Summary

  • RFC 2007 froze the training-resource templates received by October 1996 while directing updates to a more current online catalogue maintained through a Web form, email and volunteer review.
  • Category, audience, language, cost, access, revision, modification and verification fields made each claim inspectable, but they did not prove continued availability, synchronized editions, durable quality or training results.

One entry in RFC 2007 says that a training item had been revised in September 1994, that its catalogue record was modified in January 1996, and that the record was last verified in January 1996. Those are not three spellings of one date. They belong to three different acts and, potentially, three different owners.

The resource author revised the material. A catalogue maintainer changed the record. Someone performed whatever check lay behind “last verified.” Keeping those clocks separate was the strongest feature of the Network Training Materials catalogue. It prevented a later edit to metadata from silently making old courseware look newly revised, and it prevented a content date from pretending that an access path had just been checked.

RFC 2007 was published in October 1996 as an Informational IETF document and FYI 29. Jill Foster, Margaret Isaacs and Mark Prior wrote it for the TRAINMAT working group at a moment when the Internet’s user population and service range were growing faster than trainers could produce material. Librarians, computing-service staff and subject support teams all faced the same duplication problem: each could write another guide, exercise or presentation, or they could pool pointers to work already judged useful.

The answer was a catalogue, not a curriculum. It covered user guides, resource guides, trainer guides, presentation material, workshop exercises, self-paced material and online collections. Its notes used a broad but not unlimited definition: material should facilitate training or seek to change behaviour; a command reference card alone did not qualify. An international collection should normally avoid resources tied to one service or network, though sufficiently general national or regional work could be accepted.

That editorial boundary mattered because the catalogue was selective. It aimed to collect quality training material, not everything available. Items two or more years old required careful scrutiny. A resource also needed an established way to obtain it, although it did not have to be free. These were judgement rules for inclusion. They were not a published, repeatable experiment measuring what a learner retained.

The RFC was a cut-off, not the operating catalogue

RFC 2007 called the project a “living document.” The phrase did not make the RFC text mutable. Section 5 states that the document contains completed templates received by its publication date and that online versions may be newer. Section 4 is even more explicit: because the online catalogue would be more current than the RFC version, it should be the basis for updates.

The responsibility split was practical. The RFC preserved a durable public snapshot. The Web version could update sections separately. An HTML form was the preferred route for new or revised entries; a completed template could also be emailed. The person submitting a new template became responsible for the entry, while a small international editorial board of TRAINMAT volunteers reviewed new and revised records. Updates were performed on a best-effort basis and remained subject to review.

This design produced two kinds of memory. The RFC answered: what had the catalogue received and accepted by October 1996? The online system tried to answer: what does the catalogue currently say? Neither answer should be silently substituted for the other. If a mutable page overwrites every prior state, later readers cannot reconstruct what was known at the cut-off. If the frozen RFC is treated as current inventory, old coordinates acquire an authority the document expressly denied them.

A template made claims separable

The Trainmat template did more than format prose. It separated identity, intended use, access and custody. A handle identified the entry at its source. Category, country, title, author, organization, audience level, language, keywords and description explained what the resource claimed to be and for whom. Cost and access type described the path to obtain it. Email, FTP, Gopher, Web and offline publications each had their own fields for addresses, commands, URLs, login details, file size, media type, requirements, publisher and reference.

The final fields recorded the catalogue’s own work: notes, last verification date, last modification date and the name and email of the last modifier. The form automatically allocated the handle and added the record-modification date. This made it possible to ask whether a resource’s content, access route and catalogue entry had moved together or drifted apart.

The examples show why the distinction cannot be collapsed. Some records name several languages; that proves a multilingual claim was entered, not that every edition was revised together. Some have no latest-revision date. Some list unknown cost. At least two say Record-Last-Verified-Date: Never and still appear in the published snapshot. Inclusion therefore did not require a completed verification event, and a complete-looking record did not mean every fact was present.

“Verified” needs an object and a method

RFC 2007 defines the last-verification field as the date when an item’s contents were last verified. It does not define the test, the observations retained, the duration of validity or continuous monitoring. A date can show that a check was recorded; it cannot prove that a URL worked a day later, that a mail server delivered the file, that a video remained purchasable or that required software still opened the format.

Nor does checking one edition establish agreement among language versions. Language is a catalogue field, while content revision is one date. Without edition-specific fingerprints and dates, a multilingual row cannot say whether French, German, Japanese and English copies contain the same lesson. The catalogue could accurately repeat a supplier’s statement and still lack the evidence to compare every file.

Editorial review has a similarly precise boundary. TRAINMAT volunteers decided whether an item belonged in the catalogue and applied a degree of quality control. That is real curation. It proves an acceptance decision under the working group’s criteria at that time. It does not prove that learners completed the material, changed behaviour, passed a test or produced an operational result.

The document’s honesty lies in these limits. It kept “Never,” unknown costs, empty dates and several access modes visible rather than converting absence into certainty. Its metadata did not eliminate decay. It preserved enough separation to notice decay later.

Sources and the boundary of this history

The primary record is RFC 2007, cross-checked against its RFC Editor record and IETF Datatracker entry. The IETF 27 proceedings and the RIPE RTC charter archive preserve the earlier ambition to collect, classify and reuse network-training material. RFC 2150 records RFC 2007 in the FYI series, while RFC 1175 marks the adjacent bibliography territory this article does not claim.

These sources establish a historical catalogue design and its recorded entries. They do not establish present availability, modern training practice, the quality of any named resource or a learner outcome.