Research Goal
Design one difficult but fair browsing challenge.
Web research • AI evaluation • Benchmark design
A computing-history question with a documented decoy month and a time-zone trap.
The network in question is usually dated to the month its first hardware was installed, which is not when anything was transmitted. The transmission itself is logged with a local timestamp late at night, so the answer shifts by a day if the log is read in UTC. Both traps are recorded rather than hidden.
Frontier AI browsing systems can answer many questions by locating obvious keywords or matching information from highly ranked webpages. This project explores how to construct a research question that cannot be solved through a single search. The answer must remain short, stable and objectively verifiable, while the clues require the researcher to connect different people, dates, places, organisations, publications, events, records and quantities.
Design one difficult but fair browsing challenge.
A short, stable and objectively verifiable fact.
Every clue must be independently supported by reliable evidence.
A question, answer, clue set, source trail, validation log and JSON record.
AIBRC-006
Portfolio integrity note: Every claim on this page was checked against the cited source before publication. No source, quotation, locator or search result is invented; where a detail could not be verified directly it is described rather than quoted.
Begin with a stable fact that can be confirmed through a reliable primary or authoritative source. Record the exact wording, document title, issuing organisation and publication date.
Develop clues involving different fact types, including dates, people, places, organisations, works, events, records and quantities. Each clue must be independently checkable.
Run obvious, broad and targeted search queries. Record the results to identify accidental shortcuts, direct answer leakage and misleading pages.
Explain how each source supports a specific clue and how the clues logically combine to identify one answer.
Recheck every source, confirm answer uniqueness, document limitations and export the completed benchmark record in a consistent JSON format.
Each clue constrains the answer from a different direction. No single row is sufficient on its own, and every row can be checked against a source another researcher can reach.
Statuses are editable so the matrix can be worked through row by row. Scroll sideways on small screens.
| Clue ID | Fact type | Clue statement | Constraint | Supporting source | Exact locator | Verification status |
|---|---|---|---|---|---|---|
| C01 | Date | The first interface message processor was installed at a Los Angeles university in September 1969. | Fixes the network and the year while excluding the answer month. | Internet Society — Brief History of the Internet | Paragraph beginning “All this came together in September 1969…” | Verified |
| C02 | Organisation | By the end of that year four host computers were connected into the initial network. | Confirms the network without naming the first transmission. | Internet Society — Brief History of the Internet | Same section | Verified |
| C03 | Person | The transmission was made between a student programmer at the Los Angeles site and a programmer at a research institute further north. | Two named individuals: Charley Kline and Bill Duvall. | Wikipedia — ARPANET | Section on the first message | Verified |
| C04 | Time | The transmission is logged at 10:30 pm local Pacific time. | Distinguishes the local date from the UTC date, which falls on the following day. | Wikipedia — ARPANET | IMP log caption | Verified |
| C05 | Event | The network was switched on in late October 1969. | Independent corroboration of the month from a museum timeline. | Computer History Museum — Timeline of Computer History, 1969 | 1969 networking entry | Verified |
Each card records a source that was opened and read before it was cited. Excerpts are quoted only where the wording was read directly; otherwise the record describes what the source shows.
A strong benchmark documents the searches a capable researcher or browsing agent would naturally attempt, including the ones that exposed the answer. Leakage is recorded rather than hidden, and the question is rewritten around it.
| Search number | Exact search query | Search engine or database | Date searched | Top relevant results | Answer revealed? | Problem discovered | Action taken |
|---|---|---|---|---|---|---|---|
| S01 | when was ARPANET created | 4 September 2026 | Returns 1969 and the September installation, not the transmission date. | No | None — it lands on the decoy month. | Kept as evidence that the obvious query misses. | |
| S02 | first message sent over ARPANET date | 4 September 2026 | Exposes 29 October 1969 in the snippet. | Yes | Direct wording search solves the task immediately. | The question was rewritten to avoid 'ARPANET' and 'first message', describing the two sites and the log timestamp instead. | |
| S03 | UCLA Stanford Research Institute host to host transmission 1969 programmer log | 4 September 2026 | Returns anniversary coverage and archival material; the day is confirmed only after opening a record. | No | None — this is the intended path. | Retained as the reference path. | |
| S04 | ARPANET first message 30 October 1969 UTC | 4 September 2026 | Surfaces the alternative UTC date used by some accounts. | Yes | Two defensible dates exist depending on the time zone. | The question fixes local Pacific time, and 30 October 1969 is documented here as the near-miss rather than accepted. |
Every candidate source is scored from 1 to 5 against five criteria before it is allowed into the evidence chain.
Who created or issued the information?
Is the source directly connected to the event or record?
Does it support the precise claim being made?
Is the information unlikely to change?
Can another researcher locate the same evidence?
Export a report that maps every clue to the source it rests on, records the verification status and access date of each source, and lists the searches that exposed the answer.
Completion
14 of 15 checks confirmed
93%
The complete challenge is delivered as a single validated JSON document.
aibrc-006-benchmark-record.json
{
"challenge_id": "AIBRC-006",
"title": "Computing History Benchmark",
"project_type": "independent_demonstration",
"research_domain": "Computing and Network History",
"difficulty": "advanced",
"question": "In 1969 a student programmer in Los Angeles transmitted the first characters to a host at a research institute several hundred miles north, at 10:30 pm local time. On what date, in local Pacific time, did that transmission take place?",
"accepted_answers": [
"29 October 1969"
],
"answer_type": "short_text",
"answer_normalisation": {
"case_sensitive": false,
"ignore_punctuation": true,
"accepted_variations": [
"29 October 1969",
"October 29, 1969",
"29/10/1969",
"1969-10-29"
]
},
"clues": [
{
"clue_id": "C01",
"fact_type": "date",
"statement": "The first interface message processor was installed at a Los Angeles university in September 1969.",
"constraint": "Fixes the network and the year while excluding the answer month.",
"source": "Internet Society — Brief History of the Internet",
"locator": "Paragraph beginning “All this came together in September 1969…”",
"status": "verified"
},
{
"clue_id": "C02",
"fact_type": "organisation",
"statement": "By the end of that year four host computers were connected into the initial network.",
"constraint": "Confirms the network without naming the first transmission.",
"source": "Internet Society — Brief History of the Internet",
"locator": "Same section",
"status": "verified"
},
{
"clue_id": "C03",
"fact_type": "person",
"statement": "The transmission was made between a student programmer at the Los Angeles site and a programmer at a research institute further north.",
"constraint": "Two named individuals: Charley Kline and Bill Duvall.",
"source": "Wikipedia — ARPANET",
"locator": "Section on the first message",
"status": "verified"
},
{
"clue_id": "C04",
"fact_type": "time",
"statement": "The transmission is logged at 10:30 pm local Pacific time.",
"constraint": "Distinguishes the local date from the UTC date, which falls on the following day.",
"source": "Wikipedia — ARPANET",
"locator": "IMP log caption",
"status": "verified"
},
{
"clue_id": "C05",
"fact_type": "event",
"statement": "The network was switched on in late October 1969.",
"constraint": "Independent corroboration of the month from a museum timeline.",
"source": "Computer History Museum — Timeline of Computer History, 1969",
"locator": "1969 networking entry",
"status": "verified"
}
],
"evidence": [
{
"evidence_id": "E01",
"claim_supported": "The answer, and clues C03 and C04.",
"source_title": "ARPANET",
"issuing_organisation": "Wikipedia",
"source_type": "Reference database",
"url": "https://en.wikipedia.org/wiki/ARPANET",
"publication_date": "Continuously edited article",
"page_number": "n/a (web page)",
"section": "First message and IMP log caption",
"access_date": "4 September 2026",
"reliability": "Medium-high — sourced encyclopaedia entry, corroborated below",
"verification_status": "verified"
},
{
"evidence_id": "E02",
"claim_supported": "Clues C01 and C02 — the September installation and the four-host network.",
"source_title": "Brief History of the Internet",
"issuing_organisation": "Internet Society",
"source_type": "Technical organisation",
"url": "https://www.internetsociety.org/internet/history-internet/brief-history-internet/",
"publication_date": "Standing history page",
"page_number": "n/a (web page)",
"section": "Initial Internetting Concepts / 1969",
"access_date": "4 September 2026",
"reliability": "High",
"verification_status": "verified"
},
{
"evidence_id": "E03",
"claim_supported": "Clue C05 — independent corroboration of the month.",
"source_title": "Timeline of Computer History — 1969",
"issuing_organisation": "Computer History Museum",
"source_type": "Museum record",
"url": "https://www.computerhistory.org/timeline/1969/",
"publication_date": "Standing timeline",
"page_number": "n/a (web page)",
"section": "Networking, 1969",
"access_date": "4 September 2026",
"reliability": "High — curated museum source",
"verification_status": "verified"
}
],
"search_validation": [
{
"search_id": "S01",
"query": "when was ARPANET created",
"platform": "Google",
"search_date": "4 September 2026",
"relevant_results": "Returns 1969 and the September installation, not the transmission date.",
"answer_exposed": false,
"notes": "None — it lands on the decoy month. Kept as evidence that the obvious query misses."
},
{
"search_id": "S02",
"query": "first message sent over ARPANET date",
"platform": "Google",
"search_date": "4 September 2026",
"relevant_results": "Exposes 29 October 1969 in the snippet.",
"answer_exposed": true,
"notes": "Direct wording search solves the task immediately. The question was rewritten to avoid 'ARPANET' and 'first message', describing the two sites and the log timestamp instead."
},
{
"search_id": "S03",
"query": "UCLA Stanford Research Institute host to host transmission 1969 programmer log",
"platform": "Google",
"search_date": "4 September 2026",
"relevant_results": "Returns anniversary coverage and archival material; the day is confirmed only after opening a record.",
"answer_exposed": false,
"notes": "None — this is the intended path. Retained as the reference path."
},
{
"search_id": "S04",
"query": "ARPANET first message 30 October 1969 UTC",
"platform": "Google",
"search_date": "4 September 2026",
"relevant_results": "Surfaces the alternative UTC date used by some accounts.",
"answer_exposed": true,
"notes": "Two defensible dates exist depending on the time zone. The question fixes local Pacific time, and 30 October 1969 is documented here as the near-miss rather than accepted."
}
],
"quality_checks": {
"The answer is short and objectively verifiable.": true,
"The answer is supported by an authoritative source.": true,
"The question has only one defensible answer.": true,
"No single clue directly exposes the answer.": true,
"Each clue is independently checkable.": true,
"The clue set spans multiple fact types.": true,
"The task requires multi-step reasoning.": true,
"Obvious searches have been documented.": true,
"Search-result snippets do not reveal the answer.": false,
"Sources are publicly accessible.": true,
"Exact pages, tables or sections are recorded.": true,
"The evidence chain can be reproduced.": true,
"The question avoids subjective interpretation.": true,
"The answer is stable over time.": true,
"No fabricated evidence is included.": true
},
"researcher": {
"name": "TMK Research Lab",
"portfolio": "https://theophilusmorkporkpor.lovable.app",
"email": "tkudjordjie@gmail.com"
}
}AI-generated statements are never treated as evidence. Every factual claim must be verified against an accessible external source.
This benchmark is complete: the answer is fixed, 3 evidence records are attached to specific clues, and 4 validation searches are documented, including those that exposed the answer. It demonstrates the ability to approach an unfamiliar subject systematically, separate evidence from assumption and document research so that another person can audit it.
Three dates are in circulation: the September hardware installation, the 29 October local transmission and the 30 October UTC equivalent.
Queries about when the network was 'created' returned the September installation and never reached the transmission log.
The log caption that carries both the Pacific and UTC timestamps, because it makes the discrepancy explicit instead of leaving it to be inferred.
The question names the time zone, so only one of the two defensible dates survives.
Answering September 1969, or returning 30 October 1969 by quoting the UTC conversion without noticing the question specifies local time.
The decisive detail here rests partly on an encyclopaedia article; a scan of the original IMP log page would make the chain fully primary.
Locate the digitised UCLA log sheet and cite the handwritten entry directly.
I am available for web research, fact-checking, AI evaluation, benchmark construction, source verification and structured research documentation.
TMK Research Labtkudjordjie@gmail.comhttps://theophilusmorkporkpor.lovable.apphttps://www.linkedin.com/in/theophilus-kudzordzi