US 5226141

I'll search for information on this patent number and check for any litigation activity.

Let me check specifically for any Federal Circuit or district court litigation involving this patent.

Search Notes / Caveats

• I searched for US 5,226,141 specifically. No results were returned tying that patent to any CAFC (Federal Circuit) 2026 docket, nor to any active district-court infringement litigation. I found no evidence of current enforceability or pending appeal — see status below.
• Important disambiguation: several search hits for the bare string "5226141" refer to U.S. Trademark Registration No. 5,226,141 (the mark "NORTH," Reg. date 06/20/2017, registrant Askov Holdings, LLC, Class 25 apparel; subject of TTAB Opposition No. 91240810). That is a trademark, not the patent, and has no relationship to US 5,226,141. I did not auto-merge them.
• Everything below is grounded in the authoritative full text of US 5,226,141 provided to me. Where I lack authoritative information (e.g., litigation history not visible in my sources), I say so explicitly.

---

Bibliographic Summary — US 5,226,141 A

Assignment chain (per record): Touch Technologies, Inc. → security interest to San Diego National Bank (1998, released 2000) → Computer Associates International, Inc. (2001) → CA, Inc. (2014 merger). An assignment from inventor Esbensen to Touch Technologies (effective 1989-07-14) was also recorded in 2005.

Status caveat: the patent is recorded as expired for failure to maintain / end of term, so it is not currently enforceable. I have no authoritative evidence of any 2026 CAFC activity involving this number.

---

Abstract (as issued)

"A variable length cache system keeps track of the amount of available space on an output device. The capacity of the cache system is continuously increased so long as it is less than the available output space on the output unit. Once the size of the cache system exceeds the available output space on the output unit, which is less than the total space available on the output unit by a predetermined amount, the contents of the cache memory are flushed or written to the output device and the size of the cache memory is reduced to zero."

---

Plain-Language Overview of the Independent Claims

The invention addresses a tape-drive bottleneck: a write cache can only be as large as the "past-end-of-tape" margin historically allowed, because if the cache held more data than could be written before the physical end of tape, data would be lost. The patent dynamically grows the cache based on how much room is actually left on the tape, and flushes it just before the tape runs out (using a nominal capacity deliberately set below the real physical capacity as a safety margin).

Claim 1 — Apparatus (core concept).
An apparatus that resizes a data cache feeding an output device. It (a) computes an "available storage capacity parameter" from the device's nominal maximum capacity (which is deliberately less than the device's actual capacity); (b) compares the cache's current contents against that parameter; (c) lets the cache grow if there is still room; and (d) flushes the cache to the device if the cached data exceeds the remaining room.

Claim 5 — Method (mirror of claim 1).
The same four steps (determine available-capacity parameter; compare cache contents to it; increase cache if room remains; transfer cache to output device if cache exceeds the parameter), recited as a method.

Claim 9 — Cache memory system (sum-based variant).
A dynamically adjustable cache system for an output device whose effective/nominal maximum is less than its actual capacity. It: tracks how much data was already transferred to the device; forms a sum of that prior amount plus the cache's preset capacity; compares the device's maximum capacity to that sum and emits a signal when capacity still exceeds the sum; and increases cache capacity in response to that signal.

Claim 19 — Method of operating a cache system (sum-based variant).
Method counterpart of claim 9: provide a cache; determine prior transferred amount; sum it with present cache capacity; compare the device's nominal maximum capacity (a chosen amount below actual maximum) against the sum; generate a signal if capacity still exceeds the sum; and grow the cache when that signal is present.

Claim 26 — Computer-system data cache (remaining-capacity variant).
In a computer writing through a cache to a finite-size storage medium, the cache includes: means for determining the remaining unwritten storage capacity of the medium, and means for increasing cache storage capacity as blocks of data pass through, growing the cache up to what the remaining medium capacity can hold.

Claim 29 — Improved caching method (remaining-capacity method variant).
Method counterpart of claim 26: determine the medium's remaining storage capacity and grow the cache in response to data blocks passing through, up to the amount the remaining medium capacity can store.

Claim 32 — Data cache (addressable-space variant).
A cache buffering data between an originating means and a finite-capacity storage means, comprising an addressable memory holding data in transit, plus means for increasing the addressable space of that memory as it accepts data, in accordance with the availability of storage capacity of the destination.

Claim 34 — Data-buffering method (addressable-space method variant).
Method counterpart of claim 32: hold data in an addressable memory intermediary and increase the addressable space as data is accepted, according to the destination's available capacity.

Dependent claims worth noting

• Virtual memory implementations: claims 2, 6, 11, 16, 21, 22 add virtual-memory management (allocating pages from a free-page list to grow the cache arbitrarily, potentially beyond main memory).
• Magnetic tape / serial media: claims 14, 15, 27, 28, 30, 31 tie the output device to a linearly movable medium / magnetic tape and reference the end-of-tape marker, computing cumulative inches of tape (data ÷ density, plus inter-record gaps) versus inches remaining.
• Reducing/clearing the cache after flush: claims 17–18, 23–25 (claim 25 reduces cache capacity to zero).
• Literal anomaly: claim 15 recites "A cache memory as in claim 16…", i.e., a forward (out-of-order) dependency on a later claim. Read literally as written; it appears to be a drafting error in the issued patent, but I am treating the number as-is per your instructions.

---

Supporting Disclosure Details

• Two algorithms are disclosed: (1) a physical write-head position (total bytes written ÷ density + inter-record gaps); (2) a logical write-head position (cached bytes ÷ density + gaps, added to physical position). The patent states these predict positions within ~2% of actual for conventional tape drives.
• Safety proximity: caching stops when the logical head position comes within roughly 150 feet of end-of-tape for a 2400-foot tape (~5% of the tape), after which operation reverts to conventional device-driver writes.
• Prior art cited (7 references) includes US 4,258,418 ("Variable capacity data buffer system," IBM), US 4,315,312 ("Cache memory having a variable data block size," NCR), US 4,449,580? (no — US 4,445,170, Zilog), US 4,499,539 (IBM), US 4,524,416 (Honeywell), US 4,868,737 (Tandberg Data), and US 4,947,319 ("Arbitral dynamic cache using processor storage," IBM).
• Embodiments: FIG. 4 = virtual-memory-based dynamic cache manager in host; FIG. 5 = cache manager inside the tape unit with manually settable density/length switches; FIG. 6 = standalone variable-capacity cache unit between computer and drive. A program module named "TATHROTTLE" is referenced as attached to the specification for performing the calculations.

---

Bottom Line

US 5,226,141 is a 1993 Esbensen/Touch Technologies patent (now CA, Inc.) on a variable-capacity write cache for tape drives, whose central idea — size the cache to the remaining writable space on the medium rather than a fixed cap, and flush before exceeding a nominal (safety-discounted) capacity — is captured across eight independent claims in apparatus, system, and method form. It is expired as of ~2010, and my searches surfaced no CAFC 2026 docket or active litigation concerning it. The only prominent "5226141" match in current dockets is an unrelated trademark registration of the same number, which I have flagged rather than conflated.

I'll search for litigation records involving US 5226141.

Let me run more targeted searches on litigation databases.

Let me check litigation-specific databases directly.

Findings: Litigation Involving US Patent 5,226,141

Conclusion: I found no evidence of any patent litigation involving US 5,226,141. I searched multiple times across patent-litigation-oriented queries and general web sources and could not identify a single district court case, ITC Section 337 investigation, CAFC appeal, or PTAB proceeding in which US 5,226,141 was asserted, challenged, or otherwise at issue.

What I checked and what came back

Note: I was cut off by a tool-call limit before I could run two further queries, so my PACER/Unified Patents coverage is based on general web results rather than a direct docket lookup.

⚠️ Important collision warning — do not conflate these

Because you instructed literal interpretation of identifiers, I want to flag that the alphanumeric string 5226141 appears in several unrelated contexts, none of which is this patent:

• US Trademark Registration No. 5226141 — "NORTH," registered 2017-06-20 to Askov Holdings, LLC, Class 25 (clothing). It was the subject of a TTAB cancellation counterclaim in Askov Holdings, LLC v. Our Life 2.5 LLC, Opposition No. 91240810. This is a trademark registration, not the patent.
• NSN/part number 5226141 — a U.S. Army Tank Automotive Command thrust-washer bearing (NSN 3120-00-350-3004).
• CourtListener investment record ID 5226141; a Finnish telephone number; etc.

None of these relate to US 5,226,141.

Contextual facts from the patent record (relevant to why litigation is unlikely/expired)

• Patent: US 5,2266141 — "Variable capacity cache memory"; inventor Daniel M. Esbensen; application 07/794,689 filed 1991-11-18 as a continuation of Ser. No. 07/380,629 (filed 1989-07-14, now abandoned); granted 1993-07-06.
• Ownership chain (from the assignment records in the patent text): Original assignee Touch Technologies, Inc. → security interest to San Diego National Bank (1998, released 2000) → Computer Associates International, Inc. (recorded 2001-01-12, effective 2000-10-26) → CA, Inc. (merger recorded 2014-01-22). A 2005 assignment from Esbensen to Touch Technologies was recorded with an effective date of 1989-07-14.
• Legal status: Expired – Lifetime, with an anticipated expiration date of 2010-07-06 (17 years from grant). Maintenance fees were paid through the 12-year fee (2005), with a late-payment surcharge.
• International family: PCT publication WO1991001525A1, "Variable capacity cache memory" (same title/inventor family).

Because the patent expired in July 2010, it could not be asserted in a new infringement action today. Any historic cases would have had to be filed during the ~1993–2010 enforcement window, close to the 1990s/2000s era when Computer Associates held it.

Bottom line

Based on the sources searched, there is no known litigation involving US Patent 5,226,141. I could not confirm this against PACER, the CAFC docket, or the Unified Patents litigation portal directly due to the tool-call limit, so I would characterize this as "no litigation found in the available sources" rather than a categorical negative. If you need a definitive clearance, a direct PACER/RECAP docket search on the patent number (and a review of the full assignment chain at CA, Inc.) would be the appropriate next step — note that pre-2001 records for Touch Technologies, Inc. and the Esbensen-entity filings may not be fully digitized.

I'll verify the structured "no proceedings" signal against public sources before finalizing.

The first searches hit a coincidental trademark registration number, not the patent. Let me run more targeted checks on the patent itself and any assertion history.

Proceedings overview

Total AIA trial proceedings on US 5,226,141: 0 — no IPRs, no PGRs, no CBMs are on file. The USPTO Open Data Portal returns an empty proceeding set for this patent, and my independent web checks (Google Patents, PTAB-related search surfaces, litigation reporting) surfaced no petition, no institution decision, no FWD, and no Federal Circuit appeal — so the defensive posture is the strongest possible in one respect and the weakest in another: there is no IPR estoppel to lock a defendant into, and no claim has ever been tested at the Board. The real story for a defendant here is not "the patent is hardened" — it's that the patent expired on 2010-07-06 and is off-term, which is a far bigger defensive fact than any PTAB outcome would be.

⚠️ Disambiguation warning (do not conflate): search results for "the '141 patent" frequently refer to US 7,260,141 (Lionra Technologies / Apple, IPR2023-00796). That is a different patent, a different owner, and a different technology (baseband beamforming). It has nothing to do with US 5,226,141. Separately, "Registration No. 5226141" in TTAB Opposition 91240810 is a trademark registration (the mark "NORTH," Askov Holdings), not this patent. Neither should be cited as PTAB activity on this patent.

---

No proceedings to itemize

Because the canonical proceedings list is empty, I am not fabricating proceeding numbers, panels, or dispositions. What I can confirm from the authoritative patent text and the structured record:

Why no AIA trial was ever likely. PGR was categorically unavailable (effective filing date 1989 — PGR reaches only patents with effective filing dates on or after 2013-03-16). CBM was limited to covered business-method patents and sunsets 2020-09-16; a tape-drive cache-management patent was never a viable CBM candidate. That leaves IPR — and § 315(b)'s one-year bar from service of an infringement complaint means any IPR window closed long ago, on top of the patent going off-term in 2010.

Procedural events that do exist (from the file, not PTAB): security interest to San Diego National Bank (recorded 1998-07-02, released 2000-08-25); assignment to Computer Associates International (2001-01-12); assignment from Esbensen to Touch Technologies (2005-07-22); CA, Inc. merger (2014-01-22); maintenance fees paid at years 4, 8, and 12 (final fee 2005-06-01).

---

Strategic summary

Claim status: 35 of 35 claims are UNTESTED at the Board — but all 35 are EXPIRED. No claim of US 5,226,141 has been canceled, confirmed, or even construed in an AIA trial. Claims 1–35 (apparatus claims 1–4, 9–18, 26–28, 32–33; method claims 5–8, 19–25, 29–31, 34–35) remain as issued and, per the file, are not the subject of any certificate of cancellation or reexamination certificate. The controlling fact for a defendant is not claim scope but term: the patent's 17-years-from-grant term ran from 1993-07-06 to 2010-07-06, and the 20-years-from-earliest-filing alternative (1989-07-14 → 2009-07-14) is shorter. Google Patents' legal-status record confirms "Expired – Lifetime / Anticipated expiration 2010-07-06." An expired patent cannot be infringed, and § 286's six-year damages lookback for any pre-expiration conduct closed in 2016. A demand letter asserting this patent today is asserting a dead property.

Estoppel landscape: there is none, and that cuts in the defendant's favor. Because no IPR was ever instituted against any claim, § 315(e)(2) estoppel never attached to anyone. No petitioner, real party in interest, or privy is barred from raising any § 102 or § 103 ground. If you are nonetheless facing assertion (e.g., in a contract, indemnity, or warranty dispute over historical conduct), you are free to run invalidity on the full prior-art universe in district court or ITC, and you may seek ex parte reexamination at the USPTO — which, unlike IPR, carries no petitioner-estoppel and no statutory 1-year bar and remains available even for expired patents because reexam addresses validity of the claims as issued. The seven references cited on the face of the patent are the natural starting kit: US 4,258,418 (IBM, "Variable capacity data buffer system"), US 4,865,737 (Tandberg Data, "capacity of a related storage medium is checked when buffering data"), US 4,494,539 (IBM, limiting allocated data-storage space), US 4,315,312 (NCR, variable data block size cache), US 4,945,319 (IBM, arbitral dynamic cache using processor storage), US 4,445,170 (Zilog), and US 4,524,416 (Honeywell). US 4,258,418 and US 4,865,737 in particular read directly on the "size the cache to the remaining media capacity" concept of independent claim 5/19/26/29/32.

Pattern signals: none of the usual PTAB tells are present. There is no serial petitioner (no party filed even one IPR, let alone multiple). There is no PTAB-appeal history — no FWD, so nothing for the patent owner to have appealed. There is no defensive aggregator (no Unified Patents, RPX, or similar) in the chain; the assignment chain is purely corporate (Touch Technologies → Computer Associates → CA, Inc.) plus a 1990s bank security interest. The absence of any AIA challenge is itself the signal: patents that are actively asserted in high-stakes litigation reliably attract IPRs, and this one never did — consistent with a patent that reached end-of-term in 2010, before the AIA trial regime even came online.

---

Recommended next steps

1. If you received a demand letter citing US 5,226,141, lead with expiration, not invalidity. The patent expired 2010-07-06 (17 years from the 1993-07-06 grant; recorded as "Anticipated expiration" in the file and status "Expired – Lifetime"). No infringement is possible for post-2010-07-06 conduct, and any damages on pre-expiration conduct are time-barred by 35 U.S.C. § 286 (six-year lookback, closed 2016). Ask the sender to identify the accused acts and their dates; a well-pled letter cannot, and that is usually the end of it.

2. Preserve the invalidity record anyway. If the dispute is really about historical conduct (indemnity, escrow, M&A reps and warranties), you are not estopped by anything — no IPR was instituted, so § 315(e)(2) is a non-issue. Ground your invalidity position on the seven face-cited references above, prioritizing US 4,258,418 and US 4,865,737, and consider a § 302 ex parte reexamination request, which has no petitioner-estoppel and no § 315(b) timing bar.

3. No trial-stage milestones to track. There is no institution decision deadline, no oral hearing, and no FWD due date, because there is no proceeding. The PTAB's statutory one-year trial clock (35 U.S.C. § 316(a)(11)) is not running against anyone. The only date that matters is 2010-07-06.

4. Verify the emptiness independently before you rely on it. My check of the ODP structured record plus public sources found zero AIA trials, and the patent's 2010 expiration makes new filings commercially pointless — but the ODP ingest can lag. Before filing anything that asserts "no PTAB activity," re-confirm on USPTO PTAB E2E / Patent Trial and Appeal Board End-to-End and the USPTO Patent Center for US 5,226,141. If a filing does surface, treat it as a post-term anomaly and re-run this analysis.

Bottom line: this is not a "hardened by surviving IPR" patent — it is an expired 1993 patent that no one ever bothered to challenge at the Board because it was already off-term. For a defendant, that is the best possible posture: no live infringement exposure, no current claim construction to fight, no estoppel constraining your invalidity theories, and a documented expiration date you can put in the first paragraph of your response.

I'll verify the chain against external sources before writing up the analysis.

Source note: reel/frame and conveyance data below are taken from the Google Patents legal-events record for US 5,226,141 (https://patents.google.com/patent/US5226141/en), which reproduces the USPTO assignment records. Correspondent-of-record names are not exposed in that feed and I could not retrieve the Assignment Center record pages in this session — see the data-gap note under Signal 3. Nothing below is inferred beyond what the recorded events state.

Inventors

Pattern notes:

• No co-inventors and no inventor departures to trace — a single-inventor patent, so the "all inventors left within 12 months" fire-sale tell cannot be assessed.
• Chain-of-title anomaly (notable): the inventor→assignee assignment (Reel 016290/0753) was not recorded until 2005-07-22, ~16 years after filing and ~4.5 years after Touch Technologies had already conveyed the patent to Computer Associates International, Inc. (Reel 011457/0549, effective 2000-10-26). The 2005 filing carries an effective date backdated to the 1989 filing date, which is the signature of a nunc pro tunc / confirmatory assignment filed to cure a missing link in the recorded chain rather than a new economic transfer. The same 2005 window contains a late maintenance-fee payment with surcharge (Year 11, paid 2005-06-01) — the record shows a title-and-annuity hygiene exercise in 2005.

Original assignee

Touch Technologies, Inc., a California corporation (Sunnyvale, CA), named as original assignee on the issued patent and in the USPTO reassignment records (Assignors: Touch Technologies, Inc.; Reel 011457/0549 context).

• Primary line of business: not determinable from the patent record. The disclosure is squarely a tape I/O write-caching subsystem for virtual-memory minicomputer operating systems (the specification explicitly cites "VAX-type virtual memory computer systems marketed by the digital equipment corporation"), which indicates a small storage-management software vendor. I could not confirm from any source I retrieved a shipped product embodying the claims — mark that as unverified, not as "no product."
• Financial profile inferred from recorded events: the company granted a security interest over its IP to San Diego National Bank in 1998 (Reel 009289/0845, effective 1998-04-03) and had it released in 2000 (Reel 011084/0796, effective 2000-08-25) — the profile of a small private company borrowing against its patent portfolio. Two months after the release it sold the patent out to Computer Associates.
• Current status: not determinable; Touch Technologies has held no interest in this patent since the October 2000 conveyance. No bankruptcy, dissolution, or change-of-name filing appears in this patent's record.

Assignment timeline

Chronological list of every recorded assignment in the record:

• 1989-07-14 (executed) / recorded 2005-07-22 — Reel 016290/0753

  • Conveyance: Assignment of Assignors Interest (recorded nunc pro tunc; effective date stated as the 1989 filing date)
  • Assignor: Daniel M. Esbensen
  • Assignee: Touch Technologies (California)
  • Correspondent: not exposed in the retrievable record (see Signal 3 data gap)
  • Context: chain-of-title curative filing — confirms inventor→company ownership 16 years late, four years after the patent had already been sold onward.

• 1998-04-03 (executed) / recorded 1998-07-02 — Reel 009289/0845

  • Conveyance: Security Interest (secured financing, not an ownership transfer)
  • Assignor: Touch Technologies, Inc.
  • Assignee: San Diego National Bank (California)
  • Correspondent: not exposed in the retrievable record
  • Context: securitization — Touch Technologies pledges its IP as collateral to a local bank.

• 2000-08-25 (executed) / recorded 2000-09-07 — Reel 011084/0796

  • Conveyance: Release of Security Interest
  • Assignor: San Diego National Bank
  • Assignee: Touch Technologies, Inc. (California)
  • Correspondent: not exposed in the retrievable record
  • Context: lien satisfaction — collateral released six weeks before the patent is sold.

• 2000-10-26 (executed) / recorded 2001-01-12 — Reel 011457/0549

  • Conveyance: Assignment of Assignors Interest
  • Assignor: Touch Technologies, Inc.
  • Assignee: Computer Associates International, Inc. (New York)
  • Correspondent: not exposed in the retrievable record
  • Context: acquisition — small-vendor IP asset sale into a large public software company (CA). This is the only arm's-length economic transfer in the chain.

• 2012-03-27 (executed) / recorded 2014-01-22 — Reel 032113/0267

  • Conveyance: Merger
  • Assignor: Computer Associates Think, Inc.
  • Assignee: CA, Inc. (New York)
  • Correspondent: not exposed in the retrievable record
  • Context: internal corporate reorg — merger vesting title in CA, Inc. Recorded ~3.7 years after the patent's anticipated expiration of 2010-07-06, i.e., pure housekeeping on an expired asset.

Link not evidenced in this record: the 2001 assignment names Computer Associates International, Inc. as assignee, but the 2014 merger names Computer Associates Think, Inc. as assignor. The intermediate link (CA International → CA Think, or a name change) is not reflected in this patent's legal-events list. That is an observation about the record, not an assertion that the link is missing.

Timeline diagram

timeline
    title Ownership of US 5226141
    1989 : Filed by Touch Technologies Inc
         : Daniel Esbensen named sole inventor
    1993 : Patent issued
    1998 : San Diego National Bank lien recorded
    2000 : Bank lien released
         : Assigned to Computer Associates Intl
    2005 : Confirmatory inventor assignment recorded
    2010 : Patent term expires
    2014 : Merger into CA Inc recorded

NPE / troll-pattern signals

1. Shell-entity transfer — NOT PRESENT. Every assignee in the chain is either an operating corporation (Touch Technologies, Inc.; Computer Associates International, Inc.; CA, Inc.) or a chartered bank acting as secured lender (San Diego National Bank). There is no "IP / Patents / Licensing / Holdings / Ventures" assignee, no registered-agent address, and no single-purpose LLC anywhere in Reels 009289/0845, 011084/0796, 011457/0549, 016290/0753, or 032113/0267.

2. Known asserter in the chain — NOT PRESENT. No assignee matches Acacia/ACT, Marathon, Intellectual Ventures, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, or any Spangenberg entity. CA, Inc. is a large operating software vendor; if CA appears in asserter directories it would be as an operating-company plaintiff, not a PAE. Note that this patent's term is the controlling fact: anticipated expiration 2010-07-06 means no one in this chain has been able to assert it for the last ~16 years.

3. Repeat correspondent across the chain — UNKNOWN / NOT ASSESSABLE. This is the one signal I cannot close out, and I will not guess at attorney names. The Google Patents legal-events feed carries assignor, assignee, conveyance, execution/recording dates and reel/frame, but not the correspondent of record. Closing this item requires pulling the five records in the USPTO Assignment Center by patent number (https://assignmentcenter.uspto.gov/ or https://assignment.uspto.gov/patent/index.html) and reading the correspondent field on Reel 009289/0845, 011084/0796, 011457/0549, 016290/0753, and 032113/0267. There is no observable pattern of a single recording agent in what is retrievable, but absence of a name in a feed that omits names is not evidence.

4. Cascading transfers — NOT PRESENT. Five recorded events spread over 1998 → 2000 → 2001 → 2005 → 2014, i.e. ~16 years and an average gap of four years. There is no <24-month run of chained LLCs, and the assignees share no common naming or address pattern (bank, California software vendor, New York software companies).

5. Pre-litigation transfer — NOT PRESENT / NOT APPLICABLE. I found no infringement litigation naming US 5,226,141. The final recorded conveyance (executed 2012-03-27, recorded 2014-01-22, Reel 032113/0267) postdates the patent's anticipated expiration (2010-07-06), so it could not have been arranged to enable assertion.

6. Bankruptcy fire-sale — NOT PRESENT. No Chapter 7/11 conveyance appears in the record. Reel 009289/0845 is an ordinary secured-financing pledge, and it was released at Reel 011084/0796 (effective 2000-08-25) rather than foreclosed. The sequence — pledge 1998, release Aug 2000, sale to CA Oct 2000 — is consistent with a small vendor monetizing IP while solvent enough to satisfy its lender, but the record does not establish distress. No CA bankruptcy: CA, Inc. was acquired by Broadcom in 2018.

7. Privateering — NOT PRESENT. No operating-company → NPE transfer and no NPE assertion on CA's behalf. The chain runs operating company → operating company, and never touches a licensing vehicle.

8. Defensive aggregator — NOT PRESENT. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. It terminates at CA, Inc. (Reel 032113/0267), and the patent is expired.

Verdict

Operating-company assertion — with the express caveat that the "assertion" prong is unproven for this specific patent; read this as the "not an NPE" bucket.

Justification: the ownership chain is entirely operating entities and a secured lender — Reel 011457/0549 (Touch Technologies, Inc. → Computer Associates International, Inc., effective 2000-10-26) and Reel 032113/0267 (merger vesting title in CA, Inc., effective 2012-03-27) — with the only other entries being a bank pledge (Reel 009289/0845) and its release (Reel 011084/0796). Zero of the eight NPE signals are present; the sole unresolved item is the correspondent-of-record field for the five reels, which the retrievable feed does not carry. Because the patent's anticipated expiration was 2010-07-06 and the final conveyance was recorded in 2014 on an already-expired asset, no assertion of US 5,226,141 by the current owner (CA, Inc.) is possible, and none was found.

Verify in USPTO Assignment Center: https://assignmentcenter.uspto.gov/ (mirror: https://assignment.uspto.gov/patent/index.html) — search patent number 5226141.

I'll verify the patent and its cited references via web search, then analyze each citation.

Prior-Art Analysis — US 5,226,141 A

Note: I've interpreted all identifiers literally as given. I used the authoritative full text supplied in your message for the patent's own citations, and web search to characterize each reference. I do not have access to the USPTO file history / examiner's stated §102-§103 grounds, so the anticipation mapping below is my own analysis, flagged for confidence, not a restatement of a record I could not fetch.

---

1. The patent being examined

Independent claims: 1 (apparatus), 5 (method), 9 (cache system), 19 (method), 26 (system), 29 (method), 32 (data cache), 34 (method). Claims 2–4, 6–8, 10–18, 20–25, 27–28, 30–31, 33, 35 are dependent.

Core inventive concept: the cache capacity is grown block-by-block so long as it stays below the remaining writable capacity of the output medium (a nominal maximum set a safety amount below actual capacity), and is flushed when the cached data would exceed that remaining space. The cache can therefore expand to nearly the whole medium (the tape), rather than being capped at the past-end-of-tape region.

Source: https://patents.google.com/patent/US5226141/en

---

2. The seven examiner-cited references

These are the references listed under "Patent Citations (7)" in the patent. I address each in turn.

(a) US 4,258,418 A

• Full citation: US 4,258,418 A — "Variable capacity data buffer system," International Business Machines Corporation. Filed 1978-12-28; published/granted 1981-03-24.
• Description: Buffers data between a processor and a serially-reusable, over-runnable I/O device (magnetic tape, floppy disc, teleprocessing). The data storage device has a maximum capacity value; logic establishes a threshold capacity value less than the maximum and keeps the stored data at that threshold while transferring. It explicitly targets minimizing I/O device start/stop delays — the same problem US 5,226,141 addresses.
• Potential §102 relevance: Claim 1 and claim 5 (a "maximum storage capacity… less than an actual storage capacity of the output device" reads onto the threshold-below-maximum teaching). Also conceptually relevant to claims 9 and 19.
• Confidence / caveat: Low for strict §102. US 4,258,418 maintains a fixed threshold and does not determine the remaining capacity of the output medium, nor does it grow the cache in dependence on that remaining capacity. It is better characterized as a §103 reference for the "threshold less than maximum" limitation. I could not fully verify the reference's own claim text beyond the abstract/summary returned in search.

(b) US 4,315,312 A

• Full citation: US 4,315,312 A — "Cache memory having a variable data block size," NCR Corporation (inventor Schmidt). Filed 1979-12-19; published/granted 1982-02-09.
• Description: A cache in a processor whose data-block size (number of words per block) is variable via a block-size register and inhibit logic on the index address, used to improve the hit ratio for different task types.
• Potential §102 relevance: Claims 1 and 5 broadly (a cache with dynamically variable storage/Block characteristics). Also claim 4 (sequentially addressable memory locations / addressable blocks).
• Confidence / caveat: Low for §102. The variability here is block size/granularity to optimize hit ratio — not cache capacity driven by remaining space on an output device. Distinct purpose; unlikely to anticipate the "available output capacity" limitations. Treat as §103/background.

(c) US 4,445,170 A

• Full citation: US 4,445,170 A — "Computer segmented memory management technique wherein two expandable memory portions are contained within a single segment," Zilog, Inc. Filed 1981-03-19; published/granted 1984-04-24.
• Description: Segmented memory management in which two expandable memory portions share a single segment (per the title as recorded in the supplied text). Directed to memory segmentation/expansion, not to media-space-aware caching.
• Potential §102 relevance: Most plausibly cited for claims 2, 6, 13, 16 (expandable/contiguous storage allocation) — i.e., the general mechanism of growing addressable storage.
• Confidence / caveat: Low and unverified. My web-search budget was exhausted before I could retrieve this reference's full text, so I am characterizing it only from the title recorded in the patent. I am explicitly not fabricating its claim content. Recommend direct retrieval before relying on it.

(d) US 4,499,539 A

• Full citation: US 4,499,539 A — "Method and apparatus for limiting allocated data-storage space in a data-storage unit," International Business Machines Corporation. Filed 1982-12-20; published/granted 1985-02-12.
• Description: Limiting the space allocated within a data-storage unit.
• Potential §102 relevance: Claim 9 and claim 19 (determining a storage limit / comparing to it and acting on the comparison), and possibly claim 1.
• Confidence / caveat: Low–moderate. The title indicates limitation of allocated space, which is adjacent to — but not the same as — growing the cache up to the remaining media capacity. The "comparing maximum storage capacity to the sum, generating a signal" mechanics of claims 9/19 are the most likely target. Full text not retrieved in this session.

(e) US 4,524,416 A

• Full citation: US 4,524,416 A — "Stack mechanism with the ability to dynamically alter the size of a stack in a data processing system," Honeywell Information Systems Inc. Filed 1980-04-15; published/granted 1985-06-18.
• Description: A stack mechanism whose size can be dynamically altered in a data-processing system.
• Potential §102 relevance: Claims 1, 2, 5, 6 (means for dynamically increasing storage capacity / adding storage space when a predetermined value is reached).
• Confidence / caveat: Low for §102. Dynamic size alteration of a stack is a generic capacity-change teaching; it lacks the output-medium-availability determination that is central to the independent claims. Best treated as §103 support for the "increase capacity" element only. Full text not retrieved.

(f) US 4,868,737 A — the closest reference

• Full citation: US 4,868,737 A — "Method and arrangement in which capacity of a related storage medium is checked when buffering data," Tandberg Data A/S (inventor Pahr, Per Olaf). Filed 1986-12-22; published/granted 1989-09-19.
• Description: A buffer memory for intermediate storage of data to be written to a storage medium (explicitly including a magnetic tape drive, as well as floppy/disk/optical). The currently available (logical) storage capacity of the buffer is modified in dependence on the available storage capacity of the storage medium: it equals the buffer's physical capacity while the medium has more space, and is dynamically reduced (linearly or in steps) to match the medium's remaining space as the medium fills — so that all buffered data can always be recorded. It even measures tape position by counting drive/stepping-motor pulses and accounts for BOT / LP / EW (early warning) / EOT hole markers, and notes the medium may be a magnetic tape with an end-of-tape (EOT) indicator.
• Why it matters: This is the most factually on-point reference — same problem (tape start/stop), same domain (tape buffering), and the same core mathematical comparison (buffer capacity vs. remaining medium capacity).
• Potential §102 relevance: Claims 9, 19, 26, 29, 32, 34 (determining remaining storage capacity of the finite medium and sizing an intermediate/cache memory accordingly), and claim 28/31's tape/EOT limitation. Strongest candidate for a §102 rejection on those independent claims.
• Confidence / caveat: Moderate–high on subject matter; medium on strict anticipation. A key textual difference: in Tandberg the buffer's physical capacity P is fixed and the logical/available capacity is only reduced as the medium fills — it never grows beyond P. US 5,226,141 claims (e.g., claim 1's "means for enabling the current capacity of the data in the cache memory to increase," and the Summary's "size of the cache can automatically increase… almost the entire capacity of the medium") require increasing capacity up to the medium's remaining space. That growth limitation is a likely distinguishing point. So Tandberg is a very strong §103 combination reference and a plausible §102 reference for the method/limit-comparison independent claims, but probably not a clean §102 anticipation of claim 1 as a whole.
• Sources: https://patents.google.com/patent/[US4868737A](/patent/US4868737A)/en

(g) US 4,947,319 A

• Full citation: US 4,947,319 A — "Arbitral dynamic cache using processor storage," International Business Machines Corporation. Filed 1988-09-15; published/granted 1990-08-07.
• Description: A data cache in a computer operating system that dynamically adapts its size in response to competing demands for processor storage. An "arbiter," entered cyclically, samples users' wait states and decides whether to increase or decrease the cache ceiling by g pages; pages are taken from IBM expanded storage (a pool of 4096-byte pages) via the Expanded Storage Manager / Page Migration Program. Moving averages, trend tests, and a max-size test are used. The patent notes the same technique can manage primary memory with competing virtual-memory and data-cache demands.
• Potential §102 relevance: Claims 2, 6, 16, 22 (virtual-memory/expanded-storage means, allocating additional storage units/pages to increase cache capacity), and the "increase capacity" mechanics of claims 1, 5, 9, 19. Directly on point for the virtual-memory embodiment of FIGS. 4/7.
• Confidence / caveat: Moderate for the virtual-memory/dependent claims; low for the independent capacity-vs-output-space claims. The trigger in US 4,947,319 is memory contention (I/O-wait vs. page-fault wait), not the remaining printable/writable capacity of the output device — which is the heart of US 5,226,141. So it is a strong §103 combination reference (dynamic growing cache implemented with allocatable pages) rather than a standalone §102 anticipation.
• Sources: https://patents.google.com/patent/US4947319 ; summary from http://patentimages.storage.googleapis.com/fb/71/e1/31a63981d6c052/US4947319.pdf

---

3. Summary mapping (claimed anticipation / strongest relevance)

---

4. Candid limitations of this analysis

• These are the examiner's citations, not necessarily the examiner's §102 grounds. A reference appearing in the "References Cited" list can have been cited for §103, as a background/technical reference, or under a different statutory basis. The supplied Google Patents text does not include the office-action reasoning, so I cannot state which specific rejection (if any) each reference supported.
• I could not retrieve the full text of US 4,445,170, US 4,499,539, and US 4,524,416 within the tool-call budget; my descriptions of those three rest on the titles recorded in the patent and general knowledge, and I have flagged the relevant claim mappings as unverified. I am not asserting their claim content as fact.
• "Potentially anticipates" ≠ "invalidates." Strict §102 requires a single reference to disclose every element of the claim as arranged. On the record available, the strongest single-reference candidate is US 4,868,737 (Tandberg Data), but even it appears to reduce an available buffer capacity rather than grow a cache up to the medium's remaining space — leaving the "increase up to the amount storable in the determined remaining capacity" limitation (claims 1, 26, 29, 32, 34) as the likely point of novelty over it.
• The WO 1991001525 A1 publication (listed under "Similar Documents") is the international counterpart of this same invention, not third-party prior art against it.

If you want, I can re-run targeted searches to pull the full claim text of US 4,445,170 / 4,499,539 / 4,524,416 so the marginal references can be assessed at the same depth as Tandberg and the IBM references.

I'll verify the content of each prior-art reference from the page's "Citations" (examiner-cited) section before building the § 103 combinations. Note the "Cited By" entries mostly post-date the 1989 priority date, so they are not prior art — I'll flag that.

Obviousness Analysis — US 5,226,141 under 35 U.S.C. § 103

Builds on the prior sections (status/claim overview already given). This section adds the § 103 work product and does not re-describe the claims' general structure.

---

0. Scope, dates, and evidentiary limits (read first)

Governing law. US 5,226,141's effective filing date is 1989-07-14 (parent app. 07/380,629; the 1991-11-18 filing is a straight continuation). It is therefore a pre-AIA patent, and § 103(a) as construed in Graham v. John Deere and KSR Int'l v. Teleflex governs. Because "obvious to try" and the flexible KSR rationales are in play, the absence of an explicit teaching-suggestion-motivation in a reference is not dispositive.

Evidentiary limits I must disclose. I retrieved and read full text/abstracts for US 4,868,737, US 4,947,319, and US 4,315,312. For US 4,258,418, US 4,499,539, US 4,445,170, and US 4,524,416 I obtained only front-page/grant data and titles from the page's Citations table — my characterizations of those four are drawn from their titles and issue data, not from column-line readings, and I flag them as lower-confidence. Where a limitation turns only on those four, I say so.

Date note (contradiction flag): the session header says "Today's date: 2026-09-30" while the task text says "Current Date: April 26, 2026." This does not affect the analysis, but I am flagging the inconsistency rather than silently picking one.

---

1. Which references in the "Prior Art" section are legally available

Two critical legal points:

1. The two § 102(e) references survive § 103(c). Pre-AIA § 103(c) disqualifies § 102(e) art only if it and the application were commonly owned at the time the invention was made. Tandberg Data A/S and IBM are unrelated to Touch Technologies, Inc. — so 4,868,737 and 4,947,319 remain available despite issuing after 1989-07-14. This is exactly why the examiner cited them.
2. Everything in the page's "Cited By" tables is not prior art. US 5,340,653, 5,586,303, 5,772,410, 5,995,982, 7,478,218, and the Tivo/Centripetal families all post-date 1989-07-14 by years. Likewise WO 1991/001525 A1 in "Similar Documents" is the PCT sibling of the same Esbensen/Touch Technologies invention and is not prior art to itself. Any analysis that treats the "Cited By" list as prior art is legally wrong.

---

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

Weighing the In re GPAC factors: the field is 1980s computer operating systems and storage subsystems; the problem (I/O bottleneck at a serial storage device) and its solutions (queued I/O + data cache + device driver) were well-developed by 1986–1989. A POSITA here is a bachelor's-level CS/EE engineer (or equivalent) with ~2–4 years' experience in OS/device-driver development, or a master's with ~1–2 years, and would be familiar with: (a) tape drives, end-of-tape reflectors and past-EOT margins; (b) I/O request queues and write-behind caches; (c) tape data density, block sizes and inter-record gaps; and (d) virtual memory page allocation. This level is important because every arithmetic step the patent relies on (bytes ÷ density + inter-record gaps) is, on the patent's own admission, computed from "fixed and known factors."

---

3. Isolating the Δ over the art

The patent's own Background (§ "Description of the Preferred Embodiment") admits the FIG. 2 system — an application I/O request "intercepted before it can pass to the device driver," placed in an "I/O request queue 12," with data "moved to a data cache 14" — as known prior art. That admission removes from the inventive delta: caching tape writes at all, the queue, the interception, the device driver interface, and the flush-to-tape mechanism. What remains is only:

• Δ1 — sizing/growing the cache as a function of the remaining writable capacity of the medium (rather than a fixed cap);
• Δ2 — using a nominal maximum that is deliberately less than actual capacity (the 150-ft-of-2,400-ft margin);
• Δ3 — growth mechanism (arbitrary storage, contiguous or via virtual-memory pages);
• Δ4 — the trivial arithmetic that converts bytes to inches via density + IRG length.

Δ1 and Δ2 are the only substantive deltas. Δ3 and Δ4 are the easy ones.

---

4. Ground I — The primary combination: US 4,258,418 + US 4,868,737

4a. What each reference supplies

US 4,868,737 (Tandberg) — the key secondary reference. This is the closest art, and it is very close. Its abstract/specification (verified from the Google Patents text) states:

"the currently available storage capacity of the buffer memory is equal to its physical storage capacity as long as the available storage capacity of the storage medium is greater than this physical storage capacity of the buffer memory. The currently available storage capacity of the buffer memory is dynamically diminished when the available storage capacity of the storage medium is less than the physical storage capacity … Thus, all data stored in the buffer memory are always storable on the storage medium."

It further discloses:

• Magnetic tape as the storage medium, with holes BOT, LP, EW (early warning), and EOT — an explicit margin marker (EW) before the physical end (EOT), which is precisely the structural analogue of the claimed "maximum storage capacity being less than an actual storage capacity.";
• a detector DE that determines the available storage capacity A of the medium, by "counting the clock pulses output by a drive motor, or … measuring the time during which the magnetic tape is moved," and by "measuring the length of the magnetic tape already recorded … since the full length of the magnetic tape is known" — i.e., the same byte-count/density logic the '141 patent uses;
• a control unit CU that compares buffer contents against the (medium-derived) logical capacity and blocks further write-in unless the data in the buffer is below that capacity;
• a signal S1 emitted by the data store to the buffer memory to change its capacity.

That maps to Δ1 (capacity determined from remaining medium capacity), Δ2 (EW margin), and the comparison/transfer steps. Its one shortfall is direction: in Tandberg the buffer's physical capacity P is fixed and only its logical capacity diminishes.

US 4,258,418 (IBM) — the primary reference for Δ3/direction. Its title, Variable capacity data buffer system, discloses a data buffer whose capacity is variable in a data handling system. It is the reference that removes Tandberg's "fixed physical capacity P" constraint and supplies the "means for enabling the current capacity … to increase" element of claims 1/5 and the "means for increasing the storage capacity" of claims 9/19/26/29 and the "means for increasing the addressable space" of claims 32/34. (Low-confidence caveat: I did not read 4,258,418's specification; the mapping rests on its title and the examiner's citation of it.)

4b. Claim 1 element chart (Ground I, with 4,499,539 for the margin)

Claims 5 (method mirror), 9 and 19 (sum-and-signal variants) map the same way, with the "sum of prior transferred amount + present cache capacity" limitation supplied by Tandberg's A = (total length − recorded length) computation — a comparison of the same two quantities, merely algebraically rearranged — and the "signal" limitation supplied by Tandberg's S1 signal or by the ordinary output of a comparator. Claims 26, 29, 32, 34 drop the nominal-vs-actual limitation entirely and are covered by 4,258,418 + 4,868,737 alone.

---

5. Motivation to combine (the KSR rationales)

The combination is motivated on at least six independent grounds, any one of which suffices:

1. Same field / analogous art. Both references are data-buffering apparatus interposed between a host and a storage device — the identical field, and the identical problem statement (the patent's own "traffic jam"/bottleneck rationale).
2. Same problem, complementary solutions. 4,258,418 solves buffer sizing; 4,868,737 solves buffer-vs-medium capacity matching. Combining a variable-capacity buffer with a medium-capacity tracker is the self-evident union of two solutions to one problem. KSR: "a known technique to improve similar devices in the same way."
3. The functional relationship in Tandberg is already capacity-dependent. Tandberg states L is determined "dependent on the available storage capacity of the storage medium," and that the modification "can occur linearly or in steps, and in the simplest case, in one step." A POSITA reading L = min(P, A) has already been given the functional link; once the buffer is made variable (418), letting L track A upward as well is the direct reading of the reference, not a new invention.
4. Design incentive / market pressure. The patent's own stated goal — "the net result would be more concurrency … which in turn speeds up the whole system" — is a classic design need. Both references are optimized for the same throughput objective, so there is a positive incentive to combine.
5. Predictable result / finite solution set. Given "how much room is left on the medium," the only three options for cache size are grow, hold, or shrink. KSR holds a finite set of predictable, known-option solutions obvious to try.
6. Direction is a predictable variation. KSR expressly: "a given course of action often has no difference other than direction." Tandberg's diminution and the '141's growth are the same equation with the inequality flipped; that is a variation in direction, not in principle. The result — no more data in the buffer than the medium can hold — is identical in both.

Two further, narrower rationales: (i) the choice of storage mechanism (contiguous RAM, an auxiliary panel, or virtual-memory pages taken from a free-page list) is a routine implementation choice once a variable-size cache is desired (see Ground III); and (ii) the 150-ft-in-2,400-ft margin is optimization of a result-effective parameter — the patent concedes industry media standards "provide more than enough excess writable media beyond the end-of-tape reflector" — which is within ordinary skill under In re Aller / In re Applied Materials, and is corroborated by Tandberg's own EW marker.

---

6. Secondary grounds for the dependent claims

Ground II — add US 4,947,319 (IBM) → claims 2, 6, 11, 16, 21, 22 (virtual-memory growth).
I verified this reference's text: a data cache that "dynamically adapts its size," with an arbiter that cyclically decides whether to increase the "data cache ceiling", and a cache manager that can release pages to an "available pool." It is page-addressable (4,096-byte pages) and expressly "exploits either or both of real or expanded storage." Its stated objects are that the cache be "not static in size" and "dynamically adapt its size." That is the entire subject matter of claims 2/6/11/16/21/22 and of the FIG. 4 embodiment. Motivation: to make the write cache exceed main memory — the exact benefit the '141 specification touts ("arbitrarily large and far exceed the size of the main memory"). Secondary rationale: KSR's "obvious to try" — pulling pages from a VM free list is one of a small number of known ways to grow a buffer.

Ground III — add US 4,445,170 (Zilog) and/or US 4,524,416 (Honeywell) → claims 3, 4, 10, 12, 13, 20.
4,445,170 discloses "two expandable memory portions" within a segment — supplying "additional storage … linkable to said cache memory by a data transfer link" (claims 3, 10, 12) and the linkable-block framing of claims 10/20. 4,524,416 discloses a stack mechanism "with the ability to dynamically alter the size" of the stack — supplying both a dynamically resizable store and the sequentially/contiguously addressable structure of claim 4 and the "solid state memory means contiguously addressable with said cache memory" of claim 13. Motivation: expanding a segment/stack on demand is the standard technique for growing a store; KSR "known technique."

Ground IV — add US 4,315,312 (NCR) → the "block/file"-granularity aspects.
Verified text: a cache data buffer whose block size is variable, set by a block size register, with an inhibit circuit and fetch counters controlling words transferred per block. It supports the "increase … by an amount corresponding to the next received block or file" mechanism and the adjustable-cache-geometry concept. I grade this the weakest of the set: it varies block size, not capacity, so it is best used as corroboration rather than as a primary reference.

Ground V — flush-and-reset claims 17, 18, 23, 24, 25.
Tandberg supplies the reduction step ("logical, possible storage capacity … can be reduced linearly or in steps … in one step"), and 4,947,319 supplies reclamation to a free pool. Reducing the cache to zero after a completed flush is conventional buffer reinitialization — a routine design choice with no unexpected result (In re Harza / In re Kuhle-type reasoning). Claim 25 is essentially unpatentable as a bare post-flush reset.

Ground VI — tape-specific claims 14, 15, 27, 28, 30, 31.
Tandberg supplies "magnetic tape," BOT/LP/EW/EOT markers, drive-motor pulse counting, and computing recorded length from a known total length. The remaining arithmetic (bytes ÷ bytes-per-inch + Σ inter-record-gap inches) is the patent's own EQU1 and is admitted to be derived from "fixed and known factors." Under In re Gulack, a limitation reciting a mathematical calculation with no additional physical consequence adds no patentable weight; alternatively, computing a head position from byte count is an obvious engineering step for a POSITA. Claims 28/31's "inches remaining before the end-of-tape marker" is directly taught by Tandberg's EW/EOT arrangement.

---

7. The one real obstacle — Tandberg's "inverse direction" — and why it fails as a teach-away

The patent owner's best § 103 argument is that 4,868,737 affirmatively teaches a fixed physical buffer that is only reduced: "[t]he currently available storage capacity of the buffer memory is dynamically diminished when the available storage capacity of the storage medium is less than the physical storage capacity of the buffer memory." Read strictly, L = min(P, A) never exceeds the hardware.

Three answers:

• (a) It is not a teaching away. A reference's disclosure of a single finite-capacity hardware embodiment is not a statement that growth is inoperative or disfavored. Under KSR, "the mere disclosure of a single embodiment does not teach away." Nothing in Tandberg says a larger buffer is undesirable; its stated goal (never overflow the medium) is equally served by a growing buffer.
• (b) The two-directional function is already disclosed. Tandberg's L is a function of A. L = min(P, A) rises with A until it saturates at P. Remove the fixed P (which 4,258,418 and 4,947,319 both do) and the same disclosed functional relationship yields growth. The '141 claim is, on these facts, a substitution of variables rather than a new mechanism.
• (c) The result is identical. Tandberg's guarantee — "all data stored in the buffer memory are always storable on the storage medium" and "no more data than can be recorded on the storage medium are stored on the buffer memory at any time" — is verbatim the functional requirement of independent claims 26, 29, 32 and 34. Where the reference already achieves the claimed result by a capacity-matching rule, changing the sign of one inequality to reach the same result by the same rule is, at most, an obvious variation.

I would expect this to be the contested battleground in any IPR, with the outcome turning on the Board's willingness to read Tandberg's L(A) relationship as bidirectional once P is made variable.

---

8. Claim-by-claim assessment

Bottom-line grade: I would rate Ground I as making out a prima facie case of obviousness for every independent claim, strongest for 26/29/32/34 and most contestable for 1/5/9/19 (because of the "nominal below actual" limitation, which leans on 4,499,539 — the reference I could characterize only from its title).

---

9. Secondary considerations (Graham factor 4)

I found no evidence in the available record of: commercial success, copying, industry praise, unexpected results, or a long-felt-but-unsolved need. Specifically:

• The patent asserts throughput/concurrency benefits, but assertions in a specification are argument, not Graham evidence, and the patent's own Background concedes the fixed-length-cache FIG. 2 system was already deployed — i.e., the need was being addressed in the art.
• A "long-felt need" theory is undercut because Tandberg filed on the very capacity-matching problem in December 1986, over 2½ years before the '141 priority date, from a different country and field of use. That is powerful evidence the problem and its solution were known to POSITAs.
• No litigation nexus is available: as established earlier, the patent expired ~2010-07-06 with no litigation found, so there is no infringement-driven evidence of value.

---

10. Auxiliary (non-§ 103) invalidity observations

Not § 103, but relevant to any invalidity posture and worth flagging because the earlier section already noted the anomaly:

• Claim 15's forward dependency. "A cache memory as in claim 16" depends on a later claim. Read literally (as instructed), this is a defective dependency that a court or the Board would likely hold indefinite under § 112(b), and — if it renders the claim's scope indeterminate — potentially not amenable to a § 103 analysis at all until construed.
• § 112(f) breadth. Claims 1, 5, 9, 19, 26, 29, 32 and 34 are substantially means-plus-function recitations ("means for …"), so their scope is limited to the corresponding disclosed structure (the dynamic cache manager program / TATHROTTLE module) and equivalents. That narrows them, which helps the patent owner but also means prior art need only match the disclosed structure — and Tandberg's detector DE + control unit CU is recognizably that structure.
• § 101 residual. The "algorithms" of claims 27/28/30/31 are mathematical calculations on admitted known quantities; if characterized as pure calculation, the Gulack reasoning above makes them obvious rather than ineligible, but the § 101 overlay is worth noting for a full invalidity opinion.

---

11. Bottom line

1. The two-limb core of the invention — grow a write cache, size it to the remaining tape capacity, flush before exceeding a safety-discounted nominal capacity — is the union of two closely related references. US 4,258,418 supplies the variable-capacity buffer; US 4,868,737 supplies everything else that matters: determining remaining medium capacity by counting motor pulses/measuring recorded length against a known full length, magnetic tape with BOT/EW/EOT markers, a comparison of buffer contents against medium-derived capacity, a capacity-change signal, and — critically — the guarantee that "no more data than can be recorded on the storage medium are stored on the buffer memory at any time," which is verbatim the requirement of independent claims 26/29/32/34.
2. Motivation is strong and over-determined: same field, same problem, same result, a disclosed capacity-dependent functional relationship, a finite set of predictable options, and a direction-of-adjustment change that KSR expressly classifies as a predictable variation.
3. The strongest single attack is Ground I against claims 32 and 34, which lack the "nominal < actual" limitation and whose "increase the addressable space … in accordance with the availability of the data storage capacity" reads almost directly onto Tandberg's control unit CU modifying "the logical storage capacity L" of the buffer as a function of the medium's available capacity A.
4. The weakest link is the "maximum storage capacity is less than the actual storage capacity" limitation of claims 1/5/9/19, which depends on my low-confidence reading of US 4,499,539 and, secondarily, on characterizing Tandberg's EW (early-warning) hole as the claimed margin. A full opinion should pull the 4,499,539 and 4,258,418 specifications (columns/lines) before finalizing that leg.
5. Two references the examiner cited are only § 102(e) art (4,868,737 and 4,947,319) because they issued after the 1989-07-14 priority date — but they survive § 103(c) because the assignees are unrelated, so they are properly available. The patent owner's remaining countermeasure is a Rule 131 swear-behind to an invention date before 1986-12-22, which the 1989 parent filing makes unlikely but not impossible.
6. Caution against the page's most common trap: the 44–82 entries in the "Cited By" tables and the Similar Documents list (including WO 1991/001525 A1, the patent's own PCT sibling) are not prior art and must not be used in any § 103 combination.