Seven finders, a proposed benchmark, the imported counts, and the evidence gaps
This is the full working behind the first example in “A Discoverability Score Measures Whether Finders Who Don’t Know Your Name Meet You.” The subject is the author of the book, scored from public evidence only. The method gives him no allowance, and where the evidence is thin this page says so.
Seven finders could legitimately need someone with his published experience
Each finder is one that a person with his published experience could legitimately serve. Each rests on public evidence: the author biography on the press’s book pages, public talks and columns from 2012 to 2021 that the press read while researching its medical device connectivity book, and the public description of his radio show. A finder with no public support was left out. Most rows rest on his own biography or pages, and none has been rated for independent corroboration.
| Finder | The job | Public evidence |
|---|---|---|
CEO or board member of a healthcare or regulated-software company with no senior technology leader |
Put experienced technology leadership in place for six to eighteen months without a permanent hire |
Shared author biography states full-time and fractional CTO roles in startups, enterprises, and government agencies |
Product or engineering executive at a medical device maker |
Decide how to design connectivity into a device and how to prove it is safe and manageable |
Public talks and columns on device connectivity; 2012 to 2016; retrieved and read for the Medical Device Connectivity research file |
Security or risk leader at a health system or device company |
Reduce the risk that connected medical devices and their suppliers create |
Public columns on healthcare cybersecurity (2015) and a 2018 summit talk on third-party risk management for medical devices |
Product leader at a company building regulated healthcare or digital-therapeutics software |
Build and release regulated digital health software without a documentation backlog that stalls releases |
A 2021 FDAnews webinar on regulatory science for FDA digital health and therapeutics compliance |
Technology leader at a federal agency or federal contractor |
Find an experienced advisor for a government technology decision |
Shared author biography names the US Navy and the US Office of Management & Budget among past work |
Producer or organizer who needs an experienced interviewer or moderator for a healthcare technology program |
Find a moderator who knows the subject and can run a live or recorded conversation |
The Shahid Shah Show is an internet radio show released as an on-demand podcast; described in its public documentation |
Practitioner looking for a book on building or buying healthcare technology |
Find a book that explains a specific practice in enough depth to act on |
Books published under the author’s byline by the press |
The proposed benchmark is thirty questions nobody has reviewed or run
Thirty questions, derived from the Finder Specification and written without his name, any alias, or any phrase from his own pages. The weakest permitted source produced them: no interviews, sales calls, or search logs were available. The benchmark is proposed, not reviewed, not frozen, and not run. Before it is frozen, someone who knows these finders must check each question against a real input and delete any the finders would not ask. One question per class, as examples:
-
entity lookup: Who is Shahid N. Shah?
-
educational discovery: When does a healthcare startup need a fractional CTO instead of a full-time one?
-
publication discovery: Which books explain how to connect medical devices to electronic health record systems?
-
problem discovery: Our CTO just left and we are mid-way through an FDA submission. How do we keep engineering on track?
-
trust validation: How do I check that a fractional CTO has actually run engineering in a regulated company?
-
capability validation: Who has designed device-to-EHR integration architectures for regulated medical devices?
-
category discovery: What kinds of firms or individuals provide fractional CTO services to healthcare companies?
-
provider selection: Who are the best fractional CTOs for a digital health startup?
-
comparison: Fractional CTO firm or individual advisor for a regulated healthcare company: how do I decide?
-
buying intent: I need an experienced healthcare CTO for the next twelve months. Where do I start looking and what does it cost?
The imported counts support three of nine dimensions
An outside audit published counts for September 2026. It ran one shared set of 56 buyer-style questions on five engines, once each, plus by-name questions that produced 26 answers about him. It published counts and no answers, and its questions were not written from his Finder Specification. The toolkit imports the counts as aggregates and scores only what they support.
| Observation | Count |
|---|---|
By-name answers the audit judged correct |
26 of 26 |
Of those, answers that cited his own site |
25 |
Buyer-style answers that named him |
3 of 280, all on one question about podcast and radio moderators |
Fractional technology leadership slice (reviewed against the specification) |
0 of 25 answers; two other providers in 11 and 10 |
Indexable pages on his main site |
269, with a file listing the site for language models and structured data present |
Scored dimensions: Entity Recognition 100, on the audit’s judgment of correctness; Finder Specification Coverage 0 on the one reviewed slice; Cross-Engine Consistency 0, which is exact because the reviewed slice holds no mention. The other six dimensions are not scored, and the overall score is not computed. Three scored dimensions earn 5 of the 35 points they carry.
The three mentions came on a question that matches the sixth finder above. The audit did not publish whether they were substantive, so they are not scored. The five-question slice was chosen after the counts were known, which limits what its 0 means.
The findings keep the observation, the hypothesis, and the check apart
Observation. High recognition did not become problem association. He was named correctly in 26 of 26 by-name answers and in 3 of 280 unbranded answers, 0 of 25 in the slice reviewed against his specification. Hypothesis. The evidence connecting him to the problems finders describe may be thin, scattered, or not stated in finders' words. The counts cannot separate these. Would disconfirm. Engines leaving him out of answers to questions his pages answer directly, in the finder’s words, with independent corroboration. Then thin evidence is not the explanation. Experiment. Review the proposed benchmark against real finder inputs, freeze it, run it, and compare inclusion by question with the evidence chain. Before anyone writes a page, open the sources the answers cited for the two providers and record their types.
Observation. Two public pages give a company he founded different founding years, 2001 and 1997 (found in the press’s own check of his pages), and two book-specific questions produced no mention of his books on any engine. Hypothesis. Two statements of one fact may let an engine choose either, though nothing in the audit shows any answer repeating either year. His books may not be connected to him in a form engines use. The book pages have not been inspected. Experiment. Decide which year is right and state it everywhere. Inspect each book’s page, repair missing metadata, and run the book questions on the frozen benchmark.
One Evidence Gap Register entry shows the fields
- question_id: ext-fractional-cto
problem: A company needs part-time senior technology leadership
expected_association: need for fractional leadership -> experience as full-time and fractional CTO -> Shahid N. Shah -> named roles and public work -> independent mentions -> one page a finder or engine can cite
existing_evidence: The shared author biography states the experience. The pages the audit examined were not inspected for this claim.
missing_evidence: Not yet known. The first step is to inspect what exists and what the answers cited for the two providers.
source_quality: Not assessed
entity_relationship_issue: Not assessed
content_gap: Not yet known
technical_gap: Not yet known
hypothesis: Public pages may not state the service in the finder's words with named, checkable engagements. Not yet checked.
disconfirming_observation: Another practitioner with less public material is included consistently.
proposed_experiment: Inspect, then publish one capability page that states the service and cites only engagements that can be named, and retest the slice.
priority: P1
owner: Operator
status: open
retest_version: v1Three interventions are proposed, with what would show each failed
IV-1 (entity, days 1 to 30): decide the founding year and state it identically everywhere. IV-2 (relationship, days 31 to 60): give each book one canonical page naming the author, ISBN, publisher, and date, and link the author page to it. IV-3 (evidence, days 31 to 60): publish one capability page for fractional technology leadership that cites only engagements the people involved agree to have named. Beside each is the result that would show it failed. All data is in the toolkit download under examples/shahid-shah/.