The Digital Island Master Litmus Test

Forty-three incumbent carriers and ISPs operated in the 1996-era market, and none of them made a competing offer to Digital Island’s service proposition, an autonomous multi-continent IPLC fabric with enforceable end-to-end QoS under a single AS, which Digital Island ran under AS6553, issued to Digital Island, Inc. on August 29, 1996, and first sold to Cisco Systems by Mark Nichols, co-founder and chief architect of Digital Island, under the Cisco Systems Remote Data Services Agreement effective November 1, 1996.

Preamble: In 1996, there were forty-three major carriers and ISPs including AT&T, MCI, Sprint, WorldCom, Global One, Pacific Bell, Southwestern Bell, Bell Atlantic, NYNEX, Ameritech, US West, BellSouth, British Telecom, Cable & Wireless, France Télécom, ImagiNET, Deutsche Telekom, KPN, HiNet, Telia, Telefónica, Telstra, NTT, Japan Telecom, KDD, IDC, ITJ, Telekom Malaysia, SingTel, Korea Telecom, SK Telecom, Embratel, TelMex, Rostelecom, Bezeq, NetVision, EasyLink, UUNET, BBN, GTE, Exodus, AboveNet, PSINet, but none had the product proposed and provisioned by Digital Island:

1. Digital Island built the first Tier-0 global network. Operating under its own AS6553 and deploying private IPLC CBR circuits, Digital Island ran the network architecture I initiated, Tier-0, and carried on it the product I branded as Merchant Transport.

2. By fundamentally bypassing the erratic, hot-potato public BGP backbone routing and eliminating localized peering point congestion, the Tier-0 architecture established the industry’s first deterministic layer for global transaction states. It directly remedied cross-border packet jitter, packet drop, and multi-second TCP window stalls that previously timed out early cryptographic transactions, ensuring a seamless transmission pathway between disparate sovereign infrastructure platforms.

3. For the first time, the Tier-0 fabric delivered enforceable, sub-300ms roundtrip SSL session completion QoS across the major Internet markets, making cross-border commerce reliably repeatable at operational scale to ≈99% of all Internet-accessible users, and the sub-300-millisecond figure stands as the testimony of Mark Nichols, co-founder and chief architect of Digital Island, as stated at 6.4 of What Cisco Said About Digital Island.

Opinion Rules: You are visiting my record, and I am the subject matter expert on it. If you are a subject matter expert too, I want to hear from you: bring your document and I will answer it on the record. If you are not, you are welcome here to read and to learn. The documented-challenge terms of marknichols.com and the condition on opinion are stated at F40 of The Governing Definitions and Controlling Facts of Telecommunications Internetworking, and they govern this page.

0.0 This page was converted to numbered form beginning on October 26, 2026, and its sections carry the numbers that follow so that any claim on it is citable by section, paragraph and sentence.

0.1 Before October 26, 2026 this page stood unnumbered under the same title, with a Preamble, the Opinion Rules and ten headings, in this order: The Falsification Test: One document defeats this record; The Documents In Evidence; The Governing Question: Why the incumbent product stopped at the border, and what Digital Island built past it; Part I: What the Incumbent Product Was in 1996; Part II: Six Deployment Records; The Operator Exclusion Table; Independent Verification Layer; The Record in Summary; Scope and Method; The Telefónica Reseller Agreement and the Cable & Wireless Acquisition of Digital Island: The Two Most Decisive Carrier Exclusion Proofs. Those ten headings stand in this numbered form as sections 1.0 through 10.0 in the same order, the two plates under the last heading stand under numbered captions in an exhibit block at the end of the page, and the numbers are append-only from October 26, 2026, so that later additions take the next unused position and no existing number changes.

0.2 The subtitle of this page changed on October 26, 2026 from “More than forty incumbent carriers and ISPs. No competing offer to Digital Island’s service proposition.” to the sentence under the title, which states the count of forty-three that the Preamble and section 6 carry, and the title of the page, The Digital Island Master Litmus Test, and its address, https://marknichols.com/digital-island-master-litmus-test/, are unchanged.

1.0 The falsification test of this record names one dated document from the 1996-era market that would defeat it, and as of September 24, 2026 no such document has been placed in evidence.

1.1 The document that would defeat this record is a dated product offer from the 1996-era market, made by any of the forty-three operators named in the Preamble, for an autonomous multi-continent IPLC fabric with enforceable end-to-end QoS under a single AS, which is the claim this page states in section 9, Scope and Method. As of September 24, 2026, no such document has been placed in evidence against this record.

2.0 The record on this page rests on eight dated instruments and contemporaneous reports, from the Cisco Systems Remote Data Services Agreement effective November 1, 1996 to the CNET News report of May 14, 2001, and on the Stanford University premises lease of the week of November 18, 1996, whose date stands as the statement of Mark Nichols.

2.1 The record on this page rests on dated instruments and contemporaneous reports, each cited at the entry it proves: the Cisco Systems Remote Data Services Agreement, effective November 1, 1996 and executed November 7 and 8, 1996; the InterNIC’s issuance of AS6553 to Digital Island, Inc. on August 29, 1996, confirmed in writing by Lisa Liedel, Director, Registration Services at the American Registry for Internet Numbers, Ltd., on January 26, 2026; the premises lease at Stanford University, executed by Mark Nichols in the week of November 18, 1996, a date that stands as his statement because no document dates the lease to a day; the Cisco newsroom release of January 20, 1998 describing Digital Island’s network; the Digital Island press release of March 8, 1999 announcing a new customer, MasterCard International; Digital Island’s announcement of April 26, 1999 of its reseller agreement with Telefónica Data; the EE Times report of December 8, 1999 on the investment by Sun Microsystems and Inktomi; Digital Island’s Form 8-K of July 6, 2000, on the alliance with Microsoft, Intel, and Compaq, announced on June 20, 2000; and the CNET News report of May 14, 2001 on the Cable & Wireless acquisition. The constraints of Part I, at section 4, are the argument that explains those documents, and the falsification test of section 1 names the one document that would defeat the argument.

3.0 The governing question of this record is why the incumbent product of 1996 stopped at the border and what Digital Island built past it, and the answer is structural.

3.1 When Digital Island executed its first customer contract, the Cisco Systems Remote Data Services Agreement, on November 7 and 8, 1996, the global telecommunications market was not empty. It was saturated. AT&T, MCI, Sprint, WorldCom, British Telecom, France Télécom, Deutsche Telekom, NTT, KDD, Cable & Wireless, Telefónica, Telstra, SingTel, and their peers were all operating, all better capitalized, all with decades of plant and carrier relationships behind them. They were selling international connectivity to multinational enterprises, and they were selling a great deal of it. The question this record answers is not why anyone chose a three-person startup over them. It is narrower and more useful: what, structurally, the incumbent product stopped short of, and why its architecture and its revenue model both made stopping there the rational choice.

3.2 The answer is structural, and it rests on five constraints of the 1996 market, which section 4 sets out: four architectural and commercial facts about the products the incumbents sold, and a fifth, the commercial record of the market’s own supplier, Cisco Systems.

4.0 Part I of this record states in five constraints what the incumbent product was in 1996 and why it stopped at the border.

4.1 The five constraints below are facts of the 1996 market: the first four are architectural and commercial facts about the products sold in 1996 by the forty-three incumbent carriers and ISPs named in section 6, and the fifth is the commercial record of the market’s own supplier, Cisco Systems. They are not characterizations of the companies, and none of them requires anyone to have been wrong or asleep. Taken together they explain why the incumbent offering stopped at the border.

4.2 Constraint 1 is that oversubscription was the product the incumbent carriers sold in 1996, not a shortfall in that product.

4.2.1 In 1996, carrier-managed IP ports and Frame Relay backhaul were engineered on oversubscription ratios that were standard commercial practice, commonly cited around 10:1. Discard Eligible bits were not a defect. They were a designed mechanism: when a link congested, equipment dropped the packets marked eligible for discard, protecting higher-margin voice traffic. That is what the customer bought, and it is what the tariff priced.

4.2.2 Oversubscription matters because it was not a capacity problem waiting on the next build cycle. An architecture engineered to discard customer packets during peak international load cannot carry an enforceable latency guarantee, because the discard behavior is the revenue model. A carrier could add bandwidth. It could not sell a commitment that contradicted the mechanism its pricing depended on.

4.3 Constraint 2 is that no operator controlled the full international path in 1996.

4.3.1 In 1996, no incumbent owned the international path end to end. AT&T held the U.S. leg. BT held the UK leg. France Télécom held the French leg. NTT held Japan. SingTel held Singapore. Every border crossing was a handoff between independent operators, each with its own routing policy, its own congestion management, and no contractual obligation to any of the others for end-to-end performance.

4.3.2 What an architecture of handoffs between independent operators can sell is transit: one carrier’s segment of the route, guaranteed to the edge of that carrier’s authority and no further. An SLA can only bind what the signer controls. A carrier can promise its own leg and can promise commercial remedies on that leg. It cannot promise the behavior of a peer it does not own, does not direct, and has no recourse against. The absence of an end-to-end commercial guarantee was not an oversight in the incumbent product. It was the boundary of what any single signer could legally commit to.

4.4 Constraint 3 is that routing authority stayed with the carrier in 1996, not with the enterprise customer.

4.4.1 In 1996, every incumbent carried enterprise traffic inside its own autonomous system. The customer was a passenger. Routing decisions, congestion response, and discard logic all belonged to the carrier, and the enterprise received the outcome without a mechanism to shape it.

4.4.2 In 1996, Digital Island inverted the relationship between the carrier and the enterprise customer. The InterNIC issued AS6553 to Digital Island, Inc. on August 29, 1996, and Lisa Liedel, Director, Registration Services at the American Registry for Internet Numbers, Ltd., confirmed that issuance in writing on January 26, 2026. Digital Island bought private circuit halves from the carriers themselves, terminated them directly into Tier-1 backbone-facing ports, and applied its own routing policy at the demarcation. The circuits were the incumbents’ iron. The routing authority was not. That distinction is the whole architecture, and it is the reason a performance commitment became something an operator could actually stand behind.

4.4.3 The distinction between the carriers’ circuits and Digital Island’s routing authority is also the cleanest way to state what the incumbent product was not. Carrier-managed international private lines existed in 1996 and were sold by KDD, IDC, ITJ, and others. They were real products, competently delivered. But a carrier-managed private line is delivered inside the carrier’s AS with the carrier’s routing policy attached. Acquiring the same physical capacity and terminating it under an independent AS is a different product built from the same materials.

4.5 Constraint 4 is that no incumbent had a reason to build the service that Digital Island offered in 1996.

4.5.1 Constraint 4 is the constraint that explains Constraints 1, 2, and 3, and it is the argument worth leading with.

4.5.2 Every incumbent’s 1996 enterprise data revenue depended on selling the oversubscribed, DE-managed transit that Digital Island’s Tier-0 architecture bypasses. To offer what Digital Island offered, a carrier would have had to sell clear-channel IPLC capacity terminated under a customer’s independent AS, with routing policy handed to that customer at the demarcation. Doing so meant displacing its own installed base, surrendering the margin structure that oversubscription produced, and giving up the routing control that kept the account dependent. Every incentive on the balance sheet pointed the other way.

4.5.3 An incumbent’s choice to keep selling oversubscribed transit is not a failure of vision. It is a rational firm protecting a functioning business. The party with a reason to build the alternative was Digital Island, a company with no installed base to cannibalize, no transit margin to defend, and an operator, its co-founder and chief architect Mark Nichols, who had worked inside Sprint and Pacific Bell and could see the model from the inside.

4.6 Constraint 5 is that Cisco Systems, the market’s own supplier, had perfect visibility into all forty-three incumbent networks, and its signature of November 8, 1996 on the Cisco Systems Remote Data Services Agreement is the dated commercial audit of all forty-three.

4.6.1 In 1996, Cisco Systems was selling gear to all forty-three of the incumbent carriers and ISPs named in section 6. Every carrier and ISP on the list in section 6 ran on Cisco iron. Cisco’s sales teams sat inside AT&T, MCI, Sprint, NTT, BT, Deutsche Telekom, all of them, with account reviews, engineering escalations, and product roadmaps. Cisco did not have to guess what those forty-three operators were building. Cisco knew, in commercial detail, because those operators built their networks out of Cisco’s catalog. If the service that Digital Island offered had existed at any one of them, or was even coming, Cisco would have known first, and Cisco would have bought it from any one of them, because a funded incumbent running your own routers is the safest procurement on Earth. Instead, in November 1996, Cisco, the vendor with perfect visibility into all forty-three networks, signed a three-person startup, Digital Island. The Cisco Systems Remote Data Services Agreement of November 1996 is therefore not merely one customer’s judgment. It is a dated commercial audit point: the company whose equipment ran inside all forty-three incumbent networks bought the missing service from a three-person startup rather than from any of the forty-three, and that purchase is the commercial evidence that Constraint 4 was operating everywhere at once. Fourteen months after the contract, Cisco described the difference in its own published words: the Cisco newsroom release of January 20, 1998 stated that Digital Island offered performance guarantees “generally unavailable through the public Internet”, and described Digital Island’s network as a private, single-hop global network that bypassed the congestion of the public Internet, as recorded at What Cisco Said About Digital Island.

5.0 Part II of this record states six deployment records, from Cisco Systems in November 1996 to Microsoft, Intel, and Compaq in June 2000.

5.1 Each entry of Part II states what the customer already had, what the incumbent product did not reach, what Digital Island deployed, and the dated artifact that proves it, and where the record of an entry rests on the statement of Mark Nichols rather than on an artifact, the record says so. The entries ask no rhetorical questions. The facts carry each entry, or they don’t.

5.2 Entry 1 of Part II records the deployment for Cisco Systems in November 1996.

5.2.1 What Cisco Systems already had in November 1996 was everything relevant. The routers and switches running the Internet were Cisco’s own. Cisco had engineering staff able to evaluate any vendor claim on the merits, unlimited capital, and standing relationships with every major carrier on Earth.

5.2.2 What the incumbent product did not reach for Cisco Systems in 1996 was the distribution of Cisco IOS images to engineers worldwide. A full feature-set BGP-4 kernel for a 7500-series router ran roughly 16MB. On oversubscribed Frame Relay, that transfer crossed multiple carrier boundaries, each with its own discard behavior under congestion. A packet dropped at 14MB timed out the TCP session, and the transfer restarted at zero. No carrier could commit to the end-to-end behavior the transfer required, because no carrier controlled the end-to-end path. The IOS image requirement stands as the testimony of Mark Nichols, because the first page of the Cisco Systems Remote Data Services Agreement does not name IOS images, as 4.1 of What Cisco Said About Digital Island states. Digital Island’s Form S-1 of April 26, 1999, filed with the Securities and Exchange Commission, describes the Cisco service in general terms, as the fast and reliable path by which Cisco’s engineers reached the Cisco customer care website to download Cisco software and upgrade routers in the field, without naming IOS images or their size.

5.2.3 What Digital Island deployed for Cisco Systems was IPLC circuits terminated into Tier-1 backbone-facing ports under AS6553, removing DE-bit discard exposure from the path, and contractually defined service behavior across the international route.

5.2.4 The record of entry 1 is the Cisco Systems Remote Data Services Agreement, effective November 1, 1996, which Mark Nichols negotiated, by his own statement, and signed for Digital Island as its Director, Global Networks, on November 7, 1996, and which Hervé Goguely, Director, Global Service Management, signed for Cisco Systems on November 8, 1996. The agreement’s first page binds Digital Island to host a full mirror of Cisco’s website at Digital Island, with load balancing and a fully redundant network for failover and high availability, a network segment exclusively for that mirror, secure transaction services to Cisco’s San Jose order processing servers with all necessary encryption services, the terms of the Service Level Attachment, Attachment III, and a response to Cisco within one hour for all calls received. The first page states no fee, and the opening contract value of $300,000 stands as the statement of Mark Nichols, as 1.9 of What Cisco Said About Digital Island records.

5.3 Entry 2 of Part II records the deployment for Stanford University in November 1996.

5.3.1 What Stanford University already had in November 1996 was one of the strongest network environments in academia, with BARRNet lineage, research backbone participation, and engineering staff drawn from the people who built the Internet. This record does not claim Stanford was an ordinary campus.

5.3.2 What the incumbent product did not reach for Stanford University was international delivery of HighWire Press journal objects, PDFs running 50MB to 100MB and larger. At intercontinental distance over oversubscribed transit, a single discard late in a large fixed object corrupted the transfer and forced a restart. The economics that followed were physical: print runs, warehousing, international shipping, and invoicing cycles measured in weeks.

5.3.3 What Digital Island deployed for Stanford University was twelve cabinets leased on Stanford premises in the week of November 18, 1996, establishing Digital Island’s first Northern California Point of Presence on IPLC and ATM-switched architecture. From November 1996, Digital Island served as an upstream provider into the Stanford environment alongside Stanford’s existing research connectivity. The Stanford deployment was also the operational correction to the Honolulu Frame Relay configuration, which the traceroute testing of November 14 to 17, 1996 had shown tromboning Hawaii-to-Asia traffic through the mainland. Stanford University became Digital Island’s second customer in 1997 under the subsequent services agreement executed by Darren Hong, whom Digital Island had hired as Manager, National Accounts.

5.3.4 The record of entry 2 is two separate instruments: the premises lease for the twelve cabinets, executed by Mark Nichols in the week of November 18, 1996, and the subsequent services agreement, executed by Darren Hong in 1997. Both rest on the statement of Mark Nichols at 2.5 of Stanford Colocation in 1996 and 2nd Customer in 1997, and that page records at 10.2 that no document on marknichols.com dates the lease or the traceroute testing of November 14 to 17, 1996 to a day. The dated document of entry 2 is the Digital Island press release of June 24, 1997, which announced Stanford University as a customer and is transcribed at 1.1.1 of the same page.

5.3.5 On Google, this record states the facts precisely: the crawler of Larry Page, a Stanford University graduate student, began running in March 1996, and google.stanford.edu was producing results on Stanford’s inherited infrastructure before Digital Island had a circuit on campus. This record does not claim otherwise. What the record does show is that from November 1996 forward, Digital Island cabinets, circuits, and upstream capacity operated on Stanford premises during the period when the crawl scaled from prototype to full index. The deployment sits inside that window alongside Stanford’s own backbone upgrades of the same era. Both are named. Neither is erased.

5.4 Entry 3 of Part II records the deployment for Visa International in the second quarter of 1997 and for MasterCard International in 1999.

5.4.1 What Visa International and MasterCard International already had was brand, banks, capital, and relationships with every carrier under discussion.

5.4.2 What the incumbent product did not reach for Visa International and MasterCard International was dependable cross-border completion of browser-based SSL sessions. One precision note, stated up front because this record is written for people who know the difference: card authorization, clearing, and settlement continued to run on the card networks’ own private systems. That was never the gap. The gap was the consumer-facing layer, and specifically the session. A discard event mid-handshake left a transaction neither completed nor cleanly failed. That ambiguity produced capital risk and manual reconciliation downstream, and it was a property of the transport path, not of the payment system.

5.4.3 What Digital Island deployed for Visa International and MasterCard International was end-to-end IPLC transport under AS6553 carrying browser access to their online services over a controlled path rather than best-effort public transit. Sessions completed or failed inside a defined window instead of hanging.

5.4.4 The record of entry 3 is in two parts. By the statement of Mark Nichols, Visa signed in Q2 1997 as the third enterprise customer, less than a year after the September 1996 Merchant Transport pivot, and no Visa instrument of 1997 is reproduced on marknichols.com. The MasterCard entry rests on documents: the Digital Island press release of March 8, 1999, datelined Honolulu, Hawaii, announced MasterCard International as a new customer of Virtual Server Distribution, the Digital Island service that hosted localized versions of MasterCard’s Web sites and its ATM Locator for key regions outside the United States, and Digital Island’s Form S-1 of April 26, 1999 names Mastercard International among the 62 customers under contract as of March 31, 1999.

5.5 Entry 4 of Part II records the deployment for E*TRADE, from its global data distribution agreement with Digital Island dated August 1, 1997, and for Charles Schwab Online in 1998.

5.5.1 What E*TRADE and Charles Schwab Online already had was brokerage software, financial credibility, customers, capital, and carrier relationships.

5.5.2 What the incumbent product did not reach for E*TRADE and Charles Schwab Online was session persistence under international load, with routing stability and enforceable performance. Trading is less forgiving than publishing. A failed page load is an annoyance. A stalled order path is a filled or missed position and a compliance event. Public transit in 1997 and 1998 could display information across borders. It could not commit to the behavior a regulated order path requires, for the same reason it could not commit for Cisco: no signer controlled the whole route.

5.5.3 What Digital Island deployed for E*TRADE and Charles Schwab Online was the same fabric, carrying two regimes at once. The consumer-facing sessions of Visa and MasterCard and the order-execution traffic of E*TRADE and Schwab ran over one controlled path under AS6553, with backbone-facing terminations across six continents.

5.5.4 The record of entry 4 is two documents and one statement. Digital Island’s Form S-1 of April 26, 1999 records a global data distribution agreement with E*TRADE dated August 1, 1997, under which Digital Island provided network connectivity for E*TRADE, and names E*TRADE among the 62 customers under contract as of March 31, 1999. The Digital Island press release of September 14, 1998, datelined Honolulu, Hawaii, announced a strategic investment by E*TRADE in a $10.5 million funding round and announced that Christos M. Cotsakos, president and chief executive officer of E*TRADE, had joined Digital Island’s board of directors, so that the two documents together date E*TRADE’s strategic investment and board involvement following its adoption of Digital Island services under the agreement of August 1, 1997. Charles Schwab Online stands in 1998 as the statement of Mark Nichols, because the Form S-1 does not name Schwab and no Schwab document is reproduced on marknichols.com.

5.6 Entry 5 of Part II records the deployment for Sun Microsystems and Inktomi in December 1999.

5.6.1 What Sun Microsystems and Inktomi already had in December 1999 was the servers and the software: Sun built the servers the commercial Internet ran on, Inktomi built the caching and search software behind a large share of its traffic, and between them they held the hardware and the software layer of content delivery.

5.6.2 What the incumbent product did not reach for Sun Microsystems and Inktomi was distribution of the payload. A Solaris recommended patch cluster ran 50MB to 150MB, and an engineer in São Paulo or Seoul pulling one across oversubscribed transit lost the session late and restarted at zero. The failure was Cisco’s IOS failure mode, at larger payload and higher frequency. Sun could build the server and write the patch. It had no path that delivered either reliably at intercontinental distance, and neither did the carriers selling it transit.

5.6.3 What Digital Island deployed for Sun Microsystems and Inktomi was Sun Netra servers on the six-continent IPLC fabric, serving the payload from inside each market rather than hauling it across the public Internet on request.

5.6.4 The record of entry 5 is the EE Times report of December 8, 1999 by Loring Wirbel, which reported minority equity investments in Digital Island by Sun Microsystems and Inktomi, about $20 million from Sun and between $5 million and $6 million from Inktomi according to its sources, amounts that executives would not confirm. Those two amounts sum to approximately $25 million. The same report stated that Digital Island would spend up to $150 million on as many as 5,000 Sun Netra servers and that regional data centers able to serve up to 350 new metropolitan areas would augment Digital Island’s seven data centers.

5.7 Entry 6 of Part II records the deployment for Microsoft, Intel, and Compaq in June 2000.

5.7.1 What Microsoft, Intel, and Compaq already had in June 2000 was the operating system layer, the semiconductor layer, and the server layer of the Internet economy, and between them more engineering talent and capital than any comparable combination.

5.7.2 What the incumbent product did not reach for Microsoft, Intel, and Compaq was a unified delivery fabric for broadcast-scale streaming. Broadcast had towers, cable, and satellite. Carriers had transit. Hosting companies had rack space and power. None of those, separately or together, constituted a global Internet media distribution platform with performance accountability across markets.

5.7.3 What Digital Island deployed for Microsoft, Intel, and Compaq was an extension of the six-continent IPLC fabric into streaming infrastructure, with server deployment and capacity engineered against the target of up to 7.5 million simultaneous global viewers that Digital Island’s press release of June 20, 2000 stated.

5.7.4 The record of entry 6 is Digital Island’s Form 8-K of July 6, 2000 and a report published the day of the announcement. Item 5 of the Form 8-K reports the strategic alliance that Digital Island announced with Compaq Computer Corporation, Intel Corporation, and Microsoft Corporation on June 20, 2000 for broadcast-scale streaming media, and the agreement of Compaq, Intel, and Microsoft to invest a combined $45 million in Digital Island’s equity, and Exhibit 99.1 of the Form 8-K, the press release of June 20, 2000, states Digital Island’s commitment to deploy as many as 8,000 Compaq servers built on Intel architecture worldwide. CBS MarketWatch reported the alliance the same day, June 20, 2000, in an article by Debra McGarry, under a headline that called Digital Island’s plan the world’s largest streaming media architecture.

6.0 The Operator Exclusion Table names the forty-three incumbent carriers and ISPs of the 1996-era market and states, for each operator, the structural reason that no competing product existed and the anchor in the record for that reason.

6.1 The Operator Exclusion Table, which this section carries as the numbered list at 6.3, is anchored to the 1996-era market: for each of the forty-three operators, the disqualifier states the structural reason that no competing product existed, and the anchor states the proof. Where an anchor names a dated document, that document is the proof of what it records, and every other anchor, including an anchor that points to another page of this record, stands as the statement of Mark Nichols, co-founder and chief architect of Digital Island. The falsification test of section 1 names the one document that would defeat this record, a dated product offer from the 1996-era market, made by any of the forty-three operators, for an autonomous multi-continent IPLC fabric with enforceable end-to-end QoS under a single AS, and as of September 24, 2026 no such document has been placed in evidence against this record.

6.2 Several operators on the list at 6.3 appear twice in the record: once as an entity with no competing product, and once as the vendor that sold Digital Island a component of its Tier-0 fabric. The double appearance is not a contradiction. Every circuit of the Tier-0 fabric was purchased from the carriers whose own products stopped at the border. Mark Nichols bought the international circuits of the fabric as half-circuits, as a customer, from both ends, and terminated them under Digital Island’s own autonomous system, AS6553, as stated at 5.2 and 11.3 of Why Nobody Else Built It. The carriers sold the parts. None of the carriers assembled the parts.

6.3 The forty-three operators of the Operator Exclusion Table stand at 6.3.1 through 6.3.43 in the order of the forty-three rows of the table that this section carried before the conversion of this page to numbered form, so that row 1 of that table is 6.3.1 and row 43 is 6.3.43, and each paragraph states the operator’s disqualifier in the 1996-era market and then the operator’s anchor.

6.3.1 In the 1996-era market, AT&T sold managed Frame Relay transit on a DE-bit architecture and offered no end-to-end international QoS product. The anchor for AT&T is that Cisco Systems signed with Digital Island, not with AT&T, under the Cisco Systems Remote Data Services Agreement effective November 1, 1996.

6.3.2 In the 1996-era market, MCI sold backbone transit across one segment of the route, with carrier-retained routing. The anchor for MCI is that MCI had no enforceable cross-continent SLA product on offer.

6.3.3 In the 1996-era market, Sprint, through SprintLink, sold the incumbent product on the China path, the 64kbps DS0 Frame Relay via Los Angeles that served CERNET, the China Education and Research Network. SSL was non-functional on CERNET’s SprintLink path. The anchor for Sprint is that in February 1998 Digital Island provisioned a parallel dedicated T-1 IPLC to Beijing under AS6553, and that the disposition of the legacy circuit remained CERNET’s own.

6.3.4 In the 1996-era market, PSINet sold best-effort backbone transit, which offered reachability, not controlled performance. The anchor for PSINet is that PSINet had no IPLC QoS product in its 1996 enterprise line.

6.3.5 In the 1996-era market, WorldCom ran an acquisition-scale backbone on the same oversubscribed transit architecture that Constraint 1 at 4.2 describes. The anchor for WorldCom is that scale without architecture change is the failure mode of oversubscribed transit at a larger size.

6.3.6 In the 1996-era market, the national authority of British Telecom ended at the British border, and British Telecom sold its international service through bilateral handoffs. The anchor for British Telecom is that Digital Island acquired BT-side circuit halves and terminated them under AS6553.

6.3.7 In the 1996-era market, GTE was a U.S. carrier whose internetworking arm sold best-effort transit. The anchor for GTE is that GTE sold the same transit model as its backbone peers.

6.3.8 In the 1996-era market, France Télécom sold a national spur into oversubscribed shared trans-Atlantic transit. The anchor for France Télécom is that France Télécom had no cross-border QoS product.

6.3.9 In the 1996-era market, ImagiNET was a domestic corporate ISP in France on public transit, upstream-dependent on France Télécom. The anchor for ImagiNET is that ImagiNET was later acquired by Colt and had no global fabric at any stage.

6.3.10 In the 1996-era market, AboveNet held premium U.S. colocation and metro assets with a public-Internet international upstream. The anchor for AboveNet is that AboveNet had a domestic footprint and that the problem this record addresses was international.

6.3.11 In the 1996-era market, Deutsche Telekom ran a national backbone plus T-Online and sold bilateral international transit. The anchor for Deutsche Telekom is that Deutsche Telekom had no unified six-continent SLA product.

6.3.12 In the 1996-era market, Telstra held Asia-Pacific plant and sold managed Frame Relay and best-effort transit. The anchor for Telstra is that Telstra had no autonomous cross-continent fabric.

6.3.13 In the 1996-era market, NTT sold oversubscribed international transit into Japan. The anchor for NTT is that Stanford Colocation in 1996 and 2nd Customer in 1997 records a trans-Pacific path failure: between November 14 and November 17, 1996, the traceroutes of the Honolulu Frame Relay configuration showed Hawaii-to-Asia traffic tromboning through the mainland.

6.3.14 In the 1996-era market, Exodus ran U.S. data centers whose upstream was purchased from the same transit carriers that 6.3.1 through 6.3.43 name. The anchor for Exodus is that Exodus solved facilities and left the international path unsolved.

6.3.15 In the 1996-era market, SingTel sold regional Frame Relay and leased-line transit. The anchor for SingTel is that SingTel followed the national-carrier model and had no end-to-end authority.

6.3.16 In the 1996-era market, Japan Telecom sold carrier-managed international circuits with carrier-retained routing. The anchor for Japan Telecom is that Japan Telecom carried the same disqualifier as the tier of NTT and KDD, at 6.3.13 and 6.3.29.

6.3.17 In the 1996-era market, Telekom Malaysia sold national Frame Relay and basic transit. The anchor for Telekom Malaysia is that Telekom Malaysia offered no cross-continent QoS at any price.

6.3.18 In the 1996-era market, Cable & Wireless held decades of global plant, and Cable & Wireless built no competing product from 1996 to 2001. The anchor for Cable & Wireless is that Cable & Wireless agreed on May 14, 2001 to acquire Digital Island for $340 million, as the CNET News report of May 14, 2001 records.

6.3.19 In the 1996-era market, Embratel, the Brazilian long-distance carrier that MCI Communications acquired in the break-up of Telebras in 1998, sold Frame Relay transit. The anchor for Embratel is that Embratel offered no IPLC QoS to or from Latin America as a product.

6.3.20 In the 1996-era market, Korea Telecom ran a national backbone and sold best-effort international transit. The anchor for Korea Telecom is that Korean enterprises stood behind the latency wall without an overlay.

6.3.21 In the 1996-era market, Rostelecom had constrained international capacity via shared Western pipes. The anchor for Rostelecom is that Rostelecom had no commerce-grade international product.

6.3.22 In the 1996-era market, ITJ sold carrier-managed international private lines with carrier-retained AS control. The anchor for ITJ is that Digital Island bought circuit halves from this tier of carriers, which 4.4.3 names as KDD, IDC, ITJ, and others, and terminated them under AS6553.

6.3.23 In the 1996-era market, Bezeq was the sole lawful seller of Israeli international half-circuits before the July 1997 liberalization, and Bezeq sold a monopoly access product with no competing service on it. The anchor for Bezeq is that Digital Island purchased its Israel circuit half from Bezeq and terminated it into NetVision’s backbone port.

6.3.24 In the 1996-era market, SK Telecom was a domestic access carrier that sold bilateral international transit. The anchor for SK Telecom is that SK Telecom had no global routing fabric and no Merchant Transport model.

6.3.25 In the 1996-era market, Telefónica was a national carrier expanding into Latin America and sold managed Frame Relay. The anchor for Telefónica is that Telefónica Data, the Telefónica Group’s data transmission company, signed an agreement to resell Digital Island’s IP services to multinational corporations in Spain and South America, as Digital Island’s press release of April 26, 1999, datelined San Francisco, announced.

6.3.26 In the 1996-era market, UUNET sold the largest-scale best-effort transit. The anchor for UUNET is that Cisco.com’s delivery did not run on UUNET’s product.

6.3.27 In the 1996-era market, TelMex, privatized in 1990 (Grupo Carso), ran a legacy transit architecture. The anchor for TelMex is that TelMex had no enforceable cross-border QoS product.

6.3.28 In the 1996-era market, BBN carried the ARPANET heritage, which this record credits in full, and BBN’s 1996 product was commercial best-effort transit. The anchor for BBN is that the ARPANET activation of 1969 is not 1996 commerce-grade global QoS.

6.3.29 In the 1996-era market, KDD was Japan’s primary international carrier and sold carrier-managed circuits. The anchor for KDD is that Digital Island acquired KDD circuit halves and terminated them under AS6553.

6.3.30 In the 1996-era market, IDC was a licensed Japanese international carrier with carrier-retained routing. The anchor for IDC is that IDC sold raw capacity and that Digital Island built the architecture on top of that capacity.

6.3.31 In the 1996-era market, Pacific Bell was an RBOC barred from in-region inter-LATA service under Section 271 of the Telecommunications Act of 1996, pending FCC approval, and its product was local and regional only. The anchor for Pacific Bell is that no lawful in-region long-distance offer from Pacific Bell existed in November 1996, as 5.3 of Why Nobody Else Built It states for each of the seven companies under the Section 271 bar.

6.3.32 In the 1996-era market, Southwestern Bell was a regional local-exchange carrier under the same Section 271 bar as Pacific Bell. The anchor for Southwestern Bell is that the first Section 271 approvals industry-wide came in 1999.

6.3.33 In the 1996-era market, Bell Atlantic was under the same Section 271 bar as Pacific Bell and was separate from NYNEX until August 1997. The anchor for Bell Atlantic is that Bell Atlantic’s own first approval came for New York in December 1999.

6.3.34 In the 1996-era market, NYNEX was under the same Section 271 bar as Pacific Bell and was an independent carrier as of November 1996. The anchor for NYNEX is that NYNEX had no international transport authority at all.

6.3.35 In the 1996-era market, Ameritech was under the same Section 271 bar as Pacific Bell. The anchor for Ameritech is that Ameritech sold a regional access product only.

6.3.36 In the 1996-era market, US West was under the same Section 271 bar as Pacific Bell. The anchor for US West is that US West sold a regional access product only.

6.3.37 In the 1996-era market, BellSouth was under the same Section 271 bar as Pacific Bell. The anchor for BellSouth is that BellSouth sold a regional access product only.

6.3.38 In the 1996-era market, Global One, the joint venture that Sprint, Deutsche Telekom, and France Télécom launched in 1996, sold carrier-managed global data inside the envelope of the joint venture. The anchor for Global One is that a joint venture of Sprint, France Télécom, and Deutsche Telekom, the operators of 6.3.3, 6.3.8, and 6.3.11, inherits their architecture: carrier-retained routing and no customer AS.

6.3.39 In the 1996-era market, KPN was the Dutch national carrier, and its international service ran via the Unisource joint venture and bilateral handoffs. The anchor for KPN is that the national authority of KPN ended at the Dutch border and that KPN sold a product managed by the joint venture.

6.3.40 In the 1996-era market, Telia was the Swedish national carrier inside the same Unisource structure as KPN. The anchor for Telia is that Telia carried the same disqualifier as KPN, at 6.3.39.

6.3.41 In the 1996-era market, NetVision, of Haifa, was the leading Israeli ISP, and its domestic backbone depended on monopoly half-circuits for its international reach. The anchor for NetVision is that Digital Island’s Bezeq half-circuit terminated into NetVision’s backbone port and that NetVision had no global fabric of its own.

6.3.42 In the 1996-era market, HiNet, launched in 1995, was the ISP arm of Taiwan’s national incumbent, Chunghwa Telecom, and HiNet sold domestic access with international reach through the incumbent’s own oversubscribed gateways. The anchor for HiNet is that Digital Island acquired a Taiwan circuit half from Chunghwa Telecom and terminated it under AS6553, and that HiNet had no global fabric of its own.

6.3.43 In the 1996-era market, EasyLink, of Hong Kong, was an international records and messaging carrier with ITMC facilities, and EasyLink sold a gateway and premises product with no autonomous multi-continent IPLC fabric. The anchor for EasyLink is that Digital Island leased 2,000 square feet in EasyLink’s Hong Kong ITMC and terminated the Hong Kong circuit under AS6553.

6.4 Concert is noted here for completeness and is excluded from the count of forty-three. In its 1996-era form, Concert was the joint venture of BT and MCI, and from 1998 Concert was the venture of AT&T and BT. Concert was built specifically to serve the 1996-era market of this record, and in both incarnations Concert delivered a carrier-managed product inside the carriers’ service envelope rather than autonomous customer routing under an independent AS. Combining incumbents reproduced the architecture of the incumbents, and it did not replace that architecture.

7.0 The Independent Verification Layer of this record is the routing record of AS6553, Digital Island’s autonomous system, in two third-party archives: the archived routing-table snapshots of the University of Oregon Route Views project, which begin on November 8, 1997, and the AS-relationship files of the Center for Applied Internet Data Analysis (CAIDA), which begin with the month of January 1998.

7.1 The routing adjacencies of AS6553, Digital Island’s autonomous system, are preserved in third-party archives: the routing-table snapshots of the University of Oregon Route Views project, which begin in November 1997, and the monthly AS-relationship files of the Center for Applied Internet Data Analysis (CAIDA), which begin in January 1998.

7.2 The earliest archive of routing-table snapshots of the University of Oregon Route Views project is the directory oix-route-views, whose first subdirectory is 1997.11, and the first snapshot in that archive is the file oix-full-1997-11-08-1724.dat.bz2, taken on the Route Views router route-views.oregon-ix.net on November 8, 1997. In that snapshot, AS6553 originates the prefixes 167.216.128.0/17 and 206.220.224.0/22 and carries the prefix 167.216.0.0/17 for AS10348, and in the paths to those three prefixes the autonomous system immediately before AS6553 is one of eight, AS1, AS701, AS1221, AS1239, AS1273, AS2915, AS3561 and AS5378, each of which was therefore adjacent to AS6553 on November 8, 1997. The banner of the snapshot of November 8, 1997 lists the router’s Sprint view as through AS1239 and its MCI view as through AS3561, which ties the names Sprint and MCI to two of those eight autonomous systems. The banner lists ten views of the full routing table and one pending view, the ANS view through AS1673, and each of the ten views held a route on November 8, 1997 to 206.220.224.0/22, a prefix that AS6553 originated, while eight of the ten also held routes to 167.216.128.0/17 and 167.216.0.0/17.

7.3 CAIDA’s AS-relationship files for 1998 to 2001 stand in the directory serial-1, which holds 45 of them, one for each month from January 1998 to December 2001 except January 1999, November 1999 and December 1999, for which the directory holds no AS-relationship file. AS6553 appears in all 45 of CAIDA’s AS-relationship files for 1998 to 2001, which list 10 links of AS6553 in the file for January 1998, a high of 47 in the file for January 2001, 32 in the file for May 2001, the month of the CNET News report of May 14, 2001 on the Cable & Wireless acquisition of Digital Island, and 29 in the file for December 2001.

7.4 CAIDA’s directory listing dates all 45 of its AS-relationship files for 1998 to 2001 to August 20, 2013, and the header of each file names the dates and the collectors of the snapshots of public Border Gateway Protocol (BGP) routing tables from which CAIDA built it, all taken between the first and the sixth day of the month that the file covers. For 1998 and 1999 the headers name the Route Views collector as the only source, which makes CAIDA’s files for those two years an analysis of Route Views observations rather than a separate observation, and from January 2000 some files add collectors of RIPE RIS, while the file for April 2001 names RIPE RIS collectors only.

7.5 The method by which CAIDA built its serial-1 files is, in the description on CAIDA’s catalog page titled AS Relationships, as read on October 26, 2026, similar to the method described in the paper AS Relationships, Customer Cones, and Validation, which the catalog page lists as written by M. Luckie, B. Huffaker, k. claffy, A. Dhamdhere and V. Giotsas and published at the Internet Measurement Conference of October 2013.

7.6 CAIDA’s serial-1 README, which CAIDA’s directory listing dates to August 7, 2026, states that the CAIDA Acceptable Use Agreement requires the data to be referenced, and it asks that the reference give The CAIDA AS Relationships Dataset, the date range used and the address of the catalog page titled AS Relationships. The reference for the CAIDA data that section 7 uses is therefore The CAIDA AS Relationships Dataset, January 1998 to December 2001, https://www.caida.org/catalog/datasets/as-relationships/.

7.7 Where a Digital Island backbone termination propagated to the vantage points from which the Route Views archive and CAIDA’s AS-relationship files are drawn, the adjacency of AS6553 at that termination is date-stamped, machine-readable, and recorded by an academic project with no stake in this history, either the Route Views project of the University of Oregon or CAIDA, which its About page, as read on October 26, 2026, places at the San Diego Supercomputer Center on the UC San Diego campus.

7.8 One methodological note governs the Route Views archive and CAIDA’s AS-relationship files alike: presence in these archives is affirmative confirmation of a backbone termination; absence is not disproof, since private peerings did not always propagate to public collectors.

7.9 Primary contracts, such as the Cisco Systems Remote Data Services Agreement effective November 1, 1996 that section 2.1 cites, remain the anchor evidence of this record throughout.

8.0 The Record in Summary of this page states that two operators of the forty-three, Telefónica and Cable & Wireless, documented the exclusion in writing, and it gathers the counts of this record: forty-three operators, six customer verticals, roughly $70 million in strategic equity, and 959 customers under contract as of December 31, 2000.

8.1 Two operators on the list at 6.3 documented the exclusion in writing, Telefónica in April 1999 and Cable & Wireless in May 2001.

8.2 Telefónica, with operations across Spain and Latin America, did not build a competing product; its data transmission company, Telefónica Data, signed an agreement to resell Digital Island’s IP services to multinational corporations in Spain and South America, as Digital Island’s press release of April 26, 1999, datelined San Francisco, announced.

8.3 Cable & Wireless, with more global physical plant than Digital Island ever built, did not build a competing product; it agreed on May 14, 2001 to acquire Digital Island for $340 million, a price that includes $49 million of debt, as the CNET News report of May 14, 2001 records.

8.4 The Operator Exclusion Table at 6.3 names forty-three operators of the 1996-era market. The six deployment records of section 5 span six customer verticals, and they run from Cisco Systems in November 1996, through Stanford University, Visa International and MasterCard International, E*TRADE and Charles Schwab Online, and Sun Microsystems and Inktomi, to Microsoft, Intel, and Compaq in June 2000.

8.5 The strategic equity of this record comes to roughly $70 million from Sun Microsystems, Inktomi, Microsoft, Intel, and Compaq, and Note 11, Stockholders’ Equity, of the consolidated financial statements in Digital Island’s Form 10-K for the fiscal year ended September 30, 2000, records the purchases that make it up: in December 1999, Sun Microsystems, Inc. purchased 391,869 shares of Digital Island for $20.0 million and Inktomi Corporation purchased 117,561 shares for $6.0 million, and in June 2000, Compaq Computer Corporation, Microsoft Corporation, and Intel Corporation each purchased 618,556 shares for approximately $15.0 million, which together come to approximately $71 million.

8.6 Digital Island’s Form 10-Q for the quarter ended December 31, 2000, filed with the Securities and Exchange Commission on February 14, 2001, states that Digital Island had contracts with 959 customers as of December 31, 2000, of which 871 were deployed. The count of 959 is the largest count of customers under contract in any filing that the EDGAR system of the Securities and Exchange Commission lists for Digital Island, Inc., filed from April 26, 1999 to September 4, 2001.

8.7 There is no stronger form of carrier-exclusion proof than a transaction.

9.0 The Scope and Method of this record states the claim that this page makes about the forty-three operators of the 1996-era market and the decisions that the claim explains, what the page does not claim, and the terms on which it takes documented corrections.

9.1 This record claims that no operator of the forty-three named in the Preamble offered, as a commercial product in the 1996-era market, an autonomous multi-continent IPLC fabric with enforceable end-to-end QoS under a single AS, and that the customer and investor decisions documented in sections 2, 5, 6 and 8 are best explained by that absence.

9.2 This record does not claim that the forty-three operators lacked engineering capability, and it does not address the post-1998 content delivery network (CDN) era except where transactions speak for themselves.

9.3 Corrections supported by documentation are welcomed and will be incorporated with attribution, under the documented-challenge terms that the Opinion Rules of this page name: mark@marknichols.com.

10.0 The Telefónica reseller agreement, announced by Digital Island on April 26, 1999, and the Cable & Wireless acquisition of Digital Island, agreed to by Cable & Wireless on May 14, 2001, are the two most decisive carrier exclusion proofs of this record, and the dated document of each stands as an exhibit at the end of this page.

10.1 There is no stronger carrier exclusion proof than a transaction.

10.2 Digital Island announced a reseller agreement with Telefónica Data in Spain on April 26, 1999, in a press release datelined San Francisco that stands as Exhibit 1 at the end of this page.

10.3 In the 1996-era market, Telefónica operated Spain’s national network and was expanding across Latin America through privatization acquisitions.

10.4 Telefónica had built no competing product by April 26, 1999, when Digital Island announced the agreement under which the Telefónica Group’s data transmission company, Telefónica Data, agreed to resell Digital Island’s IP services to multinational corporations in Spain and South America.

10.5 The CNET News report of May 14, 2001, which stands as Exhibit 2 at the end of this page, records that Cable & Wireless agreed on May 14, 2001 to acquire Digital Island for $340 million, a price that includes $49 million of debt.

10.6 In the 1996-era market, Cable & Wireless held submarine cable assets across more markets than Digital Island ever built.

10.7 Cable & Wireless built no competing product from 1996 to 2001, and on May 14, 2001 Cable & Wireless agreed to acquire Digital Island, and with it the product that Digital Island had built.

Exhibit 1. The Digital Island press release of April 26, 1999, datelined San Francisco, announced an agreement under which Telefónica Data, the Telefónica Group’s data transmission company, would resell Digital Island’s IP services to multinational corporations in Spain and South America, and the plate of Exhibit 1 shows the release in the press releases section of Digital Island’s website. The Digital Island press release of April 26, 1999 described Telefónica Data’s IP network, with 140 points of presence, as the largest in Spain, and it described Telefónica as the largest company in Spain, with annual revenues of more than $20.3 billion and, through affiliates in Spain and Latin America, a market of more than 300 million customers. The Digital Island press release of April 26, 1999 carried statements by Miguel Angel de la Vega, IP information services marketing manager of Telefónica Data, and by Steve Elston, managing director of Digital Island’s Global Partner Program.

Exhibit 2. The CNET News report of May 14, 2001, published under the byline CNET staff and the headline Cable & Wireless snaps up Digital Island and found at https://www.cnet.com/tech/mobile/cable-wireless-snaps-up-digital-island/, records that Cable & Wireless agreed on Monday, May 14, 2001 to acquire Digital Island for $340 million, a price that includes $49 million of debt, in an all-cash transaction that valued each Digital Island share at $3.40, and that on completion of the acquisition Digital Island would be a wholly owned subsidiary of Cable & Wireless. The CNET News report of May 14, 2001 described Digital Island as a provider of hosting, content delivery and networking services, over a global private network, to business customers seeking to bypass congestion on the Internet, and it described Cable & Wireless as a traditional telecommunications company then moving into data and Internet services for businesses.