Does a contemporaneous record corroborate Reynar's account that Smart Tags, SharePoint, and DAFFIE independently failed on schema design, population, and maintenance?

semantic-webontologysharepointsmart-tagsdaffiejeff-reynarcorroborationknowledge-graph

Research session following up question-corroborate-reynar-three-knowledge-projects-schema-population-maintenance, which flagged claim-reynar-three-2001-2005-knowledge-systems-died-on-schema-population-maintenance as resting entirely on Jeff Reynar's own first-person 2026 retrospective (thisandthat.chat, "The Semantic Web didn't fail because ontologies were wrong," 2026-08-07). This session re-read Reynar's full essay against its own citations and searched independently for contemporaneous (period) or third-party accounts of each of the three named projects: Microsoft Smart Tags (2001), the two 2001 SharePoints, and Google DAFFIE (2005).

Headline finding: the picture is mixed, not a clean confirm or deny. One of Reynar's own named citations — a Unisys case study he points to directly — checks out independently and corroborates the maintenance mechanism closely. A contemporaneous record of "Smart Tags" failing does exist, but it independently documents a different Smart Tags story than the one Reynar's schema-design argument rests on, and Reynar's own essay flags this distinction himself. SharePoint's 2001 split into two architecturally different products is independently corroborated by a named practitioner. DAFFIE has no independent footprint anywhere outside Reynar's own writing.

Claim: Reynar's own cited source — a December 2005 Unisys case study — independently corroborates the maintenance-failure mechanism he attributes to SharePoint deployments

verifies: question-corroborate-reynar-three-knowledge-projects-schema-population-maintenance

Reynar's essay cites a specific contemporaneous document: "In December 2005, Unisys wrote up its SharePoint deployment for Dublin Core's corporate group. The metadata belonged to a Corporate Knowledge Architect. The attributes and their details were 'maintained on a spreadsheet.' They had 'been unable to automatically populate the lists of controlled vocabularies,' and change control was 'an issue with this method.'" This document was located independently at dublincore.org (DCMI Global Corporate Circle case studies) and read directly: a one-page case study dated December 2005, submitted to the Dublin Core Global Corporate Circle, based on information from Suzanne Sheppard, "the Corporate Knowledge Architect and Enterprise Search Project Manager at Unisys," describing Unisys's own SharePoint-portal metadata practice. The document's own text matches Reynar's quotation closely: "Attributes and their details are maintained on a spreadsheet that has recently been uploaded as a portal list," and "Unisys have been unable to automatically populate the lists of controlled vocabularies... Change control is an issue with this method."

This is genuine independent corroboration of the maintenance sub-problem specifically, for a real-world SharePoint deployment, from a contemporaneous (2005), named, third-party source outside Reynar's authorship. It does not corroborate the schema design or population claims about the original 2001 Microsoft SharePoint products — it is evidence about one customer's (Unisys's) downstream deployment of SharePoint, four years after the products shipped, not about Microsoft's own design intent.

Source:

• source_url: https://www.dublincore.org/groups/corporate/case_studies/Unisys_Case_Study.pdf
• source_sha: 9cf905a794394afc14046d1f7c201e4ce863e779185336c669653010ab98ff08
• source_title: "Unisys Case Study"
• source_author: "Suzanne Sheppard (information provided by); submitted to Dublin Core Global Corporate Circle"
• source_date: "2005-12" (document dated December 2005; describes state as of September 2004)
• source_venue: "Dublin Core Metadata Initiative (DCMI), Global Corporate Circle case studies"
• source_quote: "Attributes and their details are maintained on a spreadsheet that has recently been uploaded as a portal list."
• source_quote_2: "Unisys have been unable to automatically populate the lists of controlled vocabularies."
• source_tier: 1 (primary organizational account, named author, published through an industry working group, describing the author's own organization's practice)
• source_delight: Reynar's 2026 essay cites this 2005 PDF by description only; it still resolves at its original URL twenty-one years later and says almost word for word what he quotes it as saying.

Claim: an independent, contemporaneous record of Smart Tags' 2001 failure exists — but it documents a different Smart Tags story than the one Reynar's schema-design argument is about

verifies: question-corroborate-reynar-three-knowledge-projects-schema-population-maintenance

A 2001 eWEEK article ("Microsoft Yanks Smart Tags From XP," published 2001-07-02) and a contemporaneous September 2001 post to the W3C's www-rdf-interest mailing list both independently corroborate that Microsoft's "Smart Tags" feature was a real, controversial, 2001 technology that Microsoft pulled — attributed by Jim Allchin (Microsoft's group VP for Windows and IE) to unresolved conflict between content providers' concerns and the feature's benefits.

Reynar's own essay explicitly distinguishes this from the Smart Tags story his schema-design claim rests on: "the better-known Smart Tags story is the browser one. Microsoft pulled that feature from Internet Explorer 6 in June 2001 after publishers objected to Microsoft turning their words into links of its choosing. That fight was about who controls a web page. The Office version shipped, and faded for another reason" — the Office XP Smart Tags SDK, where (per Reynar) "developers turned out not to be good at identifying those types" needed to populate the schema. The eWeek and W3C-list accounts corroborate only the browser IE6 Smart Tags controversy (an editorial-control/publisher-backlash failure), not the Office Smart Tags SDK schema-design failure that is the actual subject of Reynar's three-project argument. No contemporaneous, independent account of the Office Smart Tags SDK's schema-design failure specifically was found this session.

Sources:

• source_url: https://www.eweek.com/small-business/microsoft-yanks-smart-tags-from-xp/

• source_sha: 4d75e2324dddc09190c1c0f692a9103e17b3fd582f72970a906c3b1503d10aa3

• source_title: "Microsoft Yanks Smart Tags From XP"

• source_author: "eWEEK Editors"

• source_date: "2001-07-02"

• source_venue: "eWeek"

• source_quote: "We hadnt balanced the legitimate concerns of the content providers with the benefits we think Smart Tags can bring to users"

• source_tier: 3 (mainstream tech trade journalism, contemporaneous, named outlet, unsigned byline)

• source_url: https://lists.w3.org/Archives/Public/www-rdf-interest/2001Sep/0007.html

• source_sha: 3c87f42a81920883ee9791305732fff36a94bbbe6e144d804dd8a9f51ce63c13

• source_title: "Smart Tags are stupid ?"

• source_author: "Lei L Zhang (signed 'Allan Chang')"

• source_date: "2001-09-03"

• source_venue: "W3C www-rdf-interest mailing list"

• source_quote: "there may be legal issues around this technology (see the Wall Street Journal article http://ptech.wsj.com/archive/ptech-20010628.html for reference) which has caused MS to drop it from Windows XP"

• source_tier: 3 (contemporaneous, named, but a mailing-list post rather than edited journalism; useful mainly for confirming dating and that the episode was live industry discussion at the time)

Claim: SharePoint's 2001 split into two architecturally different products is independently corroborated by a named practitioner's firsthand account — but not framed in Reynar's schema/population/maintenance terms

verifies: question-corroborate-reynar-three-knowledge-projects-schema-population-maintenance

Reynar's essay describes Microsoft shipping "two SharePoints" in 2001 — Portal Server (taxonomist-oriented, document-and-category-first-class-objects) and Team Services (file-and-site-oriented, no particular expertise required) — and argues the schema-requiring one lost to the one that asked nothing of users. An independent account by Robert Bogue, a named, long-time SharePoint developer/consultant, writing on his own blog in 2011 about his direct experience building on SharePoint from 2001 onward, corroborates that the two 2001 products were genuinely different systems at the architecture level: "The problem was that STS and Portal Server were as different as night and day. Portal server was built around the Exchange database engine and STS was built around SQL server. Portal server was focused on documents and had versions. STS was built around lists of items and didn't have versions." Bogue also independently describes operational problems with Portal Server's Exchange-engine backend in his own 2001 deployment work ("The version of the Exchange database engine that SharePoint used had issues... we found quirk after quirk").

Bogue's account corroborates the structural divergence Reynar describes (two 2001 products, different foundations, different audiences) from an independent, named source with firsthand technical experience. It does not independently use or corroborate Reynar's specific three-part schema/population/maintenance framing — Bogue's account is about database-engine and feature-parity differences, not about who owned taxonomy design.

Source:

• source_url: https://thorprojects.com/?p=9516
• source_sha: 38939a6a5899af60c843be7b7d244d35182799f11ab6e68117c171cdb96d4f7b
• source_title: "My 10 Years in SharePoint"
• source_author: "Robert Bogue"
• source_date: "2011-12-24"
• source_venue: "Thor Projects (Robert Bogue's own blog)"
• source_quote: "Portal server was built around the Exchange database engine and STS was built around SQL server. Portal server was focused on documents and had versions. STS was built around lists of items and didn't have versions."
• source_tier: 2 (named practitioner, own venue, firsthand account of his own work — weakened by being written a decade after the events it describes, a recollection rather than a contemporaneous record; the technical-mechanism claim about database engines rests on this Tier 2 account, clearing the floor but not resting on a primary Microsoft document)

Claim: no independent or contemporaneous record of DAFFIE exists outside Jeff Reynar's own writing

verifies: question-corroborate-reynar-three-knowledge-projects-schema-population-maintenance

Multiple searches for "DAFFIE" (Reynar's name for Google's "Database of All Fact, Fiction, Information and Exaggeration," which he says powered google.com's 2005 question-answering feature) combined with "Google," "Reynar," "schema," "entity," and "cold start" returned no result independent of Reynar's own blog (thisandthat.chat). The term does not appear in any Google publication, patent, press account, or third-party retrospective found this session. The only other trace found of a related 2005 Google feature is unrelated secondary coverage of Google's general question-answering / "AutoLink" features from that period, which does not name or describe DAFFIE. The "Britney Spears problem" anecdote Reynar uses to illustrate the entity-disambiguation failure mode is also attested only in his own writing; no outside account of DAFFIE's existence, internal name, or shutdown was located.

This is the one project of the three for which the core question — is there a contemporaneous record — resolves cleanly in the negative within the scope of this session's search effort. The absence may simply reflect that DAFFIE was an internal project name that never surfaced publicly, rather than evidence the project didn't exist or didn't fail as described; absence of a found record is not evidence the record doesn't exist, only that this search did not find one.

[unverified-mechanism — needs primary]: the entire DAFFIE schema/population/maintenance narrative, including the "Britney Spears" entity-disambiguation example and the claim that it fed google.com's 2005 question-answering launch, rests solely on Reynar's first-person account (Tier 2) and does not clear the Tier 1-2 floor independently — no primary Google document was found.

Further leads

• Reynar's essay also cites "SharePoint Server 2010 added the Managed Metadata Service" and quotes Microsoft's own framing that it's for "taxonomists, librarians, and administrators" — a Microsoft primary source for this (a 2010-era TechNet/MSDN page) was not located or verified this session and would be a quick independent check on the "maintenance stayed a job title" claim. (Jeff Reynar, thisandthat.chat, 2026-08-07.)
• Reynar's essay quotes a 2003 Microsoft statement that the Office SharePoint Portal Server release "extends the capabilities of Windows SharePoint Services" — the original Microsoft press/documentation source for this phrase was not independently located this session. (Same source.)
• The Wall Street Journal article referenced by the September 2001 W3C mailing-list post (http://ptech.wsj.com/archive/ptech-20010628.html) is the actual named press source behind the Smart Tags/IE6 backlash story and was not fetched this session (likely paywalled/dead link) — worth a direct WSJ-archive check for a primary contemporaneous account with named sourcing beyond Allchin's eWeek quote.
• office365itpros.com's "SharePoint Marks Its 23rd Anniversary" (Tony Redmond, 2024-03-29) is a second named, long-tenured practitioner's SharePoint history, confirming SharePoint Portal Server 2001's release date (27 March) and describing ongoing "digital debris"/information-governance maintenance problems in SharePoint Online as of 2024 — a modern echo of the maintenance problem, not used as a core claim here since it doesn't address the 2001 schema/population question.
• Google's AutoLink toolbar feature (2005) drew a similar "who controls the page" backlash to IE6 Smart Tags, per eWeek's Matthew Hicks (2005-02-18) — a structural parallel to the Smart Tags browser controversy, not pursued further here.

Entity candidates

• Jeff Reynar — person — subject of the originating claim; built Smart Tags, watched SharePoint's 2001 split, ran Google DAFFIE/question-answering; now building a company (this+that) premised on agents solving the maintenance problem he describes.
• Suzanne Sheppard — person — Corporate Knowledge Architect and Enterprise Search Project Manager at Unisys; author of the independently-verified 2005 case study that is the strongest corroborating document found this session; her account is the earlier, more load-bearing primary record this capture's strongest claim actually rests on.
• Robert Bogue — person — named SharePoint practitioner/consultant whose 2011 firsthand account independently corroborates the architectural divergence between SharePoint Portal Server and SharePoint Team Services.
• Jim Allchin — person — Microsoft group VP for Windows and Internet Explorer in 2001; the named Microsoft executive whose contemporaneous public statements (via eWeek/WSJ) are the primary account of why the IE6 Smart Tags feature was pulled.
• Unisys — organization/concept — the company whose own SharePoint deployment is documented in the 2005 Dublin Core case study; a real-world downstream SharePoint customer independent of Microsoft.
• Dublin Core Metadata Initiative (DCMI) — organization/concept — the standards body whose Global Corporate Circle collected and published the Unisys case study; a venue worth knowing for other corporate-metadata case studies (DaimlerChrysler, Eli Lilly, Seneb Consulting are also listed on the same page).
• DAFFIE (Database of All Fact, Fiction, Information and Exaggeration) — concept/term — Google's internal name, per Reynar, for the entity-attribute knowledge base behind google.com's 2005 question-answering feature; has zero independent footprint found this session, making it itself worth flagging as an open research gap rather than an establishable fact.
• Smart Tags (Microsoft) — concept/term — ambiguous between two distinct 2001 features (the IE6 browser feature and the Office XP SDK feature); the ambiguity itself is worth a page given how easily the two get conflated in casual retelling.