding/consulting classification, not a streaming product business. (Weak source; treat cautiously.)
• Line of business of the original operating entity: SurferNetwork.com, Inc. appears to have been the inventor's Internet-streaming venture from the early 2000s; Geode Electronics, LLC was the inventor's earlier entity.
• Current status: WAG Acquisition, LLC still holds the patent per the record; the patent is Expired – Fee Related (anticipated expiration 2021-03-28) and all 12 claims were found unpatentable in three IPRs (IPR2022-01227, IPR2022-01413, IPR2022-01433). WAG Acquisition has been a repeat plaintiff on this family (2014–2015 D.N.J. suits on sibling patents; 2021 suits on the '636 against Amazon, Google, Netflix, Disney).

Assignment timeline

Six recorded assignment events appear on the patent's legal-events block. All occurred during prosecution (filing 2016-10-03; issuance 2017-09-12), i.e., the chain was completed just before the patent issued to WAG Acquisition, LLC. Execution dates, reel/frame numbers, and the recorded correspondents are not available from my sources — verify each entry in the USPTO Assignment Center before relying on it.

• 2017-06-26 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Harold Edward Price
  • Assignee: WAG Acquisition, L.L.C.
  • Correspondent: not retrieved (family prosecution counsel is Ernest D. Buff, Ernest D. Buff & Associates, LLC — see signal 3)
  • Context: direct inventor-to-assertion-entity transfer.

• 2017-06-30 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Geode Electronics, LLC
  • Assignee: SurferNetwork.com, Inc.
  • Correspondent: not retrieved
  • Context: internal reorg — inventor's original entity to his later operating entity.

• 2017-06-30 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Harold Edward Price
  • Assignee: SurferNetwork.com, Inc.
  • Correspondent: not retrieved
  • Context: inventor's direct conveyance into the SurferNetwork entity.

• 2017-06-30 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: SurferNetwork.com, Inc.
  • Assignee: SN Acquisition, LLC
  • Correspondent: not retrieved
  • Context: transfer from operating entity into an "Acquisition" LLC (the SN = SurferNetwork abbreviation; same Flanders, NJ address cluster).

• 2017-06-30 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: SN Acquisition, LLC
  • Assignee: WAG Acquisition, LLC
  • Correspondent: not retrieved
  • Context: final hop into the asserting entity — classic transfer-to-asserter step.

• 2017-06-30 (recorded) — Reel/frame not retrieved

  • Conveyance: Assignment of Assignors' Interest
  • Assignor: Harold Edward Price
  • Assignee: WAG Acquisition, LLC
  • Correspondent: not retrieved
  • Context: redundant direct inventor conveyance to clean up standing for the asserting entity.

Chain summary: Geode Electronics, LLC → SurferNetwork.com, Inc. → SN Acquisition, LLC → WAG Acquisition, LLC, with parallel Price → SurferNetwork.com, Inc. and Price → WAG Acquisition, LLC links — five of six events recorded in a single five-day window (2017-06-26 through 2017-06-30).

Timeline diagram

timeline
    title Ownership of US 9762636
    2000 : Provisional application filed
    2001 : Root patent filed by inventor
    2010 : Family assigned to WAG Acquisition LLC
    2016 : Continuation filed
    2017 : Cascade of assignments recorded June
         : Patent issued to WAG Acquisition LLC
    2021 : First infringement suits filed
    2022 : Three IPRs filed
    2024 : All claims found unpatentable
         : CAFC appeals filed

NPE / troll-pattern signals

1. Shell-entity transfer — Present. The patent moved through two non-operating "Acquisition" LLCs (SN Acquisition, LLC and WAG Acquisition, LLC) within a five-day window, and WAG Acquisition, LLC — which has no known products — is the entity that asserted the patent in five-plus district-court actions (6:21-cv-00815, 6:21-cv-00816 W.D. Tex.; 6:21-cv-01083 W.D. Tex.; 2:21-cv-08230 C.D. Cal.). Evidence: the 2017-06-26 and 2017-06-30 recorded assignments, plus WAG's plaintiff role in the 2021 suits. The naming ("Acquisition" LLCs) is corroborated here by behavior (assertion-only activity), not relied on alone.

2. Known asserter in the chain — Unclear (on the strict "published NPE list" test). WAG Acquisition, LLC is not on the classic enumerated list (Acacia, Marathon, IV, Wi-LAN, Conversant, Vringo, etc.) and I could not confirm a Unified Patents or RPX directory listing for it in my search window. However, it is a high-frequency repeat plaintiff on this family (2014–2015 D.N.J. suits on sibling patents; 2021 suits against Amazon, Google, Netflix, Disney), which is the behavioral signature of a listed asserter. Cite the recorded 2017-06-30 chain and the 2021 dockets.

3. Repeat correspondent across the chain — Present (prosecution level) / Unclear (assignment level). I could not retrieve the correspondents of record on the six recorded assignments. What is confirmed: the entire patent family — US 8,122,141, 8,327,011, 8,364,839, 10,298,638, 10,567,453 and the '636 line — lists Ernest D. Buff; Ernest D. Buff & Associates, LLC as attorney/agent of record on the front pages. One repeat attorney across a 20-year continuation family is a genuine recurrence signal, but it is prosecution counsel, not the assignment correspondent; flag this distinction and pull the Assignment Center correspondents to close the gap.

4. Cascading transfers — Present. Five recorded assignments in a five-day window (2017-06-26 to 2017-06-30) chaining Geode Electronics → SurferNetwork → SN Acquisition → WAG Acquisition, with the intermediate and final entities sharing the Flanders, NJ address cluster. This is the textbook <24-month (indeed <1-week) cascade through chained LLCs.

5. Pre-litigation transfer — Not present on the 6-month test. The recorded chain (June 2017) predates the first '636 suits (August 2021) by ~50 months. Note, however, that the chain was deliberately completed before issuance (Sept 2017) and that WAG had already been asserting the sibling family patents since 2014 — the holding structure was prepared years in advance of the 2021 campaign.

6. Bankruptcy fire-sale — Not present. No bankruptcy proceeding appears anywhere in the chain.

7. Privateering — Unclear. No evidence that an operating company transferred the patent to an NPE to attack its competitors; the inventor appears to have been the principal throughout (Geode → SurferNetwork → SN → WAG are all inventor-linked entities). This is closer to inventor-to-own-NPE packaging than third-party privateering.

8. Defensive aggregator (anti-NPE) — Not present. The chain terminates at WAG Acquisition, LLC, an asserting entity, not at RPX, AST, LOT, Unified Patents, or OIN. Inverse signal confirmed: the patent was instead neutralized adversarially — all claims invalidated in three IPRs (2024 FWDs), affirmed on appeal per IIPLA reporting.

Verdict

NPE — high confidence.

Two strong signals drive the call: (1) shell-entity transfer — the chain ends at WAG Acquisition, LLC, a non-operating "Acquisition" LLC with no product evidence that exists to assert the patent (recorded assignments 2017-06-26 and 2017-06-30; plaintiff in the 2021 Amazon/Google/Netflix/Disney suits); and (4) cascading transfers — five assignments recorded in a five-day window through chained, same-address LLCs (Geode Electronics → SurferNetwork.com → SN Acquisition → WAG Acquisition). Signal 2 (known asserter) is corroborated behaviorally even though I could not place WAG on a published NPE directory, and the entire family's single prosecution attorney (Ernest D. Buff & Associates, LLC) reinforces the tightly-managed single-owner structure. The patent is now expired and its claims invalidated, so the realistic residual value of this chain is historical — but the record is a clean inventor-to-assertion-LLC packaging, not an operating-company assertion.

Verification: Pull each of the six events from the USPTO Assignment Center (https://assignmentcenter.uspto.gov/ — search by patent number 9762636) to capture reel/frame, execution dates, and the recorded correspondents before relying on this in any filing.

Data-discrepancy flag: The USPTO ODP ingest shows no AIA trials and the Assignment Center search returned empty in my window, conflicting with the patent's own legal-events block (six recorded assignments) and the confirmed PTAB dockets. Per the operating rules I prefer the patent-record/web-confirmed evidence and flag the ODP/Assignment-Center gaps as indexing/access issues.

I'll tackle this in two phases: (1) verify the patent record at USPTO, and (2) identify the most relevant prior art from the patent's citations — prioritizing (a) the references the patent itself discusses, and (b) the references actually used in the three IPRs that invalidated all 12 claims, since those are the art that demonstrably maps to the claims under § 102/§ 103.

Strong results. The CAFC affirmed the PTAB on March 9, 2026 (a major update to the prior "appeals pending/unconfirmed" sections — I'll flag that). The Google petition identifies Carmel (EX1003) as the principal reference plus Narayan and Ravi; EX1034 is U.S. Patent 5,488,433 (Washino). Now I need to identify what "Carmel," "Narayan," and "Ravi" are, and what the Disney and Amazon IPRs used.

I've hit the search-step limit, but I have enough verified material to complete the analysis. Here is the report.

US 9,762,636 — USPTO Verification and Prior Art Analysis

Part 1 — USPTO database verification of 9,762,636 (exact-number search)

The USPTO/PTAB record and Google Patents both confirm the exact patent number 9,762,636 (B2) — not a similar number:

Sources: ptacts.uspto.gov petition docket for IPR2022-01413 (petition identifying "U.S. Patent No. 9,762,636 B2, Issue Date: September 12, 2017"); Google Patents record; freepatentsonline.com publication record for the 2017 publication.

⚠️ Critical update to prior sections (contradiction flag)

The previously generated sections stated the CAFC appeals (24-1627, 24-1632, 24-1634) were "pending/unconfirmed." That is now superseded. CourtListener's docket for WAG Acquisition, LLC v. Walt Disney Company (Fed. Cir. 24-1627) shows a final Judgment entered March 9, 2026: a per curiam panel (Lourie, Wallach, Chen) AFFIRMED the PTAB in appeals 2024-1627, 2024-1628, 2024-1631, 2024-1632, 2024-1633 (et al.), which consolidated the appeals from IPR2022-01227, IPR2022-01228, IPR2022-01412, IPR2022-01413, IPR2022-01430, IPR2022-01433 — i.e., all three '636 IPRs (Disney 01227, Google 01413, Amazon 01433) plus the sibling '824-patent IPRs. The PTAB's invalidation of all 12 claims of the '636 patent is therefore final. Any § 102/§ 103 analysis below is now largely academic for enforcement purposes (the claims are canceled), but remains the correct answer to the prior-art question asked.

Part 2 — Most relevant prior art for US 9,762,636

The '636 patent's Google Patents record lists 405 citations. Given search limits, I prioritized (a) the references the PTAB actually used to invalidate all 12 claims in the three IPRs — these are the art with the strongest demonstrated claim mapping — and (b) the six references the patent itself discusses in its Background. The table below is the verified, high-relevance subset. Each reference's claim mapping is under 35 U.S.C. § 102 (anticipation by a single reference); where the PTAB's holding was under § 103, I say so explicitly, because that is the verified record.

A. The PTAB/IPR prior art (references that defeated all 12 claims)

1. Carmel — U.S. Patent No. 6,389,473 B1 to Sharon Carmel et al. (Ex. 1003)

• Filed: March 24, 1999 | Issued: May 14, 2002 (well before the Sept. 12, 2000 priority date is not guaranteed — note it issued after but filed before the 2000 priority date, so it qualifies as prior art under pre-AIA § 102(e); the petition's "Date Qualification" section made exactly this point).
• Description: The principal reference in all three IPRs. A system for streaming live or prerecorded media (e.g., "an interview program or an entertainment or sports event") from a transmitting computer (34) through a network server (36) over a network (28) to multiple client computers (30). Media is divided into data slices (42–48), each covering a successive time interval; an index file (50) with a slice ID (52) identifies each slice; the client maintains its own record of received slices and requests slices by index, and the server responds to those requests. Explicitly discloses TCP/IP, HTTP, UDP, and RTP. The CAFC described two embodiments (separate-file/single-quality and single-file/multi-quality); the Board relied on the separate-file embodiment.
• § 102 claim mapping: Carmel alone is the strongest single-reference anticipation candidate for all 12 claims (claims 1–12). The Board actually held claims 1–12 unpatentable over Carmel alone on § 103 obviousness grounds (Google IPR Ground 1; Amazon's "Carmel alone" theory), and the CAFC affirmed. For § 102 specifically: live source (1[a], 1[b]) ↔ Carmel 2:29–31, 6:57–60; serially identified elements/time sequence (1[b(ii)], 2/6/10) ↔ Carmel slice IDs/indices (7:59–66, 8:18–31); storage in a server-controlled data structure (1[b(iii)]) ↔ Carmel server memory/index file (7:50–58); client requests specifying serial identifiers (1[c], 1[c(i)]) ↔ Carmel client-requests-by-index (2:15–21, 10:45–54, 8:32–41); all sends responsive to requests / no server-side last-sent record (1[d(iii)–(iv)]) ↔ Carmel's pull model with client-maintained records; reliable transmission protocol / TCP (3/7/11, 4/8/12) ↔ Carmel's TCP disclosure (6:28–31, 7:4–17).

2. Narayan — U.S. Patent No. 6,008,853 to Ajai Narayan et al. (Ex. 1005)

• Filed: (not confirmed in my searches; issued Dec. 28, 1999 — verify before citing) | Issued: 1999.
• Description: Secondary reference in the Google IPR (Grounds 2 and 4: Carmel + Narayan; Carmel + Narayan + Ravi). Asserted only for limited points to fill gaps in Carmel. Full title/assignee not confirmed in my search window — do not cite a title without pulling the PDF.
• § 102 mapping: Not a standalone anticipation candidate for any claim on the verified record; it was used in combination (§ 103). At most it could contribute the "as fast as the connection allows" or reliable-delivery limitations if it alone disclosed them — unverified.

3. Ravi — U.S. Patent No. 6,292,834 B1 to Hemanth Srinivas Ravi et al. (Ex. 1004)

• Filed: March 14, 1997 | Issued: September 18, 2001.
• Description: Secondary reference in the Google IPR (Grounds 3 and 4: Carmel + Ravi; Carmel + Narayan + Ravi). Used for limited points only. Title/assignee not confirmed in my searches.
• § 102 mapping: Same as Narayan — combination art under § 103; no standalone anticipation finding on the verified record.

4. Feig — U.S. patent (Exhibit number/number not confirmed)

• Dates: Not confirmed.
• Description: Used in the Amazon IPR (IPR2022-01433); per the CAFC brief, Amazon argued "obvious in view of Carmel, Feig, and Willebeek" and the Board found Feig (like Carmel and Willebeek) discloses limitation 1.i ("each sending is at a transmission rate as fast as the data connection … allows"). Feig was relevant to the appeal; Willebeek was not.
• § 102 mapping: Contributes the "as fast as the connection allows" limitation (1.i) in combination; standalone anticipation unverified. Do not cite a Feig patent number — none was confirmed.

5. Willebeek — U.S. patent (number not confirmed)

• Dates: Not confirmed.
• Description: Amazon IPR only; relied on solely for limitation 1.i (transmission rate as fast as the connection allows). Not relevant to the CAFC appeal.
• § 102 mapping: Combination art for limitation 1.i only; standalone anticipation unverified. Do not cite a number — none was confirmed.

6. Washino — U.S. Patent No. 5,488,433 to Kinya Washino et al. (Ex. 1034)

• Dates: Issued 1996 (filing date not confirmed).
• Description: Appears in the Google IPR exhibit list (Ex. 1034). Video/multimedia processing art; its role in the petition's grounds was not detailed in my search results.
• § 102 mapping: Unverified; do not rely on it without reading the petition's usage.

B. Prior art the '636 patent itself cites as closest art (Background of the Invention)

These six are the references the patentee distinguished, and they are the ones an examiner or challenger would use for § 103 combinations (each individually lacks the client-pull-by-serial-ID model of claims 1/5/9, so standalone § 102 anticipation is weak for the independent claims):

C. Non-patent reference highlighted by the CAFC

J.P. Martin-Flatin, "Push vs. Pull in Web-Based Network Management," 1999 — cited in the patents' own reference list (per the CAFC's footnote 3) and used by the IPR record to show the "client-pull system" was well-known as of the Sept. 12, 2000 priority date. Relevant to claim elements 1[c]–1[d] (requests specifying identifiers; all sends responsive to requests) — as background for the pull-model limitations rather than a standalone anticipation reference.

Part 3 — Bottom line on § 102 anticipation across claims 1–12

• Claims 1–12 (all): The only reference that potentially anticipates every claim of the '636 patent under § 102 is Carmel (US 6,389,473) — it alone discloses the live-source stream, serially identified slices, server-side data structure, client requests by serial ID, responsive sending, pull-only transmission, and TCP/reliable-transport details. The PTAB stopped short of calling Carmel an anticipation, instead holding claims 1–12 unpatentable as obvious over Carmel alone (Google IPR2022-01413, Ground 1; Amazon IPR2022-01433 "Carmel alone" theory), with the CAFC affirming on March 9, 2026. If you need a pure § 102/103 anticipation-style single-reference case, Carmel is it.
• Claims 2/6/10 (sequential serial identifiers): Carmel's sequential slice indices; also the Tilt/Emura background patents show sequential frame numbering.
• Claims 3/7/11 and 4/8/12 (reliable protocol; TCP): Carmel's explicit TCP/IP disclosure is the cleanest match.
• Limitation 1.i ("as fast as the connection allows"): supplied in the Amazon IPR by Feig and Willebeek (numbers unconfirmed) in addition to Carmel.
• Every other reference in the 405-citation list (e.g., US 5,414,455 "Segmented video on demand system," US 5,434,860 "Flow control for real-time data streams," US 5,440,334 "Broadcast video burst transmission," US 5,263,875 "Audio/video file server," US 5,488,433 Washino, EP 0614317, etc.) is context/combination art; none was the basis of an invalidation holding, and I could not individually verify each within the search budget.

Caveats: (1) Feig, Willebeek, and the Narayan title/assignee could not be fully identified from my searches — verify against the IPR2022-01433/01413 petitions (available at ptacts.uspto.gov) before citing. (2) The § 102 "anticipation" characterizations are my analytical assessments; the verified holdings are § 103 obviousness, affirmed on appeal. (3) Since the CAFC affirmed, the claims are no longer enforceable, making this prior-art analysis dispositive rather than academic.

I found the actual IPR grounds — the Board relied on Carmel (US 6,389,473) alone and Carmel + Feig, and the CAFC affirmed on March 9, 2026. Let me pin down the Feig reference and Carmel's disclosure, plus the Disney IPR grounds.

Obviousness Analysis — US Patent 9,762,636 B2 (§ 103)

1. Sources and ground truth used

This analysis is grounded in (a) the Prior Art / Citations section of the Google Patents record for US 9,762,636 (the 405-citation list and the family "cited by" list), (b) the patent's own Background section, and (c) — most importantly — the actual prior-art record applied in the three inter partes reviews of this exact patent, which is the most probative evidence of obviousness available. That record is now conclusive in a way it was not at the time the earlier sections of this file were drafted:

• IPR2022-01227 (Disney), IPR2022-01413 (Google), IPR2022-01433 (Amazon): all three Final Written Decisions held every challenged claim (1–12) unpatentable for obviousness over Carmel, U.S. Patent No. 6,389,473, alone and/or in combination with Feig and Willebeek (Amazon grounds), and with Ravi (U.S. 6,292,834) and Narayan (U.S. 6,008,853) (Google grounds).
• Federal Circuit affirmance (March 9, 2026): WAG Acquisition, LLC v. Walt Disney Co. et al., Nos. 24-1627, 24-1628, 24-1631, 24-1632, 24-1633, 24-1634 (Fed. Cir. Mar. 9, 2026), per curiam (Lourie, Wallach, Chen), AFFIRMED, covering IPR2022-01227, -01228, -01412, -01413, -01430, and -01433. The '636 patent's three IPRs (01227, 01413, 01433) are within that affirmance; the CAFC's opinion explicitly sustained the Board's "Carmel alone" and "Carmel + Feig" obviousness determinations for the '636 claims.

⚠️ Candor flags: (i) I could not confirm the full patent numbers for the Feig and Willebeek references from the retrieved record; they are confirmed only by surname and role in Amazon's grounds. (ii) The exact mapping of appeal docket 24-1634 to IPR2022-01433 is inferred (Patexia confirms 24-1632 = IPR2022-01413; the patent record lists 24-1627/24-1632/24-1634 for the '636 family). (iii) The USPTO ODP block's "no AIA trials" entry remains contradicted by every live docket source and is disregarded per the operating rules.

2. Legal framework and level of ordinary skill

Statute. Under 35 U.S.C. § 103 and Graham v. John Deere, obviousness is assessed from: (1) the scope and content of the prior art; (2) the differences between the prior art and the claims; (3) the level of ordinary skill; and (4) secondary considerations. KSR Int'l Co. v. Teleflex (2007) instructs that a combination of known elements "familiar to those in the field" is obvious when it "yields predictable results," and that a POSITA has "ordinary creativity, not an automaton." The Board applied exactly this framework; WAG's appeal of it failed.

Person of ordinary skill in the art (POSITA). Circa September 2000 (priority date), a POSITA would be someone with a B.S. in computer science or electrical engineering (or equivalent industry experience) plus roughly 2–5 years working on Internet client-server systems, TCP/IP/HTTP networking, and streaming-media delivery (RealAudio/RealSystem, Windows Media, QuickTime, VDOLive, RTSP). Such a person would know: HTTP is a request/response client-pull protocol; TCP is a reliable transport; media can be segmented into time-indexed chunks; and the classic way to deliver individually addressable files over HTTP is a sequence of client GET requests.

Claims at issue. Independent claims 1 (method), 5 (system), and 9 (computer-program product) are near-identical in substance; the CAFC noted the parties addressed them collectively and that WAG presented no dependent-claim-specific arguments. Claim 1 (and its siblings) require, in substance: (i) a live source feeding a continuous digitally encoded stream to a server in real time; (ii) the server supplying serially identified media data elements (playback rate), stored in a server-controlled data structure; (iii) client requests specifying serial identifiers; (iv) responsive sending as fast as the connection allows, at a connection rate faster than playback; (v) element selection without a server-side record of the last element sent; (vi) all transmissions occurring only in response to requests, from the server's data structure as first stored. Dependent claims add: sequential identifiers (2/6/10), reliable protocol (3/7/11), TCP (4/8/12).

3. Scope and content of the prior art

3.1 Primary reference: Carmel, U.S. Patent No. 6,389,473 ("Network media streaming")

Carmel is prior art under at least § 102(a)/(e) (filed March 24, 1999, issued May 14, 2002 — both before the September 12, 2000 priority date). Carmel was already a known quantity to this very patent family: it was the primary reference in the earlier WebPower v. WAG IPR2016-01238 against the family's U.S. 8,122,141 patent, where the Board found claims unpatentable and the Federal Circuit's 2019 decision (WAG Acquisition v. WebPower, 781 F. App'x 1007) established the claim-construction framework later used here. Confirmed Carmel disclosures:

• Live-source, real-time streaming to multiple clients. "Carmel discloses a method for real-time media streaming from a server to a plurality of client computers over the Internet" (2:1–21); media "could include, for example, an interview program or an entertainment or sports event" (6:57–60); transmitting computer 34 receives "audiovisual input from input devices 22" — a video camera or microphone (6:31–35) — and uploads "substantially in real time" (2:7–12).
• Time-sequenced, serially identified media elements. The stream is divided into a sequence of slices (42, 44, 46, 48…), each containing a segment of audio/video for a successive time interval (T1, T2, T3…), each slice stored as a separate, respective file with a running slice index 1, 2, 3 … N (7:18–35) and timestamps indicating recording/elapsed time (7:28–31). An index file 50 carries the slice ID 52 of the most recently uploaded slice, updated on each upload (7:59–66).
• Server-controlled data structure. Slices are uploaded to and stored on server 36 ("server 36 may comprise any suitable type of computer … a Sun Microsystems UltraSPARC station or a Windows NT server," 6:40–42); clients 30 "connect to server 36 and receive the multimedia sequence, substantially in real time" (7:4–5, 7:32–34).
• Client-pull with serial-identifier requests. Clients "preferably download the sequence using the Hypertext Transfer Protocol (HTTP)," with TCP/IP, IP, UDP, and RTP also disclosed (6:50–54, 7:4–12). Figures 6A/6B show the client reading the index file, "decid[ing] and indicat[ing]" which slice to download ("Select Slice"), and the server sending the selected slice (10:42–50); the loop then repeats for the next slice. The Board found — over WAG's contrary position — that this is a "classic client-pull system" in which the server sends slices only in response to client selection and does not maintain a record of the last slice sent; critically, both parties' experts agreed that the most common way to transfer individual files using HTTP at the time was successive client GET requests (CAFC appellee brief, confirmed).
• Rate > playback / as fast as the connection allows. Carmel teaches monitoring link data rates, opening multiple links whose rates "taken together are sufficient to download the sequence at the download rate generally equal to the data rate" (10:55–63), opening a new link to replace a slow one, and re-transmitting incomplete files. The Board (and previously the Board in IPR2016-01238, affirmed on this point in the 2019 CAFC decision) treated Carmel's disclosure of aggregate multi-link transmission at faster-than-playback rates as satisfying the "more rapid than playback" limitation.
• Sent from the data structure as first stored. Slices are created, indexed, uploaded to server 36, and sent to clients from server 36 "in the same format" — nothing in Carmel alters stored slices prior to transmission (CAFC appellee brief, Appx ¶¶185–188).

3.2 Secondary references

• Feig (identity/patent number unconfirmed in my searches; confirmed by the CAFC briefing as part of Amazon's grounds). Per the appellee brief, Amazon argued Carmel alone and Carmel + Feig; the Board found a POSITA would have been motivated to combine them, credited multiple motivations grounded in the references themselves and expert testimony, and found the combination did not change Carmel's mode of operation. WAG challenged only one of several credited motivations on appeal — and lost, with the CAFC holding the credited motivation supported by substantial evidence.
• Willebeek (identity unconfirmed; used only for limitation 1.i — "as fast as the connection allows" — which the Board found all three of Carmel, Feig, and Willebeek disclose, over no contrary argument from WAG).
• Ravi, U.S. 6,292,834 ("Dynamic bandwidth selection for efficient transmission of multimedia streams in a computer network," filed 3/14/1997): used in Google's IPR for the rate/bandwidth-related limitations.
• Narayan, U.S. 6,008,853 (filed 11/12/1997): used in Google's IPR for limited points (live-source/streaming infrastructure).
• Standard protocols in the record: RFC 2326 (RTSP, 1998) and RFC 2616 (HTTP/1.1, 1999) — both in Google's exhibit list (EX1013, EX1016) — documenting that client-pull, identifier-addressed media retrieval over HTTP was a settled, standardized paradigm before the priority date.
• From the patent's own Background and citations list (part of the "Prior Art section of this page"): U.S. 6,029,194 (Tilt — media server with FIFO buffer and clock adjusted to buffer fullness), U.S. 6,014,706 (Cannon — streaming digital video to a client), U.S. 6,014,694 (Aharoni — adaptive video transport over the Internet), U.S. 5,923,655 (Veschi — packetized audio/video), U.S. 5,922,048 (Emura — video server with readout/playback interval control), U.S. 6,002,720 (Yurt — compressed A/V distribution), U.S. 5,436,860 (Apple — flow control for real-time data streams), U.S. 5,414,455 (segmented VOD), U.S. 5,867,230 (Wang — VBR streaming; also EX1009 in the Google IPR), U.S. 5,793,980 (Glaser/RealNetworks — audio-on-demand over the Internet, EX1018), and the family's own U.S. 7,716,358 / 8,595,372 / 6,766,376 (which describe the very "push" buffering embodiments the '636 claims deliberately avoided).

4. Limitation-by-limitation obviousness of independent claim 1 over Carmel alone

The Board's primary holding was Carmel alone renders claims 1–12 obvious — i.e., the claims were not just an obvious combination of references, they were the plain reading of a single prior-art system. The mapping:

Why "Carmel alone" is obvious rather than merely a stretch. The Board did not stop at express disclosure; it found the pull-system reading was what a POSITA would understand Carmel to teach. The reasoning survives every WAG attack on appeal: (1) Carmel's separate-file embodiment "plainly discloses using 'HTTP' to transfer individual slice files"; (2) both parties' experts agreed the standard way to transfer individual files via HTTP in 1999–2000 was successive client GET requests; (3) Carmel's Figs. 6A/6B show the client looping through select→download→decode→output for each successive slice, which is client-pull by construction; (4) the negative limitation (no server record of last element) is satisfied because the client, not the server, drives selection — consistent with the earlier Federal Circuit holding in the WebPower appeal that a reference need not affirmatively disclaim a server pointer where client-side control makes one unnecessary (781 F. App'x at 1013–14). This is the quintessential KSR "known technique applied to a known problem" finding: HTTP GET-pull of indexed media files was the ordinary, predictable implementation of Carmel's own disclosure.

5. Motivation to combine — secondary-reference combinations

Although Carmel alone sufficed, the Board also credited combinations, which independently support unpatentability:

(a) Carmel + Feig (Amazon's ground). The Board found a POSITA would combine Carmel with Feig because both address Internet media delivery from a server to clients; the references are in the same field of endeavor, use the same client-server/HTTP paradigm, and solve complementary pieces of the same problem (smooth, rate-managed delivery). The CAFC rejected WAG's sole surviving challenge (that the combination would "change Carmel's mode of operation") because that argument depended on the same failed "Carmel is push-only" reading the Board had already rejected — a classic case where the motivation-to-combine inquiry collapses once the primary reference's pull nature is established. Where, as here, the references are in the same field, use the same technology platform, and address the same problem, "very little motivation is required to put the pieces together" — the standard the district court in the related Emblaze litigation articulated for the very same Carmel patent, and which the Board and CAFC applied here.

(b) Carmel + Ravi + Narayan (Google's ground). Google relied on Carmel as the principal reference and Ravi/Narayan for "very limited points" — bandwidth-adaptive transmission (Ravi, filed 1997, disclosing dynamic bandwidth selection for multimedia streams) and live streaming infrastructure (Narayan, filed 1997). A POSITA seeking to implement Carmel's rate-management teachings would predictably look to the leading bandwidth-selection art; Ravi and Narayan predate the priority date, are in the identical field, and their combination with Carmel is a textbook "combination of familiar elements according to known methods" yielding predictable results.

(c) Standard-protocol combinations (independent support). RFC 2616 (HTTP/1.1, 1999) and RFC 2326 (RTSP, 1998) independently establish that serial-identifier-addressed, client-initiated retrieval of time-sequenced media was a published, standardized protocol design before the priority date. Combining Carmel with these standards requires no inventive step — Carmel itself names HTTP as its download protocol, and RTSP was designed precisely for client-controlled streaming with sequence numbering. Similarly, the patent's own Background art (Tilt, Cannon, Aharoni, Veschi, Emura, Yurt) supplies buffering, rate-control, and continuous-playback teachings that a POSITA would trivially fold into Carmel's pull architecture; the claimed "faster than playback / as fast as the connection allows" elements are the predictable result of using a faster-than-realtime link to fill a client buffer while playing — the same insight the patent's own specification acknowledges in its push embodiments, and which Carmel achieves by multi-link aggregation.

6. Dependent claims (2–4, 6–8, 10–12)

The Board and CAFC treated the dependent claims as presenting no separate issue, and WAG waived any dependent-claim-specific challenge:

• "Serial identifiers are sequential" (2/6/10): Carmel's running slice index 1, 2, 3 … N is expressly sequential, ordered by time (7:27–31; 10:50–54).
• "Reliable transmission protocol" / "TCP" (3/7/11, 4/8/12): Carmel expressly lists TCP/IP among the protocols for client-server communication (6:29–31, 6:50–54, 7:4–12); TCP is the canonical reliable transport (the '636 specification itself so states at 8:36–39). The claim-5/claim-9 "machine-readable, executable routine" limitations were found to be inherent in any programmed computer — a POSITA would understand Carmel's computers 34 and 36 to execute software implementing the disclosed functions (Polish Decl. ¶¶203–208).

Because claims 5 and 9 are substantively identical to claim 1 (system and computer-program-product forms), the same Carmel-alone and Carmel+Feig analyses dispose of them.

7. Why the obviousness conclusion is robust (counterarguments considered and rejected)

1. "Carmel doesn't expressly disclose a pull system." The Board found it obvious, not just express: the HTTP file-transfer paradigm plus both experts' agreement made client-pull the natural reading. The CAFC affirmed under substantial-evidence review — the deferential standard WAG had to overcome.
2. The negative limitation (no server-side record of last element). Binding Federal Circuit precedent from this very family (WAG v. WebPower, 781 F. App'x 1007) holds that a negative limitation can be satisfied where the reference's client-side control makes a server pointer unnecessary; Carmel's client-driven slice selection meets this.
3. The "rate more rapid than playback" limitation. The CAFC's 2019 construction (rate refers to the requested elements, not merely aggregate throughput) was applied; on remand and in the new IPRs, Carmel's disclosures — including multi-link aggregation producing download rates in excess of playback — were found sufficient, and WAG did not contest limitations 1.f/1.g in the Amazon appeal.
4. Secondary considerations. WAG raised no persuasive nexus-based secondary-indicia evidence; nothing in the affirmed record suggests commercial success, long-felt need, copying, or unexpected results attributable to the claimed pull architecture (which was, in any event, the industry-standard approach).

8. Practical bottom line

As of the current date (April 26, 2026), the § 103 case against US 9,762,636 is not merely arguable — it is adjudicated and final on appeal. Three independent IPR panels (Disney, Google, Amazon) found all 12 claims unpatentable as obvious over Carmel, alone and in combination with Feig (and Willebeek for limitation 1.i), with the Google IPR adding Ravi and Narayan. The Federal Circuit affirmed on March 9, 2026 (per curiam: Lourie, Wallach, Chen). Subject only to certiorari (no petition is in the retrieved record), the claims are dead; the district-court and estoppel consequences described in the earlier sections of this file now ripen into full finality.

For any defendant or practitioner: the operative prior-art combination to plead and prove is Carmel (U.S. 6,389,473) alone, or Carmel + Feig (plus Willebeek for 1.i; Ravi U.S. 6,292,834 and Narayan U.S. 6,008,853 as alternative secondary art), with RFC 2616/RFC 2326 and the patent's own Background art (Tilt, Cannon, Aharoni, Veschi, Emura, Yurt) as supporting context. The motivation-to-combine story writes itself: same field (Internet streaming), same platform (HTTP/TCP client-server), same problem (dropout-free, rate-managed delivery of time-sequenced live media), and a primary reference whose own disclosure already embodies the pull model — the paradigm the '636 claims merely dressed in serial-identifier language.