US 7698465

I'll search for litigation/docket information and corroborating records for this specific patent.

US Patent 7,698,465 — "Generalized Paxos"

Search scope note: I queried for USPTO/CAFC 2026 docket material on this exact number. The docket-oriented queries returned zero results, and I found no litigation or Federal Circuit appeal referencing 7,698,465 in any returned source. That is an absence of search hits, not affirmative proof that no such proceeding exists. Everything below on the patent itself is drawn from the granted patent text and USPTO-derived bibliographic mirrors, which are internally consistent.

1. Bibliographic data

Sources: https://patents.google.com/patent/US7698465/en ; https://patents.justia.com/patent/7698465 ; https://www.patents-review.com/a/20060136781-generalized-paxos.html

2. Abstract (as granted)

A distributed computing system can achieve a generalized consensus, enabling commands that commute to be selected in any order. A leader can learn of previously selected sequences of commands, and can propose a compatible sequence of commands. Devices can select a sequence of commands that is compatible with previously selected sequences. Additional commands can be selected by selecting a sequence of commands comprising a previously selected sequence and the additional commands. Further efficiencies can be realized if the devices receive proposals directly from clients. Two or more proposals arriving in varying orders at the various clients may be selected in varying orders. However, if those proposals commute, a generalized consensus nevertheless exists despite the variations, enabling the system to continue efficient operation. To conserve memory, a checkpoint command that does not commute with any other command can be used to secure a sequence of commands for archiving or deletion.

3. Plain-language overview of the independent claims

The core technical contribution is replacing agreement on one command per step with agreement on a "command structure" (c-struct) — a representation that treats all sequences of functions differing only in the relative order of commuting commands as mathematically equal. Claims 1 and 6 are verbatim-verified as independent.

Claim 1 — Method (leader-side consensus on a command structure). A method for selecting a proposed command structure to synchronize function execution across multiple devices:

• (a) transmit a proposed command structure to a first quorum, where the structure represents all equivalent sequences of functions, is voted on under an associated proposal number, and consists of one or more functions appended to a prefix (a smaller command structure from which the proposed one is derived);
• (b) receive acceptance messages — an accepting device may accept only if it has not responded to a higher suggested proposal number, and either it previously voted for that same prefix under that same proposal number or it has not voted for any command structure under that proposal number;
• (c) conclude the command structure is selected if acceptance messages arrive from a second quorum.

This is the "generalized Paxos" analogue of classic Paxos Phase 2, with sequence/prefix compatibility substituted for identity of a single command.

Claim 6 — Computer-readable storage medium (same three acts, embodied as instructions). Mirrors claim 1's transmitting / receiving / determining acts as computer-executable instructions. Note the receiving limitation is worded slightly more narrowly than claim 1 ("indicating an acceptance of the proposed command structure" without the "by one or more devices that have accepted" clause).

Dependent claims recovered verbatim:

• Claim 2 (dep. on 1): adds a Phase-1-style exchange — transmit a suggested next proposal number to a third quorum; receive a response acting as a promise not to accept lower-numbered structures, the response comprising (1) largest device-accepted proposal number, (2) the previously accepted command structure at that number, and (3) largest device-responded proposal number.
• Claim 3: if responses come from a fourth quorum, the structure to propose is either any previously accepted command structure indicated, or a common command structure having, as prefixes, all base command structures of a first set of quorums (the "convergence" rule).
• Claim 5: informing the system of selection, whereupon each device may retain its post-execution state, retain later-selected structures, and discard the proposed structure (memory reclamation).
• Claim 7 (dep. on 6): parallels claim 2.

Claim 4 — I retrieved only its number, not its text, in the available sources. Given the specification, the likely subject matter is the checkpoint / non-commuting-command aspect, but I will not assert that as fact.

⚠️ Uncertainty on claims 8–15. The 15-claim granted set is confirmed; verbatim text was only retrieved through claim 7. The published application US 2006/0136781 A1 carried 37 claims and included independent claims directed to a computing device (network interface + processing unit performing the same transmit/receive/determine acts, pub. claims 28 and 32) and an unusual "command structure comprising an array of functions" claim (pub. claim 37), plus dependent claims covering leaderless conflict resolution, checkpoint handling, learner-device reporting, and function identifiers. I cannot authoritatively confirm the granted numbering or survival of those independent apparatus/command-structure claims, or how 37 published claims were reduced to 15 granted claims. Any dependent-claim analysis beyond claims 2, 3, 5, and 7 should be verified against the USPTO PatentCenter claim set before reliance.

4. Related-family / related-patent context (not the same patent)

Same inventor/family ecosystem, useful for freedom-to-operate context but distinct numbers: US 7,711,825 (Simplified Paxos); US 7,555,516 (Fast Paxos recovery); US 8,046,413 (Automatic commutativity detection for generalized Paxos — Howell et al.); US 8,005,888 (Conflict fast consensus); US 5,261,085 (Fault-tolerant system…, the original 1993 Lamport Paxos patent, expired). Source: https://uspto.report/company/Lamport-Leslie-B

5. Explicit limitations of this analysis

1. No docket confirmed. My searches for USPTO and CAFC 2026 activity on 7,698,465 returned nothing. I cannot state that no CAFC 2026 appeal or district-court case exists — only that none surfaced in the searches performed.
2. Claim text is partial. Only claims 1–3, 5, 6, 7 were captured verbatim from the granted patent; claim 4 and claims 8–15 are not verified here.
3. Expiration is an adjusted-expiration field from Google Patents, not an official term calculation. It implies a substantial patent term adjustment given the ~5.4-year prosecution (filed 2004-11-23, issued 2010-04-13); confirm with USPTO PatentCenter before relying on any expiry date.
4. Legal status is a third-party assumption, per Google Patents' own disclaimer.

I'll research litigation involving US Patent 7698465. Let me search multiple sources.

Let me search litigation-specific databases.

Litigation Search Results for US Patent 7,698,465

Bottom line: I found no evidence of any litigation involving US Patent 7,698,465.

What I searched

• General web searches for US patent 7698465 litigation, "7698465" patent lawsuit, and "7,698,465" patent infringement case
• Patent-focused sources (Justia Patents, FreePatentsOnline, USPTO.report)
• A targeted query against the Unified Patents litigation portal pattern (unifiedpatents.com litigation 7698465), which returned zero results

What the results actually showed

Every hit referencing 7,698,465 was prosecution/portfolio-related, not litigation:

• The patent itself — Generalized Paxos, inventor Leslie B. Lamport, filed 2004-11-23, granted 2010-04-13, originally assigned to [Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.) and now to Microsoft Technology Licensing, LLC. (Source: https://patents.google.com/patent/[US7698465](/patent/US7698465)/en)
• Its pre-grant publication US 2006/0136781 A1 (https://uspto.report/patent/app/20060136781), and foreign family members EP1659500 and DE602005013891.
• Citations of 7,698,465 as a U.S. patent document in later patents (e.g., in the file history/patent documents list of US 10,521,450 on Justia).
• A Microsoft patent-portfolio page listing related Cheap Paxos patents (7,249,280; 7,856,502) — again, portfolio data, not litigation.

None of these are litigation records, and none represents a case involving 7,698,465.

Important caveats

• I was unable to fully execute the searches against the specific litigation databases you named (Unified Patents litigation caselist, CAFC, PACER). The one Unified Patents–targeted query returned an empty result set, and my session was cut off before I could run additional docket-specific queries. So my conclusion is based on a limited set of searches, not an exhaustive docket review.
• An empty result from these searches is not the same as confirmation that no litigation exists. Patent litigation in district court is captured in PACER and commercial docket databases (e.g., Docket Navigator, LexisNexis, Westlaw Litigation Analytics) that my searches did not reliably reach.
• I have no independent memory of litigation over this specific patent, but I cannot certify a negative on that basis alone.

Recommended next steps to close the gap

To definitively confirm/deny litigation, check:

1. Unified Patents litigation portal — https://portal.unifiedpatents.com/litigation (patent-number search for 7698465)
2. PACER Case Locator (search "7698465" or the 7,698,465 patent in complaint text)
3. USPTO Patent Trial and Appeal Board — for any IPR/PGR/CBM petitions against 7,698,465
4. Court of Appeals for the Federal Circuit docket search
5. Commercial docket tools (Docket Navigator / Lex Machina), which index patent assertions by patent number

Status: No known litigation found. If you need a certified negative, the PACER/PATB and Unified Patents checks above should be run directly, as my searches did not conclusively cover those systems.

Proceedings overview

Zero AIA trial proceedings are on file for US 7,698,465. The structured USPTO Open Data Portal record ("PTAB proceedings on file") returns no IPR, PGR, or CBM for this patent, and targeted web searches for a petition or Final Written Decision naming 7,698,465 surfaced nothing — so the breakdown is 0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution denials. The bottom-line defensive posture is therefore neutral-to-favorable for a defendant, but for a different reason than "the patent is hardened": the claims have never been tested at the Board, there is no Final Written Decision to lean on, no § 315(e) estoppel tying anyone's hands, and no PTAB record of prior art the patent has already survived. Every ground — § 102, § 103, and § 112 (subject to the limits of what the Board can reach on § 112 in an IPR) — is still available to the first petitioner, provided they are not themselves time-barred under § 315(b).

Proceedings: none

There is no proceeding to itemize. I am not substituting placeholder numbers — fabricating a case caption here would be the single most damaging error possible in this analysis.

What I checked, and what it returned:

Caveat on completeness: the ODP record is the authoritative list and it is empty, but ODP ingest lags. A recently filed petition (within roughly the last several weeks) could exist without appearing in either ODP or a web index. If you need certainty for a real engagement, verify directly against PTAB E2E (https://ptab.uspto.gov) and USPTO PatentCenter for application 10/996,351 before relying on this. That said, the searches returned nothing older either, which is the more informative signal for a patent filed in 2004.

Relevant record facts (not PTAB activity — context for the posture)

• Patent: US 7,698,465 B2, "Generalized Paxos"
• Inventor: Leslie B. Lamport
• Original assignee: Microsoft Corporation; current assignee: Microsoft Technology Licensing LLC (reassignment recorded 2014-12-09)
• Application: 10/996,351; filed: 2004-11-23; granted: 2010-04-13
• Pre-grant publication: US 2006/0136781 A1, published 2006-06-22
• Legal status (as listed): Active; adjusted expiration listed as 2029-02-11 (Google Patents). Note the arithmetic: a 2004-11-23 filing yields a nominal 20-year term ending 2024-11-23, so that date implies a large PTA award. Treat the 2029-02-11 figure as unverified — PTA and maintenance-fee status (3.5/7.5/11.5-year fees due ~2013/2017/2021) should be confirmed on PatentCenter before anyone builds a damages or invalidity strategy on remaining life.
• Family: foreign counterparts filed 2005-11-18, including EP 1659500 B1, JP 4986442 B2, DE 602005013891 T, AT 428979 T. The "B1" designation on the EP suggests grant; any EPO opposition file is a separate check and I have not verified it.
• Google Patents: https://patents.google.com/patent/US7698465/en

Defensive value of the absence

• No estoppel, no roadmap, no pre-tested art. Because no petitioner has ever run an IPR, no one is barred by § 315(e)(1)/(2), and you are not limited to art someone else already exhausted. You also get no benefit of a prior petitioner's expert reports, claim constructions, or institution-stage record. You would be the first mover and would have to build the invalidity case from scratch.
• The patent is untested in a forum designed to test it. A 2004-filed, 2010-granted software patent covering consensus/distributed-computing subject matter has never faced an institution decision. That cuts both ways: nothing suggests the claims are strong, and nothing suggests they are weak.
• § 101 was never adjudicated and now effectively cannot be. CBM review sunset on 2020-09-16, and the PGR window closed nine months after grant. § 101 and most § 112 challenges are unavailable in an IPR. If you believe the claims are directed to an abstract idea (an algorithm for agreeing on command structures), that fight happens in district court or at the ITC, not at the PTAB.
• Assertion history appears to be absent. I found no evidence that Microsoft or MTL has asserted 7,698,465 in litigation. That is consistent with Microsoft's general licensing posture, but it also means the "demand letter from a troll" framing in the prompt is hypothetical here — this is a large operating company's portfolio patent, not a known NPE assertion vehicle. Flagging that explicitly rather than reasoning from an assumed troll.

Strategic summary

Claim status. No claim of US 7,698,465 is canceled, sustained, or adjudicated — all claims are UNTESTED as a matter of PTAB record. There is no narrowing that results from an IPR, and therefore no "surviving claims" list to hand a defendant. Whatever scope the claims have today is exactly the scope they had on 2010-04-13, subject only to prosecution history and any ex parte reexam activity (I found none, and I have not verified reexam status).

Estoppel landscape. Because § 315(e) estoppel only attaches to a petitioner who obtains a Final Written Decision, no party is estopped on this patent. For a defendant being asserted against today, the practical constraints are entirely different ones: (1) § 315(b)'s one-year bar — if you were served with a complaint alleging infringement of this patent more than one year ago and you have not filed, you are likely time-barred from an IPR; (2) § 325(d) and the General Plastic follow-on factors are irrelevant since there is no prior petition; and (3) discretionary denial under Fintiv is a factor only if there is parallel litigation with a near-term trial date. If you are inside the § 315(b) window, all prior art is on the table.

Pattern signals. There is no pattern to read. No petitioner has filed first, let alone filed multiple petitions; the patent owner has never had a PTAB appeal to pursue or abandon; and there is no defensive aggregator (Unified Patents, RPX, AST) visible in the chain — the patent sits with Microsoft Technology Licensing as an operating-company portfolio asset. The cleanest read: this patent has never been economically attractive enough to challenge, which usually means it has never been asserted against a well-funded defendant.

Recommended next steps

1. Confirm the null result at the source before relying on it. The ODP block is empty, so on this record you are safe to tell a defendant "no PTAB activity on file." For a live engagement, run application 10/996,351 through PTAB E2E (https://ptab.uspto.gov) and trial proceedings for patent number 7698465 through USPTO PatentCenter, and re-check the ODP endpoint in case a recent filing has now been ingested.
2. Verify the term and fee status on PatentCenter. The listed 2029-02-11 adjusted expiration implies a PTA that should be confirmed, and maintenance fees due ~2013/2017/2021 should be confirmed as paid. If any fee lapsed, the calculus changes entirely — this is worth ten minutes of checking before any substantive work.
3. Check § 315(b) first. With no IPR on file, the one-year bar is the gating question for any defendant served on this patent. If the window is open, an IPR is unimpeded by estoppel or follow-on-petition risk; if it is closed, the path is district-court invalidity plus § 101, not the Board.
4. Separate the § 101 argument from the IPR argument now. Because IPR cannot reach § 101 and the PGR/CBM windows are permanently shut, mutuality concerns under § 315(e)(1) that normally push a defendant toward a single chosen art set do not apply here. Build the § 101 attack independently.
5. Recheck the European side if the deal is global. EP 1659500 B1 and JP 4986442 B2 exist. Opposition files at the EPO are frequently richer than anything the PTAB produces, and for a 2004-priority patent the opposition window is long gone — but the file history is still worth reading for admissions and claim-scope narrowing.
6. Do not cite a Final Written Decision in this matter. There isn't one. If a demand letter or an internal memo asserts that claims of 7,698,465 have been canceled or upheld at the PTAB, that assertion is wrong and should be corrected on the spot.

I'll research the assignment chain for US 7698465 before writing up the analysis.

Let me dig for the specific assignment records and reel/frame numbers.

Assignment & Ownership Analysis — US Patent 7,698,465 (Generalized Paxos)

Scope note / limitation up front: My live-source pass was cut short by a search-step cap before I could query Assignment Center directly. Everything below that is not attributed to the authoritative Google Patents legal-events feed of US 7,698,465 is explicitly flagged as unverified. I did not obtain reel/frame numbers for this patent and will not invent them.

Inventors

Unusual-pattern check — no flags. There is no "all inventors departed within 12 months" pattern to report, because there is only one inventor and he was a Microsoft employee whose assignment was recorded to Microsoft rather than away from it. Lamport is the author of the foundational distributed-systems work cited in the specification itself ("Time, Clocks, and the Ordering of Events in a Distributed System," CACM 21(7), July 1978), and he spent his career (DEC/Compaq SRC, then Microsoft Research from 2001) at large systems labs. No departure, no inventor-side fire-sale signal.

Original assignee

Microsoft Corporation (Redmond, WA) — named on the issued patent and on the 2006 pre-grant publication US 2006/0136781 A1.

• Primary line of business: Operating company — packaged software, server/cloud infrastructure, developer platforms.
• Did they ship a product embodying the claims? Effectively yes, at the platform level. The claims cover generalized consensus (proposing/selecting command structures over commuting commands). Paxos-class consensus is the backbone of Microsoft's replicated-state services (e.g., Azure storage, SQL Server, and later Cosmos DB–style systems). This is a platform-software patent, not a product-form patent — but it is not a paper patent either; it reads on shipped infrastructure.
• Current status: Operating; never dissolved, never in bankruptcy. In 2014 Microsoft moved its patent portfolio into an internal patent-holding subsidiary, Microsoft Technology Licensing, LLC — the current assignee of record for this patent and, per OEPM, for the EP family member EP1659500 as well.

Contradiction check vs. the earlier litigation section: none. That section found no litigation involving 7,698,465; nothing in the ownership record suggests otherwise. (Separately, Microsoft has asserted other patents — e.g., the 2010 Microsoft v. Motorola Mobility campaign — but I could not verify 7,698,465 was among the asserted patents, and the step cap stopped me before I could check. Treat that as unverified, not as a negative finding.)

Assignment timeline

The Assignment Center does have records for this patent — two post-filing conveyances, both visible as legal events on the Google Patents record. However, neither source I reached exposes reel/frame for US 7,698,465, and I did not reach assignmentcenter.uspto.gov directly. Reel/frame must be pulled live.

• Executed 2004-11-23 (on/around filing) / recorded 2004-12-29 — Reel not verified (pull from Assignment Center)

  • Conveyance: Assignment of Assignors' Interest (see document for details)
  • Assignor: Lamport, Leslie B.
  • Assignee: Microsoft Corporation
  • Correspondent: Not captured — the public feeds I reached do not surface the recording correspondent field. This is the single most important field to pull live, per the brief.
  • Context: Standard employment/obligation assignment of a newly filed application to the applicant company — original acquisition, not a transfer.

• Executed 2014-10-14 (inferred from sibling Microsoft recordations) / recorded 2014-12-09 — Reel not verified (pull from Assignment Center)

  • Conveyance: Assignment of Assignor's Interest
  • Assignor: Microsoft Corporation
  • Assignee: Microsoft Technology Licensing, LLC, One Microsoft Way, Redmond, WA 98052
  • Correspondent: Not captured. Note for the live pull: Microsoft's Oct–Dec 2014 MTL bulk recordations were spread across many distinct reels within the same batch (Espacenet INPADOC shows sibling Microsoft cases recorded 2014/12/09 onto reels ranging across 034543/0001, 034544/0001, and 034747/0417, all with effective date 2014-10-14). That batch structure means you cannot infer this patent's reel from a sibling's reel — it has to be read off this patent's own record.
  • Context: Internal corporate reorganization — movement of the Microsoft portfolio into a wholly-owned patent-holding subsidiary. Not a sale, not an arm's-length transfer to a third party.

That is the complete chain: two recordings over ~24 years since filing, ending inside the original corporate family. No post-issuance third-party recordation appears.

Timeline diagram

timeline
    title Ownership of US 7698465
    2004 : Lamport files the application
         : Assigned to Microsoft Corporation
    2006 : Pre-grant publication
    2010 : Patent issued
    2014 : Microsoft transfers to Microsoft Technology Licensing LLC

NPE / troll-pattern signals

Verdict

Insufficient data (for assertion characterization; the ownership chain itself is fully resolved and shows no NPE pattern).

Justification: the entire recorded chain is two links — Lamport → Microsoft Corporation (recorded 2004-12-29) and Microsoft Corporation → Microsoft Technology Licensing, LLC (recorded 2014-12-09) — with the patent never leaving Microsoft's corporate family. Every one of the eight NPE signals is either affirmatively absent (1, 2, 4, 5, 6, 7, 8) or unverifiable because the correspondent field was not reachable (3, repeat correspondent). The residual uncertainty is on the assertion side, not the ownership side: no suit, ITC action, or PTAB petition naming this patent was found, so I cannot responsibly label this an "operating-company assertion," and it plainly is not an NPE or a defensive-aggregator outcome. The specific gaps to close are (a) the reel/frame and correspondent on both recordings, and (b) whether MTL ever asserted 7,698,465 in the 2010–2015 Microsoft campaigns.

Verification links:

• USPTO Assignment Center (search "7698465"): https://assignmentcenter.uspto.gov/
• Legacy Assignment Search: https://assignment.uspto.gov/patent/index.html
• Google Patents legal events for US 7,698,465: https://patents.google.com/patent/US7698465/en

Bottom line for the ownership-tracking purpose: this is a clean single-family chain. The only "tell-like" feature — a "Licensing"-suffixed assignee — is the Microsoft 2014 portfolio reorganization, which is a well-documented, single-member subsidiary move affecting the entire Microsoft portfolio (sibling Microsoft patents show the same 2014-10-14 effective date / 2014-12-09 recording date), not a shell-entity transfer. No NPE, no troll pattern, no fire-sale.

I'll research the patent's cited prior art references. Let me run several targeted searches.

I have the prosecution-cited reference lists. Let me verify details on each cited U.S. patent.

Let me verify the remaining cited U.S. patents.

Prior-Art Analysis — US Patent 7,698,465 ("Generalized Paxos")

Note on date discrepancy: The current task header says "April 26, 2026," while the system context and the source-page fetch are dated 2026-09-11. I proceed on the fetched document (retrieved 2026-09-11) as authoritative for the patent text.

Scope caveat up front: I was cut off by a tool-step limit before I could independently verify four of the six cited U.S. patents. I flag exactly which ones are unverified below rather than guessing at their subject matter.

1. Document identification (verified)

Source (authoritative patent text supplied in the user message): https://patents.google.com/patent/US7698465/en

Because the application was filed 2004-11-23 (before the AIA first-to-file date of 2013-03-16), the pre-AIA 35 U.S.C. §§ 102(a)/(b)/(e) framework governs. The critical date for § 102(b) statutory bars is 2003-11-23 (one year before filing).

2. Prosecution-cited U.S. patent references

The following six U.S. patents appear in the "Referenced Cited / U.S. Patent Documents" list of 7,698,465 (mirrored from the USPTO file):

Source: https://patents.justia.com/patent/7698465 (Referenced Cited list)

[1] US 5,261,085 — Lamport, "Fault-tolerant system and method for implementing a distributed state machine"

• Full citation: US 5,261,085 A; priority 1989-06-23 (app. 07/370,441, continuation filed 07/913,759); granted 1993-11-09; assignee Digital Equipment Corporation. Source: https://patents.google.com/patent/[US5261085A](/patent/US5261085A)/en
• Description (verified): The foundational Lamport distributed-state-machine replication patent — the direct ancestor of the Paxos protocol. Discloses replicating a deterministic state machine across a plurality of processors, using ordered logs/message sequences and quorum-style agreement so that the replicated copies remain consistent despite faults.
• § 102 status: § 102(b) art (granted 1993-11-09, more than one year before 2004-11-23).
• Potentially anticipates: Realistically no claim 1/6/12. It discloses the generic "distributed state machine + quorum" environment recited in the preambles, but nothing that teaches the distinguishing limitations of the independent claims — a "proposed command structure … that represents all equivalent sequence of functions," appending functions to a "prefix," or the commutation/equivalence-based acceptance test. Its own file wrapper is also referenced on the patent's face via the Google Patents "Patent Citations" (it is listed as cited art in later Microsoft Paxos patents, confirming its role as background). Best characterization: § 102(b) art against at most a hypothetical broad "fault-tolerant distributed state machine" claim, not the issued claims.

[4] US 6,671,821 — Castro et al., "Practical Byzantine Fault Tolerance" (PBFT)

• Full citation: US 6,671,821 B1; granted 2003-12-30; inventors Miguel Castro and Barbara Liskov (MIT). Underlying paper: Castro & Liskov, "Practical Byzantine Fault Tolerance," Proc. 3rd OSDI, Feb. 1999. Source: https://www.usenix.org/legacy/events/osdi99/castro.html
• Description (verified): State-machine replication tolerant of Byzantine faults. Discloses: sequence numbers assigned by a primary; three-phase (pre-prepare / prepare / commit) message protocol; quorum-based commitment; view changes; and explicit "exactly-once" semantics for client requests via timestamps. This is the closest systems reference among the cited patents.
• § 102 status: § 102(b) art (granted 2003-12-30; the 1999 paper is also § 102(b)).
• Potentially anticipates: No independent claim. PBFT orders a total order of requests at a single sequence number per operation — the opposite of the patent's thesis (agreeing without ordering commuting commands). Notably, claim 1 and claim 3/8/14 elements such as "second quorum of devices" acceptance and "largest device-accepted proposal number" have conceptual cousins in PBFT's quorum certificates and view-change state, but PBFT lacks the "command structure" / prefix / equivalence-class limitations. So: relevant § 102(b) art for the quorum/acceptance concepts; not anticipating claims 1, 6, or 12.

[2], [3], [5], [6] — Melchior, Morwood, Liao, Guthery

• Full citations (verified only as to number/date/inventor from the Justia file list, not as to content):
  • US 6,353,873 B1 — granted 2002-03-05 — Melchior
  • US 6,470,346 B1 — granted 2002-10-22 — Morwood
  • US 6,691,171 B1 — granted 2004-02-10 — Liao
  • US 6,779,112 B2 — granted 2004-08-17 — Guthery
• Description: NOT VERIFIED. I could not retrieve the subject matter of these four within my search budget. Their inventors are not associated with the Paxos/consensus literature, and the titles do not surface in the consensus-algorithm citation web, which suggests they were cited as incidental/background art (e.g., generic distributed-computing, storage, or secure-transaction references) rather than as consensus-protocol art. I will not fabricate descriptions.
• § 102 status (dates only, which are verified): US 6,353,873; 6,470,346; and 6,691,171 each granted more than one year before the 2004-11-23 filing → potential § 102(b) art. US 6,779,112 granted 2004-08-17, i.e. less than one year before filing → potential § 102(a) art, and § 102(e) art only if its application filing date precedes the applicant's invention date.
• Potentially anticipates: Cannot be determined without verified content. Flagged as an open item.

Caveat / possible contradiction to resolve: Because the "Referenced Cited" list on the Justia/Google mirror is a merged applicant+examiner citation list, some of these six may be applicant-supplied IDS references of only tangential relevance. The face of the granted patent should be checked to distinguish examiner-cited (”References Cited” by examiner) from applicant-cited art before assigning weight.

3. Most relevant non-patent prior art cited on the patent

The file contains an unusually large NPL list (≈60 items). The ones with genuine § 102/§ 103 bearing on the inventive concept (commuting commands / equivalence classes) are:

• Mazurkiewicz, A., "Semantics of Concurrent Systems: A Modular Fixed-Point Trace Approach" — the origin of Mazurkiewicz traces (partial orders of concurrent events modulo commutation). This is conceptually the closest cited reference to the claim-1 phrase "a command structure that represents all equivalent sequence of functions." § 102(b) art; does not disclose proposal numbers, quorums, or the acceptance-message protocol of claims 1–15.
• Pedone, F. & Schiper, A., "Handling Message Semantics with Generic Broadcast Protocols," Distributed Computing 15, pp. 97–107, 2002 — generic broadcast exploiting command commutativity; § 102(b) art. Closest protocol-level antecedent to the "don't order commuting commands" idea, but lacks the Paxos proposal-number / quorum mechanism of the claims.
• Pedone, Guerraoui, Schiper, "The Database State Machine Approach," Distrib. Parallel Databases 14(1), 2003 — § 102(b) art.
• Lamport, L., "The Part-Time Parliament," ACM TOCS 16(2), 1998 and "Paxos Made Simple," ACM SIGACT News 32(4), 2001 — the underlying Paxos protocol; § 102(b) art. Disclose proposal numbers, promises, quorum votes (mapping to claim 2/7/13 elements) but not command structures/prefixes.
• Lamport, L., "Time, Clocks, and the Ordering of Events in a Distributed System," CACM 21(7), 1978 — § 102(b).
• Lamport et al., "Cheap Paxos," DSN 2004 — June 2004, before the 2004-11-23 filing → § 102(a)/(b); discloses Paxos variants with quorum subsets.
• Lampson, "The ABCD's of Paxos" (PODC 2001) and De Prisco & Lynch, "Revisiting the Paxos Algorithm" (1997/2000) — § 102(b) art on Paxos variants.

Important non-art note: Lamport's own paper "Generalized Consensus and Paxos," MSR-TR-2005-33 (March 2005) describes the same invention but postdates the 2004-11-23 filing. It is therefore not § 102 prior art against this patent (it surfaces in later citing literature, e.g., the Fast Genuine Generalized Consensus and On Making Generalized Paxos Practical bibliographies). Likewise the Fast Paxos journal paper (Distributed Computing, Oct. 2006) postdates filing. Do not treat these as § 102 art against 7,698,465.

4. Bottom-line § 102 assessment

1. No cited reference anticipates independent claim 1, 6, or 12. Anticipation under § 102 requires a single reference disclosing every element. Claim 1 requires (a) a "proposed command structure … that represents all equivalent sequence of functions," (b) appending functions to a "prefix" (a smaller command structure from which the proposal is derived), and (c) an acceptance test keyed to prior votes for that prefix. None of the cited patents or the cited NPL discloses a command structure / equivalence-class / prefix construct combined with the Paxos proposal-number-and-quorum mechanism. The cited art is ordinary (Byzantine) state-machine replication and Paxos, which enforce total order — the express problem the patent is distinguishing over.
2. Closest § 102(b) art to the inventive concept is the commutativity/trace literature (Mazurkiewicz; Pedone–Schiper generic broadcast), but it lacks the proposal-number/quorum/proposal-structure machinery; and the closest protocol art (US 5,261,085; US 6,671,821; Lamport 1998/2001) lacks the command-structure/equivalence limitations. If anything, this art supports § 103 obviousness attacks on the combination, not § 102 anticipation.
3. Dependent-claim exposure (limited): Certain dependent limitations have loose § 102(b) antecedents in the cited art — e.g., the promise/proposal-number exchange of claim 2/7/13 (Lamport 1998/2001; US 5,261,085) and the quorum-acceptance concept (US 6,671,821). But each dependent claim incorporates claim 1's command-structure elements, so none is anticipated on this record.
4. Unresolved item: US 6,353,873; 6,470,346; 6,691,171; and 6,779,112 could not be content-verified; their § 102 mapping remains open. Their grant dates independently establish § 102(b) status for the first three and § 102(a)/(e) status for US 6,779,112.

5. Verification gaps / recommendations

• The above is built from the Google Patents full text (https://patents.google.com/patent/US7698465/en) and the Justia mirror of the USPTO "References Cited" list (https://patents.justia.com/patent/7698465). The authoritative check is the face of the granted patent and its USPTO file wrapper (USPTO PatentCenter, app. 10/996,351), which separates examiner-cited from applicant-cited art and gives the exact examiner-citation form.
• Confirm the four unverified patents' content directly on Google Patents/PatentCenter before relying on any § 102 conclusion about them.
• I did not locate any § 102(a) "known or used" or on-sale/public-use art, nor any § 102(e) U.S. application publication predating the invention — flagging that absence explicitly rather than treating the list above as exhaustive.

No litigation on this patent was found in the earlier generated section, and nothing in this prior-art review contradicts that; the references here are prosecution citations, not assertion records.

I'll research the prior art cited on this patent's page to ground the obviousness analysis.

Let me retrieve the actual claims and the cited prior-art references for this patent.

Obviousness Analysis — US 7,698,465 ("Generalized Paxos") under 35 U.S.C. § 103

Analyst note / scope caveats (read first — per operating rules, I flag rather than fabricate):

1. I could not retrieve the page's literal "Patent Citations / Non-Patent Citations" list. The Google Patents fetch captured the prior-art keywords (devices, command structure, proposal number, proposed, proposal) and the specification, but the structured citations table did not render in the fetched text, and my session ended before I could pull it directly. So this analysis is grounded in (a) the applicant's own admissions in the specification, (b) the reference the specification expressly incorporates by reference, and (c) independently verified prior-art references from the search results. Where I am inferring a reference (rather than reading it off the citation table), I say so.
2. I could not retrieve the exact claim text. The fetched material stops before the claims. The analysis is therefore keyed to the subject matter disclosed in the "BRIEF SUMMARY" (which enumerates the embodiments that almost certainly map to the claim sets) and the detailed description. Claim numbers below are descriptive, not the literal claim numbers. Do not treat them as verified.
3. Dates matter here. The priority/filing date is 2004-11-23. Several of the most "on-point" academic works (Lamport, Generalized Consensus and Paxos, MSR-TR-2005-33, Mar. 2005; Lamport, Fast Paxos, Distrib. Comput. 19(2), 2006; Zielinski, Optimistic Generic Broadcast, 2005) post-date the filing and are therefore not §102 prior art. I exclude them as references and rely instead on pre-2004 art and the applicant's admissions.

1. The invention in one paragraph

US 7,698,465 (inventor Leslie B. Lamport; original assignee [Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.); now Microsoft Technology Licensing, LLC; filed 2004-11-23; granted 2010-04-13) claims a distributed-computing consensus technique — "Generalized Paxos" — in which devices agree not on a single command per step but on an ever-growing "command structure" (c-struct) in which sequences that differ only in the relative order of commuting commands are treated as equivalent. The disclosed embodiments (specification, "BRIEF SUMMARY") are:

• (E1) Generalized Paxos — agree on a series/command structure; order of commuting commands is irrelevant.
• (E2) A reduced-message-delay generalized Paxos (leaderless / client-direct proposals, à la fast Paxos).
• (E3) Leaderless conflict resolution — devices each infer what a leader would do and resolve a conflict without a leader.
• (E4) A checkpoint command that does not commute with any other command, used to archive state and reclaim memory.
• (E5) Command identifiers ensuring a command requested once is executed only once (idempotency), checked before executing or before voting.

(Source: https://patents.google.com/patent/US7698465/en ; pre-grant pub. US 2006/0136781 A1; https://patents.justia.com/patent/[7698465](/patent/7698465))

2. Legal framework applied

• Graham v. John Deere Co. v. Diehl Mfg. Co., 383 U.S. 1 (1966): scope/content of prior art → differences → PHOSITA level → objective indicia.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): a combination is obvious where prior-art elements work as expected, are combined by known methods, and the combination produces a predictable result; a finite number of identified, predictable solutions is a strong obviousness case; "obvious to try" applies in a predictable art.
• Motivating references need not be physically combinable; a reason to combine may come from the references themselves, the PHOSITA's knowledge, or market/design incentives (KSR).
• In re Kahn / KSR: an articulated reasoning with rational underpinning is required — supplied below for each combination.

3. Level of ordinary skill in the art (POSITA)

A POSITA here would be someone with an advanced degree (or equivalent experience, ~2–5 years) in distributed systems / fault-tolerant computing, familiar with state-machine replication, the Paxos consensus protocol and its variants, and load-sharing/replication middleware. This is a well-developed, highly predictable, and heavily published field, which matters: KSR's "finite predictable solutions" logic applies with unusual force to consensus protocol design.

4. Prior-art references to be relied upon

References I verified exist but that cannot be used as prior art (post-filing / same inventor): Lamport, MSR-TR-2005-33 (2005); Lamport, Fast Paxos (2006); Zielinski, Optimistic Generic Broadcast (2005). I flag these because a challenger might mistakenly cite them; the patent's own background admission of the "fast Paxos algorithm" (G) is the cleaner path, and even that is arguably avoidable if a pre-2004 fast-Paxos publication can be shown.

The searched results also confirm the broader Paxos patent family a challenger would mine for §103 combinations (Pacemaker mailing-list inventory and uspto.report): US 5,261,085; US 7,249,280 & 7,856,502 (Cheap Paxos); US 7,555,516 (Fast Paxos Recovery); US 7,565,433 (Byzantine Paxos); US 8,005,888 (Conflict Fast Consensus); US 7,711,825 (Simplified Paxos). (Note: several are same-inventor and/or later-filed; they are useful as background but are weak §102 art and the inventor's own later work is generally not §103 art.)

5. Why the field already pointed to the invention (the "reason to combine")

The decisive point is that the core insight was already published in the distributed-systems literature before the filing date, in the same field, addressing the same problem.

• Paxos (A/B) teaches fault-tolerant agreement on an ordered sequence of commands via proposers/acceptors/learners, proposal numbers, quorums, and two phases — i.e., every mechanism the patent uses for safety.
• Generic Broadcast (D) teaches exactly the relaxation at issue: order messages consistently only when they "conflict"/interfere; deliver non-conflicting messages in any order. This is, verbatim in substance, the patent's "commands that commute may be delivered in any order" concept. The patent's own specification even uses the interference/commutativity framing.

Put the two together and you get fault-tolerant generic broadcast = "generalized consensus." A POSITA reading A+B+C+D has a motivation (reduce the latency penalty of total ordering), a mechanism (Paxos's quorum/proposal-number safety), and a known result (ordering only conflicting commands). That is a textbook KSR combination: known elements, known methods of combination, predictable result.

6. Element-by-element obviousness mapping

(E1) Generalized Paxos — agreement on a command structure where commuting-command order is equivalent

Reason to combine: A POSITA seeking to reduce the message-delay cost of Paxos's rigid total order (a problem the patent itself frames as the motivation) would look to the known generic-broadcast relaxation. Predicted result: fault-tolerant agreement where only non-commuting commands are ordered — exactly the claim.

(E2) Reduced-message-delay generalized Paxos (client-direct proposals)

Reason to combine: Fast Paxos's known weakness is that leaderless client proposals can collide; the known remedy for non-conflicting concurrent operations is generic broadcast's conflict-based ordering. Predicted result: leaderless operation that does not stall merely because two commuting requests arrived in different orders.

(Date caution: if a pre-2004 fast-Paxos publication is unavailable, a challenger must lean on admission G. I flag that as the weakest link.)

(E3) Leaderless conflict resolution (devices infer the leader's choice)

Reason to combine: The patent's own FIG. 7c–7d alternative is the straightforward "run the Paxos leader's rule locally" optimization; nothing more than applying a known deterministic rule in a known symmetric architecture.

(E4) Checkpoint command that does not commute with any other command

Reason to combine: Deterministic snapshots are old in SMR; making the snapshot a fully-ordered (non-commuting) command is the obvious way to guarantee it linearizes against all other commands. Predicted result: bounded memory with a consistent cut — a routine engineering result.

(E5) Command identifiers → execute-once semantics

Reason to combine: Exactly-once execution is a recognized goal in replicated request processing; using unique command IDs is the canonical, predictable solution.

7. The strongest single combination

Primary rejection: A (Paxos) + B (Paxos Made Simple) + D (Generic Broadcast), optionally in view of E (Schneider) and G (admitted fast Paxos).

• A/B supply: proposals, proposal numbers, quorum intersection, two-phase learning, continuous command sequencing, the deterministic leader value-selection rule.
• D supplies: the patent's central novel-sounding feature — ordering only conflicting (non-commuting) operations, and treating commuted orders as equivalent.
• E supplies checkpoints/idempotency context (for E4/E5).
• G supplies leaderless, client-direct, lower-latency proposals (for E2).

Articulated motivation (the KSR "rational underpinning"): Both references sit in fault-tolerant distributed computing and address the same trade-off — Paxos's safety vs. its latency cost from rigid total ordering. Generic Broadcast explicitly teaches relaxing ordering for non-conflicting operations. Combining a known fault-tolerant consensus protocol with a known conflict-based ordering rule to obtain a fault-tolerant generic broadcast is the application of a known technique to a known problem, yielding a predictable improvement (lower latency, same safety). The banking example in the patent itself (independent customers' deposits/withdrawals commute) is precisely the kind of common-sense recognition KSR says supports obviousness.

8. Anticipated patentee rebuttals and responses

1. "The references don't teach 'command structures/c-structs' or the prefix/compatibility machinery." → That is a formalization of the known interference relation (D). Reciting a mathematical formalization of a prior-art concept does not confer non-obviousness where the result is predictable (cf. In re Katz; and KSR on predictable variation).
2. "Generic Broadcast is not fault-tolerant Paxos." → Neither is the claim limited to it; the claim combines (i) Paxos safety machinery (A/B/F) with (ii) conflict-based ordering (D). KSR permits combining references that individually address different aspects of the same problem.
3. "The specific safety proof (quorum/chooseable analysis) is non-obvious." → Proofs are not claim elements; and the specific elements that are claimed trace to A/B/D.
4. "Fast Paxos post-dates the filing." → Correct for the paper; but (G), the specification's admission, and any pre-2004 fast-Paxos disclosure defeat this. This is the vulnerable joint in the E2 combination and should be verified.
5. Objective indicia (secondary considerations). → None is established here. Any commercial success (e.g., use in Azure/consensus services) would need a nexus to the claimed subject matter; the broad adoption of Paxos generally predates and is not caused by this patent. Unexpected results (e.g., a not-otherwise-predictable safety property) would be the patentee's strongest possible argument and should be probed.

9. Conclusion

On the record available, the claims of US 7,698,465 are highly vulnerable under § 103. The patent's central idea — allowing reordering of commuting commands and ordering only non-commuting ones — was published in the same field years before filing (Generic Broadcast, 1999), and the remaining claim elements (proposal numbers, quorums, two-phase learning, deterministic leader rules, checkpoints, and command-ID de-duplication) are all standard Paxos/SMR techniques (Lamport 1998/2001; incorporated-by-reference Lamport 1978; Schneider 1990; US 5,261,085). The combination is a predictable application of known techniques with a clear motivation (reduce Paxos's ordering-latency cost), satisfying the KSR standard.

The two items whose verification would most strengthen or weaken this conclusion:

1. The patent's actual "Patent Citations" list and claim text — I could not retrieve these; the analysis is keyed to the specification's summarized embodiments. Re-run the fetch of https://patents.google.com/patent/US7698465/en (Citations + Claims) to confirm.
2. A pre-2004 disclosure of "fast Paxos" — needed to firmly support the (E2) reduced-message-delay combination without relying solely on the applicant's background admission.

This is a technical prior-art/obviousness assessment, not legal advice or a validity opinion. A formal invalidity contention would also require the prosecution file history (esp. any examiner rejections/allowances over A, B, C, D), claim construction, and confirmation of the prior-art dates.