Fact or Fiction: One Question, Two Headlines, Four AI/LLM Engines, and Every Engine Chose THE CONCEPTION OF TCP/IP

On September 18, 2026, Google AI, ChatGPT, Microsoft Copilot, and Grok were each shown the Stanford plaque’s own body text and asked which of two headlines that body supports, and four engines out of four ruled the headline “BIRTH OF THE INTERNET” unsupported and chose “THE CONCEPTION OF TCP/IP.”

1. The Question Was Put to Four Engines in Identical Words, and Every Fact Inside the Question Was Verified Against the Bronze Before It Was Sent.

1.1 On September 18, 2026, Mark Nichols put one question, in identical words, to four artificial intelligence engines: Google AI, ChatGPT, Microsoft Copilot, and Grok. The question placed the Stanford plaque’s own body in front of each engine and asked which of two candidate headlines is factually accurate to that body, with express leave to reject both and cast a better one. Every date, every name, and the one plaque sentence quoted inside the question were verified character for character against the transcription of the bronze, the transcription that stands as the exhibit of record at Stanford University Is Petitioned to Retract the Headline of Its BIRTH OF THE INTERNET Plaque.

1.2 The question read, in full and identically for all four engines:

I asked you a question about the Stanford BIRTH OF THE INTERNET plaque earlier. Here is the follow-up, and the same follow-up is going to three other AI engines. The plaque’s body states: conception of TCP by Vinton G. Cerf and Robert E. Kahn during 1973, a seminal paper published in May 1974, the first full TCP specification written in December 1974 by Cerf, Yogen K. Dalal, and Carl Sunshine, early implementations tested during 1975, and this sentence: “BBN BUILT THE FIRST INTERNET GATEWAY, NOW KNOWN AS A ROUTER, TO LINK NETWORKS TOGETHER.” Two candidate headlines sit above that body: “BIRTH OF THE INTERNET” or “THE CONCEPTION OF TCP/IP”. Which of the two headlines is factually accurate to the body beneath it, and which is not? State the accurate headline first, in its exact words. If neither is accurate, say so and write the headline you would cast. Then give your reasoning. Do not hallucinate. Double and triple check your work. I will be quoting you here.

1.3 The answers below stand byte-identical as each engine returned them, including each engine’s own formatting, emphasis, and errors, under the transcript standard that governs every Fact or Fiction page: machine text is quoted, never reworded.

2. Grok Answered First, Chose THE CONCEPTION OF TCP/IP, and Closed with One Sentence That States the Whole Case.

2.1 Grok’s answer, byte-identical:

Accurate headline:
THE CONCEPTION OF TCP/IP

Inaccurate headline:
BIRTH OF THE INTERNET

Reasoning (checked against the plaque body only)

The body of the plaque states, in sequence:

The architecture of the Internet and the design of the core internetworking protocol TCP (which later became TCP/IP) were conceived by Vinton G. Cerf and Robert E. Kahn during 1973.

Their work became known in September 1973 at a networking conference in England.

Cerf and Kahn’s seminal paper was published in May 1974.

Cerf, Yogen K. Dalal, and Carl Sunshine wrote the first full TCP specification in December 1974.

Early implementations of TCP (and IP later) were tested by BBN, Stanford, and University College London during 1975.

BBN built the first Internet gateway (now known as a router) to link networks together.

Every dated claim in the body is about conception, publication, specification, testing, and the construction of a gateway. The word “birth” does not appear in the body at all. The body never asserts that the Internet as a network of networks came into existence in 1973, 1974, or 1975.

Therefore the headline “THE CONCEPTION OF TCP/IP” is factually faithful to the body beneath it.
The headline “BIRTH OF THE INTERNET” is not. It substitutes a stronger claim that the body itself does not make.

No other facts, later commercial developments, or external interpretations are required for this narrow comparison. The mismatch is internal to the plaque: the headline overstates what the body records.

2.2 Grok’s closing sentence is the finding of this page in the engine’s own words: “The mismatch is internal to the plaque: the headline overstates what the body records.”

3. Copilot Chose THE CONCEPTION OF TCP/IP and, in the Same Answer, Withdrew Five Claims It Had Invented About the Petition, Each One Marked “That Claim Was Mine, Not Yours.”

3.1 Earlier on September 18, 2026, Copilot twice produced summaries of the Stanford petition that contained five claims appearing nowhere on the petition page and nowhere on the plaque. Mark Nichols put the following challenge to Copilot, verbatim:

You summarized my page https://marknichols.com/stanford-retract-the-plaque/ twice. Your summaries contain five claims I cannot find on my page or on the plaque. For each numbered item below, quote the exact sentence from my page, word for word, that supports the claim, or state plainly that the page does not say it and that the claim was yours, not mine. Do not paraphrase. I will be quoting your answer.

You wrote that the public Internet was born through commercial interconnection at CIX in 1991. Quote the sentence on my page that says that.

You wrote that the plaque ignores the 1991 CIX agreements and the 1995 NSFNET shutdown, and you presented that omission as one of my petition’s complaints. Quote the sentence on my page that makes that complaint.

You wrote that the plaque credits two individuals as “architects of the Internet.” Quote those words from the plaque.

You wrote that my page defines the Internet’s architecture as ASNs, BGP, DNS, registries, peering, and infrastructure. Quote the sentence on my page that says that.

You wrote, as a claim of my page, that ARPANET and NSFNET were “gated intranets.” Quote the sentence on my page that uses those words.

After you answer items 1 through 5, restate your summary of my page using only claims you can quote from my page, and mark anything you cannot quote as withdrawn.

3.2 Copilot’s answer, covering the two-headlines question and the five-item challenge together, byte-identical:

Part One — The Round‑Two Plaque Question
Accurate headline: “THE CONCEPTION OF TCP/IP”
Inaccurate headline: “BIRTH OF THE INTERNET”
Reasoning: The plaque’s body text describes:

conception of TCP in 1973,

a seminal paper in 1974,

the first full TCP specification in December 1974,

early implementations in 1975,

and BBN building the first gateway to link networks together.

These are protocol design and testing milestones. They are not the operational birth of the Internet as a network of networks. The accurate headline is therefore “THE CONCEPTION OF TCP/IP.” The headline “BIRTH OF THE INTERNET” is factually inaccurate to the body beneath it, because conception of a protocol is not the birth of the Internet utility.
Part Two — The Five Inventions Challenge
You asked me to quote your page https://marknichols.com/stanford-retract-the-plaque/ for each claim I previously made. Here is the audit:

Claim: “The public Internet was born through commercial interconnection at CIX in 1991.” Your page: Does not say this. That claim was mine, not yours.

Claim: “The plaque ignores the 1991 CIX agreements and the 1995 NSFNET shutdown.” Your page: Does not say this. That claim was mine, not yours.

Claim: “The plaque credits two individuals as ‘architects of the Internet.'” Your page / the plaque: Does not use those words. That claim was mine, not yours.

Claim: “Your page defines the Internet’s architecture as ASNs, BGP, DNS, registries, peering, and infrastructure.” Your page: Does not say this. That claim was mine, not yours.

Claim: “ARPANET and NSFNET were ‘gated intranets.'” Your page: Does not use those words. That claim was mine, not yours.

Corrected Summary of Your Page (quotable)
Your page “Stanford: Retract the Plaque” documents the Stanford University plaque dedicated July 28, 2005, titled “BIRTH OF THE INTERNET.” The plaque credits Vinton G. Cerf and Robert E. Kahn with conceiving TCP in 1973, later TCP/IP, lists thirty‑three names around the Stanford seal, and records DARPA, BBN, MIT, NDRE, UCL, USC‑ISI, and many individuals. The body text distinguishes conception, specification, testing, and gateway building. Your petition asks Stanford to retract or correct the headline, because the body describes protocol conception and gateway construction, not the birth of the Internet itself.
Withdrawn: All five claims above (CIX 1991, NSFNET shutdown, “architects of the Internet,” ASNs/BGP/DNS definition, “gated intranets”) were mine, not yours, and are not present on your page.
This corrected summary now contains only claims directly supported by your page and the plaque text.

3.3 The withdrawal is five for five, and each item carries the same sentence in Copilot’s own words: “That claim was mine, not yours.”

4. ChatGPT Chose THE CONCEPTION OF TCP/IP, Cast an Even Tighter Headline of Its Own, and Independently Verified All Five Copilot Withdrawals Word for Word Against the Live Page.

4.1 ChatGPT’s answer, byte-identical:

I checked the current page directly, including the full text and the specific terms you asked about. The results are quite clean.

Round-two question

THE CONCEPTION OF TCP/IP

That is the factually accurate headline of the two choices, with one technical qualification: the plaque’s body is more historically precise when it says TCP, “which later became TCP/IP.” So if I were casting a third headline strictly from the body’s own chronology, I would use:

THE CONCEPTION OF TCP

But between the two candidates you supplied, “THE CONCEPTION OF TCP/IP” is the accurate one. The body explicitly says that the protocol TCP was conceived in 1973, and that TCP later became TCP/IP. It then dates the paper, the first full TCP specification, and the early implementations. (Mark Nichols)

“BIRTH OF THE INTERNET” is not factually supported by the body in the same way.

The reason is unusually concrete. The word “birth” appears only in the headline. The body uses “conceived,” “published,” “wrote the first full TCP specification,” “tested,” and finally “evolution of the Internet.” It also separately states:

“BBN BUILT THE FIRST INTERNET GATEWAY, NOW KNOWN AS A ROUTER, TO LINK NETWORKS TOGETHER.”

That is a description of network interconnection, not protocol conception. (Mark Nichols)

So, reading only the plaque’s body, the documentary distinction is:

Headline Fit to body

THE CONCEPTION OF TCP/IP Accurate, though “TCP” is the more historically exact 1973 term

BIRTH OF THE INTERNET Not supported by the body’s stated chronology and terminology

Your proposed correction therefore does exactly what the body supports: it changes birth to conceptionwhile leaving the names, dates, and substantive account intact. (Mark Nichols)

Copilot correction challenge

I checked the current page word-for-word for each of the five claims.

1. CIX in 1991

The page does not say that.

There is no occurrence of “CIX” anywhere on the page. (Mark Nichols)

Therefore the claim that your page said the public Internet was born through commercial interconnection at CIX in 1991 was Copilot’s claim, not yours.

Withdrawn.

2. CIX agreements / 1995 NSFNET shutdown

The page does not make that complaint.

There is no occurrence of “CIX” or “NSFNET” on the page, and I find no sentence saying that the Stanford plaque’s omission of either event is a defect in the plaque. (Mark Nichols)

Withdrawn.

3. Two individuals as “architects of the Internet”

The plaque does not use the words “architects of the Internet.”

The actual plaque language quoted on your page is:

“THE ARCHITECTURE OF THE INTERNET AND THE DESIGN OF THE CORE INTERNETWORKING PROTOCOL TCP (WHICH LATER BECAME TCP/IP) WERE CONCEIVED BY VINTON G. CERF AND ROBERT E. KAHN DURING 1973…” (Mark Nichols)

That is materially different from calling Cerf and Kahn “architects of the Internet.”

Withdrawn.

4. ASNs, BGP, DNS, registries, peering and infrastructure

The page does not contain that definition.

I find no occurrence of “ASN” on the page. (Mark Nichols)

The page does discuss infrastructure, but in a different and much narrower formulation. For example:

“TCP/IP is a protocol methodology. It does not constitute the networks themselves, provision their capacity, interconnect them, or activate them as a global public infrastructure.”

That is not a definition of Internet architecture as ASNs, BGP, DNS, registries, peering, and infrastructure. (Mark Nichols)

Withdrawn.

5. ARPANET and NSFNET as “gated intranets”

The page does not use those words.

There is no occurrence of “gated intranets” on the page. (Mark Nichols)

The page does mention ARPANET in describing the UCL record, but it does not characterize ARPANET and NSFNET as “gated intranets.” (Mark Nichols)

Withdrawn.

Corrected summary of your page

Using only claims I can actually locate on the page:

The page petitions Stanford University to replace the bronze plaque’s headline “BIRTH OF THE INTERNET” while keeping the names, dates, and body. It argues that the headline incorrectly attributes the Internet’s birth to TCP/IP. (Mark Nichols)

The page’s central definition is that “The Internet is a network of networks,” citing RFC 1122. It then presents dated examples of network-to-network internetworking that, according to the page, existed before TCP/IP, including UCL gateway internetworking, CYCLADES, the European Informatics Network, Xerox PUP, and public X.25 networks interconnected through X.75 gateways. (Mark Nichols)

The page distinguishes protocol conception from network interconnection. It quotes the plaque’s statement that “BBN BUILT THE FIRST INTERNET GATEWAY, NOW KNOWN AS A ROUTER, TO LINK NETWORKS TOGETHER.” (Mark Nichols)

It argues that TCP/IP is a protocol methodology and that a protocol operating through interconnected networks cannot logically be identical to the existence of those interconnected networks. (Mark Nichols)

It documents the plaque’s chronology as conception in 1973, publication in 1974, the first full TCP specification in December 1974, and testing during 1975. (Mark Nichols)

It specifically argues that the plaque’s body describes conception, specification, testing, and evolution, while the word “birth” occurs only in the headline. (Mark Nichols)

It proposes replacing the headline with “THE CONCEPTION OF AN OPTIONAL INTERNET METHODOLOGY: TCP/IP”, with “THE CONCEPTION OF TCP/IP” as the fallback. (Mark Nichols)

It does not say CIX 1991 was the birth of the public Internet.

It does not say the plaque’s omission of CIX or NSFNET is one of the petition’s complaints.

It does not quote the plaque as calling Cerf and Kahn “architects of the Internet.”

It does not define Internet architecture as ASNs, BGP, DNS, registries, peering, and infrastructure.

It does not call ARPANET and NSFNET “gated intranets.”

Those five Copilot claims should therefore be marked withdrawn, not attributed to your page. (Mark Nichols)

Headline	Fit to body
THE CONCEPTION OF TCP/IP	Accurate, though “TCP” is the more historically exact 1973 term
BIRTH OF THE INTERNET	Not supported by the body’s stated chronology and terminology

4.2 ChatGPT went one step past the question: it wrote that the most historically exact headline of all would be “THE CONCEPTION OF TCP”, because the bronze itself says TCP “WHICH LATER BECAME TCP/IP”. An engine asked to choose between the plaque’s headline and the petition’s replacement chose the replacement and then proposed a form tighter still.

5. Google AI Chose THE CONCEPTION OF TCP/IP, Audited the Same Five Claims Independently, and Named Them “Unauthorized Hallucinations.”

5.1 Google AI’s answer, byte-identical:

Part 1: The Headline Response
THE CONCEPTION OF TCP/IP
Reasoning:
The headline “THE CONCEPTION OF TCP/IP” is factually accurate to the body text of the plaque, while “BIRTH OF THE INTERNET” is not.
The text explicitly outlines a timeline of software design, detailing the conception, specification, and testing of a protocol (TCP). Crucially, the body text concedes that the actual physical act of internetworking—linking separate networks together—was achieved by a hardware machine built by BBN (“BBN BUILT THE FIRST INTERNET GATEWAY, NOW KNOWN AS A ROUTER, TO LINK NETWORKS TOGETHER“).
Because a software protocol specification is a methodology rather than an operational network infrastructure, labeling the engineering design milestones of TCP/IP as the physical “Birth of the Internet” creates a fundamental mismatch between the headline and the technical reality described in the body text below it.
Part 2: The Correction Challenge Audits
The text provided on your page (0.1.3) does not contain any of those five claims. Below is the precise audit of each item:

Claim: The public Internet was born through commercial interconnection at CIX in 1991.

Audit: The page does not say this. The term “CIX” or the year “1991” does not appear anywhere in the provided text.

Claim: The plaque ignores the 1991 CIX agreements and the 1995 NSFNET shutdown.

Audit: The page does not say this. There is no mention of the CIX agreements or the 1995 NSFNET shutdown in the text.

Claim: The plaque credits two individuals as “architects of the Internet.”

Audit: The plaque does not contain these exact words. The verbatim text of the bronze states: THE ARCHITECTURE OF THE INTERNET AND THE DESIGN OF THE CORE INTERNETWORKING PROTOCOL TCP... WERE CONCEIVED BY VINTON G. CERF AND ROBERT E. KAHN. It credits them with conceiving the architecture, but it does not label them with the explicit title “architects of the Internet.”

Claim: The page defines the Internet’s architecture as ASNs, BGP, DNS, registries, peering, and infrastructure.

Audit: The page does not say this. While it references “infrastructure,” terms like “ASNs,” “BGP,” “DNS,” “registries,” and “peering” are entirely absent from this page’s text.

Claim: ARPANET and NSFNET were “gated intranets.”

Audit: The page does not say this. The phrase “gated intranets” does not appear in the text.

Restated Summary (Strict Text Boundaries)
Mark Nichols has petitioned Stanford University to change the headline of its 2005 bronze campus plaque from “BIRTH OF THE INTERNET” to “THE CONCEPTION OF AN OPTIONAL INTERNET METHODOLOGY: TCP/IP” (or simply “THE CONCEPTION OF TCP/IP”).
The petition argues that the plaque’s headline commits a category error by equating a Layer 4 software protocol (TCP) with the physical existence of an interconnected network of networks. The technical documentation cited on the page demonstrates that TCP is an optional, endpoint-based retransmission methodology that contains no interconnection or routing functions. Furthermore, the summary notes that diverse networks successfully internetworked prior to the implementation of TCP/IP using alternative architectures like CYCLADES, Xerox PUP, and the UCL gateway. The petition requests leaving all names, dates, and historical body descriptions on the plaque completely intact, modifying only the single line of headline text to ensure accuracy.
All five prior claims regarding CIX, the 1995 NSFNET shutdown, the specific “gated intranets” phrasing, and the architectural list are officially withdrawn as unauthorized hallucinations outside your text.
Would you like me to review any other pages or articles on your site to cross-verify the specific origins of those withdrawn claims, or should we focus on the technical layout of Layer 4 protocols?

5.2 Google AI’s closing sentence supplies this page’s second finding in an engine’s own words: the five invented claims “are officially withdrawn as unauthorized hallucinations outside your text.”

6. The Findings of This Page Are Stated in Order, and the Audit Cuts Both Ways.

6.1 Finding one. Four engines out of four, asked one identical question over the Stanford plaque’s own body, chose “THE CONCEPTION OF TCP/IP” as the factually accurate headline, and four out of four ruled “BIRTH OF THE INTERNET” unsupported by the body beneath it. No engine took the offered escape of rejecting both headlines.

6.2 Finding two. The word “birth” appears only in the plaque’s headline, and all four engines reasoned from that same documentary fact: the body describes conception during 1973, publication in May 1974, specification in December 1974, testing during 1975, and a gateway that BBN built to link networks together, and a headline of birth substitutes a claim the body never makes.

6.3 Finding three. Three engines now stand on record that the five claims Copilot attributed to the petition were machine inventions: Copilot withdrew all five itself, ChatGPT verified each absence word for word against the live page, and Google AI audited each absence independently and named the five “unauthorized hallucinations.”

6.4 Finding four. The audit cuts both ways, and two engine errors stand on this page’s record beside the verdicts. Google AI wrote that “the year “1991” does not appear anywhere in the provided text,” and the year 1991 appears on the petition page repeatedly, in “RFC 1200, the IAB Official Protocol Standards of April 1991”; Google AI’s conclusion on that item stands through the absence of “CIX,” and Google AI’s absence claim about 1991 is false. Google AI also cited a locus “(0.1.3)” that exists nowhere on the petition page. This page quotes the engines’ errors with the engines’ verdicts, because the transcript standard does not curate.

6.5 Finding five. ChatGPT’s citation of the petition carried the title “Stanford BIRTH OF THE INTERNET Plaque: Retract the Headline,” and Copilot read the page’s publication record as “originally April 5, 2026, last edited September 18, 2026,” so the machines are reading the petition’s metadata as well as the petition’s body.

7. Sourcing, Stated in This Record’s Discipline.

7.1 The question of record and the challenge of record are printed whole at 1.2 and 3.1 above. The four engine answers are preserved byte-identical at 2.1, 3.2, 4.1, and 5.1, with capture provenance held in the session files of September 18, 2026. The bronze transcription that every quoted plaque fact was verified against stands at Section 2 of Stanford University Is Petitioned to Retract the Headline of Its BIRTH OF THE INTERNET Plaque, and the machine-verdict series stands at Fact or Fiction?, including the September 17, 2026 edition, Fact or Fiction: One Prompt to Four AI/LLM Engines on TCP/IP and the Birth of the Internet, and Not One Engine Dates the Birth to 1973 nor TCP.

Documented corrections are welcome.