The eight-second attention span — a discourse object, audited

Does the eight-second attention span claim cluster — and the goldfish comparison attached to it — hold together as a research finding when its citation chain is followed to termination, when the underlying construct ('average attention span') is examined for coherence, and when the institutional carrier's own primary data (Microsoft's EEG findings on pages 20-22) is read against the propagated claim?

1. Q1What is the source-quality rating of each link in the chain — Statistic Brain, Microsoft Canada, the secondary press outlets, the institutional consequence nodes — under the NATO Admiralty Code, with reliability and credibility rated independently per the discipline?
2. Q2Does the chain's termination at Statistic Brain (a content-mill aggregator whose listed institutional sources have explicitly disclaimed the figures) constitute structural collapse of provenance, and what charitable interpretations of the actors at each link of the chain survive that collapse?
3. Q3Does the construct 'average human attention span' admit a single seconds-measurement, and what is the relationship between that construct and (a) the three-part Sohlberg & Mateer attention framework (Sustained / Selective / Alternating) Microsoft itself uses, (b) the ACE (Attention / Connection / Encoding) EEG measurement output, and (c) the Mark-line workplace screen-switching duration?
4. Q4What is the relationship between Microsoft's page 6 slide ('human attention is dwindling') and Microsoft's page 20-22 EEG data showing differentiated outcomes (Attention declining while Encoding improves for heavy social media users in TV-only environments)? Is the report internally contradictory, and if so, what charitable reading recovers Microsoft's editorial intent without recovering the slide as supported?
5. Q5What is the implication of the Marketplace / Koehn citation (February 11, 2014, fifteen months before Microsoft's report) for Microsoft's role in the propagation chain, and what is the implication of the absence of any documented primary source predating the Marketplace citation for the upstream provenance question?
6. Q6Where does charitable interpretation recover specific actors (Microsoft's slide as honestly attributed, press deadline pressure as explanatory, Microsoft's authorial intent as dual-register), and where does it not (Statistic Brain's institutional sourcing, the propagated claim itself, the legal-scholarship and NBA-strategy institutional consequences that built strategy on demolished foundations)?
7. Q7What is the framework's structural exposure on this audit, given that the substantive citation tracing has already been performed by Schwabish (2016), Gernsbacher (c.2016), Maybin (2017), and McGinty (2017), and that the audit's contribution is structural reframing rather than novel evidence?

1. c-1aevidentiaryfactualA-1

   Pillar I — claim presented as Microsoft research vs. citation chain terminating at a non-source

   Claim A — as asserted

   Microsoft research established that human attention has dropped from 12 seconds (2000) to 8 seconds (2013), shorter than a goldfish's nine-second attention span. The figures are findings of Microsoft Canada's 2015 EEG and survey study.

   TIME (McSpadden, May 14 2015): 'according to a new study from Microsoft Corp., people now generally lose concentration after eight seconds.' National Post (May 12 2015): 'study from Microsoft Corp.' The Independent (May 13 2015): 'Microsoft surveyed 2,000 people and used electroencephalograms (EEGs).' NBC Nightly News (Lester Holt, May 15 2015): 'This according to a study from Microsoft.' The Telegraph (May 15 2015): 'Researchers surveyed 2,000 participants and studied the brain activity of 112 others.' New York Times (Egan, January 22 2016): 'A survey of Canadian media consumption by Microsoft concluded that the average attention span had fallen to eight seconds, down from 12 in the year 2000.' USA Today, NBA Commissioner Adam Silver (January 13 2017): 'A 2015 study by Microsoft revealed that the average person's attention span... is down to eight seconds — which is less than that of a goldfish.'

   Claim B — the record

   Microsoft Canada's Spring 2015 'Attention Spans' report explicitly attributes the 12/8/9-second figures to 'Statistic Brain' on page 6 of the report, in small text below the infographic. Microsoft's primary research methodology — a quantitative survey of 2,000 Canadians (Q4 2014) and an EEG neuro deep-dive of 112 Canadian respondents (December 2014, conducted by Brainsights) — measures three distinct attention types (Sustained, Selective, Alternating, derived from Sohlberg & Mateer's academic framework, page 9) using ACE (Attention, Connection, Encoding) score outputs (pages 12–13). The methodology does not generate, and the report does not claim, a single 'attention span' seconds-figure as a primary research finding. The Statistic Brain page itself, archived via Wayback Machine and verified by direct screenshot in Gernsbacher's Wisconsin packet, lists 'National Center for Biotechnology Information, U.S. National Library of Medicine, The Associated Press' as its sources without specific studies, methodologies, or sample sizes. When contacted independently by Schwabish (PolicyViz, January 2016), Gernsbacher (UW Madison, c.2016), and Maybin (BBC, March 2017), each of NCBI, NLM, and the Associated Press could find no record of underlying research supporting the figures. Statistic Brain has not responded to multiple investigator inquiries. Furthermore, the figures predate Microsoft's report by fifteen months: the same 12/8/9-second comparison appeared in a Marketplace radio segment by Kai Ryssdal on February 11 2014, recited verbatim by Harvard Business School historian Nancy Koehn ('The average American attention span in 2013 was about 8 seconds. The average attention span in 2000 was 12 seconds. And then get this kicker — the average attention of a goldfish is 9 seconds.'). Koehn herself did not respond to Gernsbacher's contact inquiry requesting source attribution. The chain terminates upstream of Statistic Brain at no documented primary source.

   Microsoft Attention Spans report, page 6 (the infographic with 'Source: Statistic brain' attribution), pages 9 (Sohlberg & Mateer framework), 11–13 (methodology and Glossary defining ACE); Statistic Brain 'Attention Span Statistics' page (Wayback captures and Gernsbacher's Wisconsin packet direct screenshot); Marketplace 'Goldfish have longer attention spans than Americans...' (Kai Ryssdal / Nancy Koehn, February 11 2014); Schwabish (PolicyViz, January 29 2016); Gernsbacher (Wisconsin Psych 532 packet, c.2016, documenting unanswered inquiries to Statistic Brain and to Koehn); Maybin (BBC News, March 10 2017); McGinty (WSJ, February 17 2017).

   Charitable interpretations — before the verdict

   • Microsoft Canada's authors may have reasonably relied on a third-party aggregator that itself claimed institutional sourcing; the duty to verify the institutional sourcing of a cited aggregator is a citation-chain audit task that consumer-research-publication conventions of the time did not require, and that Microsoft Advertising's editorial process was not designed to perform. The slide's explicit 'Source: Statistic brain' attribution is editorially honest under the conventions of the genre — Microsoft did not present the figures as their own research within the report itself.
   • The pre-Microsoft circulation through Koehn / Marketplace (February 11 2014) recovers Microsoft's role as amplifier-with-attribution rather than originator. Microsoft was operating in a pre-existing claim ecosystem in which the figures had already received mass-media circulation through a Harvard Business School historian on an American Public Media program. Charity to Microsoft on this point recovers a real fact of the propagation environment: the figures had already been laundered into respectability by an academic-titled commentator before Microsoft's slide-design step.
   • Press outlets repeating the claim with attribution to 'Microsoft research' (TIME, USA Today, the Telegraph, NBC Nightly News, the Independent, the National Post, Yahoo, the New York Times) did so under deadline pressure and in conditions where a major corporate research publication functions as a credibility heuristic. The structural pattern of marketing-research-to-press citation transfer is well-documented and predates this case. Charity to the press recovers the structural pressure but not the editorial duty: the page 6 'Source: Statistic brain' attribution is visible in the document and would have been surfaced by any verification step.
   • Statistic Brain may have had an internal aggregation process — a survey, an internal data-collection methodology, or an unattributed transfer from an upstream source the institutions named (NCBI, NLM, AP) could not retrospectively retrieve — that was never publicly disclosed. The audit cannot fully exclude this scenario without Statistic Brain's cooperation, which has not been forthcoming despite multiple investigator inquiries spanning 2016–2017. Charity here recovers the formal possibility of an undisclosed source, but not the claim's status as established research, since the absent disclosure means the figures cannot be evaluated under any reliability rubric.

   Verdict demolished — the citation chain does not survive structural audit. Charity recovers the actors' positions at multiple links but does not recover the claim. The chain terminates at a content-mill aggregator whose About Us page declares the operators are not analysts, whose listed institutional sources have explicitly disclaimed it, and whose own response to multiple investigator inquiries has been silence.

   Confidence after charity high — claim does not survive charity. The four charitable readings recover (a) Microsoft's slide attribution as honest under genre conventions, (b) Microsoft's role as amplifier of a pre-existing claim rather than originator, (c) press structural pressure as partial explanation, and (d) the formal possibility of an undisclosed Statistic Brain source. None of the four recoveries supports the claim itself: at the chain's documented terminus there is no primary research, and the institutions Statistic Brain named as sources have explicitly disclaimed any record.

   Reversal condition This contradiction would reverse if (i) Statistic Brain produces primary research methodology, sample size, and dated source documentation supporting the 12/8.25/9-second figures; (ii) NCBI, NLM, or the Associated Press locate archived material containing the figures with an authoring research team; (iii) a peer-reviewed primary source predating February 11 2014 (the Marketplace/Koehn citation date) surfaces supporting the figures; or (iv) Nancy Koehn or another pre-Microsoft amplifier produces source documentation predating their own citations. Subsequent revisions of this audit will treat substantive falsifiability material as input.

2. c-1binter docfactualA-1

   Pillar II — Microsoft's slide as accurate representation of Statistic Brain vs. dual transformation in the slide's data

   Claim A — as asserted

   Microsoft Canada's page 6 slide reports the Statistic Brain figures accurately and represents what Statistic Brain actually published at the time of the Microsoft report's compilation.

   Implicit in the slide's design — Microsoft cites 'Source: Statistic brain' without indicating any transformation of the underlying figures.

   Claim B — the record

   Statistic Brain's archived 'Attention Span Statistics' page (verified via multiple Wayback Machine captures and direct screenshot in Gernsbacher's Wisconsin packet) reports the human attention figure as '8.25 seconds' for the year 2015, not '8 seconds' for 2013 as Microsoft's slide presents. Microsoft's slide therefore performs two transformations of the Statistic Brain data: (1) rounds 8.25 down to 8, and (2) shifts the year reference from 2015 to 2013. The transformation of the year is the more consequential of the two — it relocates the figure from a then-contemporary aggregation claim to a retrospective claim about a previously-measured value. Microsoft's report (Spring 2015) cannot have been citing the Statistic Brain page in its post-2015 form (the form preserved in the verified screenshots) because the report's compilation predates the post-2015 captures. Either Microsoft cited an earlier Statistic Brain version that has since been overwritten with the '2015/8.25' figure, or Microsoft introduced both transformations during their own editorial process. The Wayback Machine preserves 924 captures of the Statistic Brain page spanning 2012-02-08 through 2025-08-18, and pre-2015 captures could resolve which scenario applies.

   Microsoft Attention Spans report, page 6 (the slide showing '12 / 2000', '8 / 2013', '9 / goldfish'); Statistic Brain 'Attention Span Statistics' page, Gernsbacher's Wisconsin packet direct screenshot (showing '2015: 8.25 seconds' / '2000: 12 seconds' / 'goldfish: 9 seconds', 'Date research was conducted: July 2, 2016'); Wayback Machine capture index (924 captures across 2012–2025).

   Charitable interpretations — before the verdict

   • Microsoft's editorial process may have rounded 8.25 → 8 for presentation simplicity, treating the decimal as marketing-irrelevant noise. Rounding numerical figures for visual presentation in marketing-research publications is a routine editorial decision and does not in itself constitute misrepresentation of the underlying source. Charity recovers the rounding step without recovering the year transformation.
   • Microsoft may have cited an earlier Statistic Brain version that displayed different numbers (e.g., a 2013 capture showing '2013: 8 seconds') and Statistic Brain subsequently updated the page to '2015: 8.25 seconds' while preserving the goldfish figure. The Wayback Machine can in principle resolve this — earlier captures (2013, 2014) would either confirm a pre-existing '2013/8' framing on the Statistic Brain page or refute it. Charity here is contingent on the Wayback verification, which the audit has not directly performed and treats as Tier-B (corroboration assessment flagged for human confirmation).
   • The transformation of both year and decimal may have been introduced by an intermediate party (a marketing infographic vendor, a creative agency, or an internal Microsoft graphic-design step) without explicit awareness by the report's analytic team. This recovers Microsoft Canada's editorial team's analytical intent but not the report-as-published, since the slide is the artefact that propagated through the press.

   Verdict partial — the decimal rounding (8.25 → 8) survives partial charity as a routine editorial step. The year transformation (2015 → 2013) does not survive charity unless an earlier Statistic Brain capture supports it; the verdict on the year transformation is therefore provisionally demolished pending Wayback verification of pre-2015 captures, with the falsifiability condition explicit and resolvable.

   Confidence after charity moderate-high — the decimal rounding (8.25 → 8) survives partial charity as a routine editorial decision; the year transformation (2015 → 2013) does not survive charity unless an earlier Statistic Brain version showed '2013/8 seconds'. The verdict differentiates these. Pending direct Wayback verification of pre-2015 captures, the verdict on the year transformation is provisional.

   Reversal condition This contradiction would reverse on the year-transformation finding if a pre-2015 Statistic Brain Wayback capture surfaces showing '2013/8 seconds' (rather than '2015/8.25 seconds') at the time of Microsoft's report compilation (~late 2014 to early 2015). The decimal-rounding finding would reverse only if Microsoft Canada documents an editorial standard requiring exact reproduction of source decimals — an unlikely standard for a marketing-research publication.

3. c-1cintra docfactualA-1

   Pillar III — Microsoft's report establishes that human attention is dwindling vs. Microsoft's report contains primary EEG data and editorial summary that contradict the dwindling framing

   Claim A — as asserted

   Microsoft's Spring 2015 report establishes, as a research finding, that human attention has dwindled from 12 seconds to 8 seconds, with the goldfish comparison serving as a biological baseline for the asserted decline.

   Microsoft Attention Spans report, page 6 — infographic headlined 'We know human attention is dwindling' with the 12/8/9-second comparison. Press repetition treating the page 6 slide as the report's primary finding (TIME, National Post, Independent, Telegraph, NBC Nightly News, NYT, USA Today, Yahoo, etc.).

   Claim B — the record

   Microsoft's report contains, on pages 4, 9–13, 17–22, and 23–24, primary research framings, framework definitions, and EEG data that contradict the page 6 slide's 'dwindling' framing. Page 4 executive summary explicitly states: 'Connected consumers are becoming better at doing more with less via shorter bursts of high attention and more efficient encoding to memory.' Page 9 cites Sohlberg & Mateer's three-part attention taxonomy (Sustained, Selective, Alternating) as 'an academic framework' for the research, distinguishing three independent attention dimensions. Pages 12–13 introduce the ACE (Attention, Connection, Encoding) score output from the EEG measurement of 112 Canadian participants conducted by Brainsights. Page 20 EEG data, for moderate/heavy social media users in TV-only environments, shows: Attention index = -9, Connection index = -5, Encoding index = +3 — i.e., attention drops in this environment but encoding (ability to commit information to memory) improves. Page 22 EEG data, for early tech adopters in TV-only environments, shows: Attention -42 vs Later adopters +21, Encoding +37 vs Later adopters -19 — a substantial encoding improvement for the group whose sustained attention has declined. Page 24 'In short' synthesis states explicitly: 'While digital lifestyles decrease sustained attention overall, it's only true in the long-term. Early adopters and heavy social media users front load their attention and have more intermittent bursts of high attention. They're better at identifying what they want/don't want to engage with and need less to process and commit things to memory.' The same document whose page 6 slide spawned the propagated 'dwindling attention' claim contains, in its primary EEG findings and its own editorial synthesis, the rebuttal data and language. The press repetition stripped this contradiction by quoting the slide and ignoring the data.

   Microsoft Attention Spans report — page 4 (executive summary), page 6 (the slide), page 9 (Sohlberg & Mateer attention model citation), pages 12–13 (Glossary defining ACE, Sustained, Selective, Alternating), pages 17–22 (EEG data on Sustained attention findings by demographic), pages 23–24 ('In short' synthesis); Gernsbacher's Wisconsin packet documenting that secondary outlets (TIME, USA Today, NYT op-ed, etc.) repeated the slide without surfacing the contradicting EEG data; NYT op-ed (Egan, January 22 2016) self-confessed: 'I tried to read the entire 54-page report, but well, you know.'

   Charitable interpretations — before the verdict

   • The page 6 slide may have been intended as introductory framing — a conversational opener for a marketing audience — rather than as a primary empirical finding of the report. Microsoft Advertising publications routinely use attention-getting opening slides without intending them as the report's empirical contribution. The internal contradiction with later EEG data is real as a published artefact but may have been editorially decoupled in authorial intent. Charity recovers Microsoft's editorial intent: the slide was framing, not finding.
   • Microsoft's report addresses two registers: a marketing/strategic register seeking summary takeaways and emotional opening hooks (page 6 slide, page 4 executive summary's headline statements) and an analytical register seeking the underlying EEG and survey data (pages 17–22). The two registers may have been editorially decoupled, with the slide serving the former and the EEG findings serving the latter. The internal contradiction is then an artefact of dual-register publication, not deliberate misrepresentation. Charity recovers the editorial structure but does not recover the slide as supported.
   • The press's role in extracting the slide and ignoring the EEG data is a separate failure that the audit attributes to TIME, USA Today, NYT, et al. — not to Microsoft. Charity to Microsoft on this point separates Microsoft's authorial role (publishing a multi-register document containing internally contradictory framings) from the press's editorial role (selecting the slide for amplification while ignoring the data that contradicts it). Both failures are real; they are distinct.

   Verdict demolished — the report-as-published contains internally contradictory framings: page 6 says 'human attention is dwindling' while pages 4, 23, and 24 explicitly state that connected consumers have improved at encoding while losing sustained focus, and pages 20–22 show the EEG data supporting the differentiated finding rather than the dwindling framing. Charity recovers Microsoft's authorial intent (dual-register publication, marketing-frame slide) but not the artefact's self-consistency. The propagated claim survives only by stripping away the rebuttal that the same document supplies.

   Confidence after charity high — the report-as-published does contain internally contradictory framings, and this is not in dispute. Charity recovers Microsoft's authorial intent (dual-register publication, marketing-frame slide rather than empirical finding) but does not recover the artefact's status as self-consistent. The press's role in stripping the contradiction is a downstream failure that does not exonerate the upstream artefact's contradiction.

   Reversal condition This contradiction would reverse if (i) the EEG indices on pages 20 and 22 are misread by the audit and the actual values support the goldfish slide's framing rather than contradict it (the audit relies on Gernsbacher's Wisconsin packet direct screenshots and the squarespace mirror PDF reading; misreading is unlikely but in principle possible); (ii) Microsoft Canada produces internal documentation showing that the page 6 slide was intended as a separate empirical claim distinct from the report's EEG findings, and that the audit's reading of 'internal contradiction' misrepresents Microsoft's editorial intent; or (iii) the textual quotations from the executive summary (page 4) and 'In short' synthesis (page 24) are misattributed by the audit, which the verbatim screenshot record makes unlikely.