TACIT KNOWLEDGE
TACIT KNOWLEDGE
METHDO DESIGN
METHDO DESIGN
CASE STUDY
CASE STUDY
A method for surfacing what users know but can’t say — so what gets designed is built around their real, unspoken needs.
A method for surfacing what users know but can’t say — so what gets designed is built around their real, unspoken needs.
Sole designer-researcher · Research-through-Design · TCS Research × IIT Hyderabad · Mar–Sep 2025
Sole designer-researcher · Research-through-Design · TCS Research × IIT Hyderabad · Mar–Sep 2025
AT A GLANCE
Tacit knowledge - the sensory judgments, timing instincts and rules of thumb experts rely on but can’t put into words - is a blind spot for conventional user research, which depends on people explaining what they know. M3 Allegory-Driven Elicitation is the method I designed to reach it, in three layers: a spoken baseline, silent observation of the same task, then an allegory that lets people describe what they can’t state directly. A five-person pilot surfaced behavior every participant had left out of their own account, seven tacit patterns that recurred independently across strangers, and knowledge categories observation alone can’t reach. The method was accepted at ICoRD 2027 (Springer).
5/5
participants showed behaviour absent from their own verbal account
07
convergent tacit patterns across independent interviews
ICoRD 2027
accepted · peer-reviewed (Springer)
DISCOVER
SCENARIO
Imagine facing a fast bowler. The ball is on you in half a second - too fast to think - and yet you are already leaning into the shot.
Imagine facing a fast bowler. The ball is on you in half a second - too fast to think - and yet you are already leaning into the shot.
Imagine facing a fast bowler. The ball is on you in half a second - too fast to think - and yet you are already leaning into the shot.
By the time you could consciously decide, the ball would be past you. So you don’t decide - you read it: the bowler’s wrist, the seam, a flicker of the shoulder, and your feet and hands are already moving to meet a ball you haven’t fully seen. Ask a batsman how they did it and they’ll shrug: “I just watched the ball.” The skill is real, trained over years, and it runs entirely below words. Ask anyone how they do the thing they do best and you hit the same wall: “I just know.”
The most valuable expertise is the part people can’t articulate — and every standard method depends on them articulating it. The knowledge that matters most is exactly the knowledge our methods are built to miss.
The most valuable expertise is the part people can’t articulate — and every standard method depends on them articulating it. The knowledge that matters most is exactly the knowledge our methods are built to miss.
The most valuable expertise is the part people can’t articulate — and every standard method depends on them articulating it. The knowledge that matters most is exactly the knowledge our methods are built to miss.
MY ROLE
Sole designer-researcher
Sole designer-researcher
I owned every stage end to end — no team to hand off to. Design decisions, fieldwork and analysis were all mine.
I owned every stage end to end — no team to hand off to. Design decisions, fieldwork and analysis were all mine.
Method design
Study design
Fieldwork
AI-assisted analysis
First-author write-up
CONTEXT
Design research at TCS Research (TRDDC), extended into the M.Des thesis at IIT Hyderabad.
TIMEFRAME
Mar–Sep 2025, continued into the thesis.
DISCIPLINE
Design Research (Research-through-Design).
DISCOVER
RESEARCH QUESTIONS
The study set out to answer two research questions — one about the limits of conventional methods, one about whether an allegory could reach past them.
RQ1 Do conventional user-research methods actually elicit tacit knowledge — and what are their limits?
Interviews, contextual inquiry and probes all depend on people putting knowledge into words; verbal reports only reach what is in focal awareness.
RQ2 Can an allegory-based method elicit tacit cues while preserving the explicit, spoken data?
It must reach the unspoken layer without discarding the declarative baseline that gives the gap its meaning.
RQ1 frames the problem M3 addresses · RQ2 is what M3 Allegory-Driven Elicitation was built to test
TIMELINE
Eighteen months: from a research blind spot at TCS to an accepted ICoRD 2027 paper.
Mar 2025 – Aug 2026 · TCS Research → M.Des thesis
Mar 2025 – Aug 2026 · TCS Research → M.Des thesis
Problem & interviews
Mar–May 2025
• Polanyi · SECI
• 7 designer interviews
• 4 core gaps
Ideation & jury
May–Aug 2025
• Workshop 1 · 13 concepts
• Final jury
• 3-day sprint
Method: M1 → M3
Sep–Dec 2025
• M1 trial · 3 people
• M2 redesign
• M3 + Phase 1.5
5-cook pilot
Apr 2026
• P1–P5 sessions
• Field-medic allegory
Analysis & write-up
Apr–Jun 2026
• AI clustering
• Affinity → 7 patterns
• Full paper draft
ICoRD 2027 accepted
Aug 2026
• Springer LNCS
• First author
Problem & interviews
Ideation & jury
Method: M1 → M3
5-cook pilot
Analysis & write-up
ICoRD 2027 accepted
Mar–May 2025
May–Aug 2025
Sep–Dec 2025
Apr 2026
Apr–Jun 2026
Aug 2026
• Polanyi · SECI
• 7 designer interviews
• 4 core gaps
• Workshop 1 · 13 concepts
• Final jury
• 3-day sprint
• M1 trial · 3 people
• M2 redesign
• M3 + Phase 1.5
• P1–P5 sessions
• Field-medic allegory
• AI clustering
• Affinity → 7 patterns
• Full paper draft
• Springer LNCS
• First author
Mar–May 2025
May–Aug 2025
Sep–Dec 2025
Apr 2026
Apr–Jun 2026
Aug 2026
Polanyi · SECI
7 designer interviews
4 core gaps
Workshop 1 · 13 concepts
Final jury
3-day sprint
M1 trial · 3 people
M2 redesign
M3 + Phase 1.5
P1–P5 sessions
Field-medic allegory
AI clustering
Affinity → 7 patterns
Full paper draft
Springer LNCS
First author
Problem & interviews
Ideation & jury
Method: M1 → M3
5-cook pilot
Analysis & write-up
ICoRD 2027 accepted
DISCOVER
01
Where conventional research breaks down
Conventional user research collects what people can say, and systematically misses what they know but can’t put into words. It isn’t a matter of asking sharper questions: verbal reports only reach what’s in focal awareness (Ericsson & Simon), and reflection after the fact rebuilds experience from memory. Hierarchy and time pressure narrow candour further, and reading tacit signals usually falls back on researcher intuition rather than a consistent framework. The expertise that matters most is exactly the part these methods are built to miss.
Conventional methods reach focal awareness only — the tacit layer sits beneath what people can report.
THE DECISION
I treated this as a signal about the method, not the participants. The easy reads were “ask better questions” or “recruit better talkers” — but my data and the literature agreed the ceiling is structural. More interviews would only buy more surface. That reframing is what turned a research finding into a method-design brief.
DISCOVER
01
Where conventional research breaks down
Conventional methods reach focal awareness only — the tacit layer sits beneath what people can report.
Conventional user research collects what people can say, and systematically misses what they know but can’t put into words. It isn’t a matter of asking sharper questions: verbal reports only reach what’s in focal awareness (Ericsson & Simon), and reflection after the fact rebuilds experience from memory. Hierarchy and time pressure narrow candour further, and reading tacit signals usually falls back on researcher intuition rather than a consistent framework. The expertise that matters most is exactly the part these methods are built to miss.
THE DECISION
I treated this as a signal about the method, not the participants. The easy reads were “ask better questions” or “recruit better talkers” — but my data and the literature agreed the ceiling is structural. More interviews would only buy more surface. That reframing is what turned a research finding into a method-design brief.
DISCOVER
01
Where conventional research breaks down
Conventional methods reach focal awareness only — the tacit layer sits beneath what people can report.
Conventional user research collects what people can say, and systematically misses what they know but can’t put into words. It isn’t a matter of asking sharper questions: verbal reports only reach what’s in focal awareness (Ericsson & Simon), and reflection after the fact rebuilds experience from memory. Hierarchy and time pressure narrow candour further, and reading tacit signals usually falls back on researcher intuition rather than a consistent framework. The expertise that matters most is exactly the part these methods are built to miss.
THE DECISION
I treated this as a signal about the method, not the participants. The easy reads were “ask better questions” or “recruit better talkers” — but my data and the literature agreed the ceiling is structural. More interviews would only buy more surface. That reframing is what turned a research finding into a method-design brief.
DISCOVER
THE METHOD
M3 reads three layers of knowledge against each other
Asking harder doesn’t reach what people can’t say. M3 sets three layers side by side, each reaching a layer of knowledge the last can’t, and each built so it can be read against the one before it.
PHASE 1
DECLARATIVE
Spoken baseline
The fullest verbal account a person can give drawn out with neutral comic prompts before any cooking begins.
what they can say
PHASE 1.5
PROCEDURAL
Silent observation
Watch the real task in their own kitchen. Present-tense questions only “what now?”, never “why did you?”
what they do but can’t say
PHASE 2
RELATIONAL
Allegory
Retell the craft as a field medic under pressure - reaching where the skill was learned and what it means.
how they relate to it
The gap between what a cook says and what I watch them do is not noise — it’s the finding.
DISCOVER
01
Where conventional research breaks down
Conventional user research collects what people can say, and systematically misses what they know but can’t put into words. It isn’t a matter of asking sharper questions: verbal reports only reach what’s in focal awareness (Ericsson & Simon), and reflection after the fact rebuilds experience from memory. Hierarchy and time pressure narrow candour further, and reading tacit signals usually falls back on researcher intuition rather than a consistent framework. The expertise that matters most is exactly the part these methods are built to miss.
Conventional methods reach focal awareness only — the tacit layer sits beneath what people can report.
THE DECISION
I treated this as a signal about the method, not the participants. The easy reads were “ask better questions” or “recruit better talkers” — but my data and the literature agreed the ceiling is structural. More interviews would only buy more surface. That reframing is what turned a research finding into a method-design brief.
DISCOVER
02
Why I moved the study into a kitchen
The method needed a phase where I watched people work. An earlier pilot with IT professionals ran into workplace confidentiality — the real task couldn’t be recorded. So I moved the pilot to home cooking: a domain thick with tacit knowledge passed down socially, where the signature-dish task could actually be performed and filmed in someone’s own kitchen.
THE DECISION
Changing domains was a deliberate trade, not a retreat. Cooking cost me direct enterprise relevance but bought the one thing the method needed to be tested at all — an unrestricted view of the real task. I found and interviewed the five cooks myself, recruiting for variation over volume — different dishes, sources and motivations — so any pattern that still recurred couldn’t be an artefact of one background.
Five cooks chosen for maximum variation — so convergence would mean something structural, not shared history.
DISCOVER
03
Phase 1 — the best spoken account first
Each session opened with comic-based prompts showing everyday cooking tensions — estimating quantities without measuring, tasting and adjusting mid-prep, recovering from a mistake — and I recorded how the person said they handle each. This is the declarative baseline: the fullest verbal account they can give under good conditions.
THE DECISION
The baseline isn’t the data — it’s the control. A behaviour later counts as a tacit cue only if it never appeared here. Capturing the strongest possible verbal account first is what makes the later gap meaningful. I used comics rather than direct questions because a neutral scenario draws out how they’d actually act instead of a rehearsed self-description.
Comic prompts instead of direct questions — a neutral scene pulls a truer account than “how do you cook this?”
DISCOVER
04
Phase 1.5 — watch, don’t ask
Then the person cooked their signature dish in their own kitchen while I filmed and stayed out of the way. The only questions allowed were present-tense — “what are you checking right now?”, never “why did you do that?” Against the Phase 1 baseline the gaps were immediate: the cook who’d called his process “random” was running a precise sensory system — oil sizzle, the clarity of the dal, spoon resistance for texture.
THE DECISION
Present-tense questioning is the whole discipline of this phase. “Why did you” invites a story invented after the fact; “what now” catches the judgment as it happens. I kept observation and questioning separate — unlike contextual inquiry — so I could see exactly where the spoken account stops and the doing continues.
Present-tense questions only. “Why did you” rewrites the past; “what now” catches the live judgment.
DISCOVER
05
Phase 2 — the allegory
Observation shows what people do, not what it means or where it came from. So the final phase handed each cook a single allegory — a field medic working under pressure — with a codebook mapping cooking to the medic’s world. They retold their practice as the medic, translated it back to plain language, then linked it to something I’d just watched them do. Through the medic’s inherited kit, every participant named the person they’d learned from and how: “I just watched her hands for years.”
THE DECISION
I assigned one fixed allegory rather than letting people choose — an earlier version proved free choice just adds cognitive load. Field medicine was picked for structural parallel (fast assessment, judgment under pressure, improvised recovery) but no literal overlap with cooking, so it displaces people from their rehearsed story without contaminating it.
A field medic, not a chef — enough structural echo to transfer, enough distance to break the rehearsed answer.
DISCOVER
06
Analysis — AI drafts, the researcher decides
AI clusters the first pass; the researcher verifies every code. The automation is declared preliminary, not load-bearing.
Every phase was transcribed and pooled with the written allegory narratives. A custom AI agent generated a first pass of thematic clusters across the corpus; I then checked each cluster back against the source, rewrote the code definitions and ran affinity analysis by hand. Nothing entered the final analysis without my direct verification.
THE DECISION
The AI is a speed layer on the first pass, never the judge. Letting it cluster raw transcripts saved the slow mechanical sort; gating every code behind researcher review kept interpretation with a person. I state plainly in the paper that this pipeline is preliminary clustering only and isn’t independently validated.
DISCOVER
07
What the three layers surfaced together
Seven patterns; five reached full convergence across strangers — evidence the structure lives in the practice, not the individual.
Across five independently run interviews, seven tacit patterns recurred — lowering heat as the first recovery move, smell as the readiness gate, sensory timing replacing the clock — five of them in all five cooks, none of whom had any contact with each other. The allegory added a layer neither talking nor watching reached: how the skill was transmitted, the emotional weight it carried, the philosophy under it. One cook: “waiting is not empty time — waiting is the method.” Another reframed his own approach as “calculated jugaad, not random jugaad.” Every participant was surprised by a skill they’d never counted as skill.
THE DECISION
Convergence across strangers is the real evidence — if independent cooks with different teachers land on the same procedural structure, the pattern belongs to the practice, not the person. I reported the weaker results too (one pattern held in only three of five) rather than trim to a clean story.
DEFINE
02
Why I moved the study into a kitchen
The method needed a phase where I watched people work. An earlier pilot with IT professionals ran into workplace confidentiality — the real task couldn’t be recorded. So I moved the pilot to home cooking: a domain thick with tacit knowledge passed down socially, where the signature-dish task could actually be performed and filmed in someone’s own kitchen.
Five cooks chosen for maximum variation — so convergence would mean something structural, not shared history.
THE DECISION
Changing domains was a deliberate trade, not a retreat. Cooking cost me direct enterprise relevance but bought the one thing the method needed to be tested at all — an unrestricted view of the real task. I found and interviewed the five cooks myself, recruiting for variation over volume — different dishes, sources and motivations — so any pattern that still recurred couldn’t be an artefact of one background.
DEFINE
03
Phase 1 — the best spoken account first
Each session opened with comic-based prompts showing everyday cooking tensions — estimating quantities without measuring, tasting and adjusting mid-prep, recovering from a mistake — and I recorded how the person said they handle each. This is the declarative baseline: the fullest verbal account they can give under good conditions.
Comic prompts instead of direct questions — a neutral scene pulls a truer account than “how do you cook this?”
THE DECISION
The baseline isn’t the data — it’s the control. A behaviour later counts as a tacit cue only if it never appeared here. Capturing the strongest possible verbal account first is what makes the later gap meaningful. I used comics rather than direct questions because a neutral scenario draws out how they’d actually act instead of a rehearsed self-description.
DEVLOP
04
Phase 1.5 — watch, don’t ask
Then the person cooked their signature dish in their own kitchen while I filmed and stayed out of the way. The only questions allowed were present-tense — “what are you checking right now?”, never “why did you do that?” Against the Phase 1 baseline the gaps were immediate: the cook who’d called his process “random” was running a precise sensory system — oil sizzle, the clarity of the dal, spoon resistance for texture.
Present-tense questions only. “Why did you” rewrites the past; “what now” catches the live judgment.
THE DECISION
Present-tense questioning is the whole discipline of this phase. “Why did you” invites a story invented after the fact; “what now” catches the judgment as it happens. I kept observation and questioning separate — unlike contextual inquiry — so I could see exactly where the spoken account stops and the doing continues.
DEVLOP
05
Phase 2 — the allegory
Observation shows what people do, not what it means or where it came from. So the final phase handed each cook a single allegory — a field medic working under pressure — with a codebook mapping cooking to the medic’s world. They retold their practice as the medic, translated it back to plain language, then linked it to something I’d just watched them do. Through the medic’s inherited kit, every participant named the person they’d learned from and how: “I just watched her hands for years.”
A field medic, not a chef — enough structural echo to transfer, enough distance to break the rehearsed answer.
THE DECISION
I assigned one fixed allegory rather than letting people choose — an earlier version proved free choice just adds cognitive load. Field medicine was picked for structural parallel (fast assessment, judgment under pressure, improvised recovery) but no literal overlap with cooking, so it displaces people from their rehearsed story without contaminating it.
DEVLOP
06
Analysis — AI drafts, the researcher decides
Every phase was transcribed and pooled with the written allegory narratives. A custom AI agent generated a first pass of thematic clusters across the corpus; I then checked each cluster back against the source, rewrote the code definitions and ran affinity analysis by hand. Nothing entered the final analysis without my direct verification.
AI clusters the first pass; the researcher verifies every code. The automation is declared preliminary, not load-bearing.
THE DECISION
The AI is a speed layer on the first pass, never the judge. Letting it cluster raw transcripts saved the slow mechanical sort; gating every code behind researcher review kept interpretation with a person. I state plainly in the paper that this pipeline is preliminary clustering only and isn’t independently validated.
DELIVER
07
What the three layers surfaced together
Across five independently run interviews, seven tacit patterns recurred — lowering heat as the first recovery move, smell as the readiness gate, sensory timing replacing the clock — five of them in all five cooks, none of whom had any contact with each other. The allegory added a layer neither talking nor watching reached: how the skill was transmitted, the emotional weight it carried, the philosophy under it. One cook: “waiting is not empty time — waiting is the method.” Another reframed his own approach as “calculated jugaad, not random jugaad.” Every participant was surprised by a skill they’d never counted as skill.
Seven patterns; five reached full convergence across strangers — evidence the structure lives in the practice, not the individual.
THE DECISION
Convergence across strangers is the real evidence — if independent cooks with different teachers land on the same procedural structure, the pattern belongs to the practice, not the person. I reported the weaker results too (one pattern held in only three of five) rather than trim to a clean story.
DEFINE
02
Why I moved the study into a kitchen
The method needed a phase where I watched people work. An earlier pilot with IT professionals ran into workplace confidentiality — the real task couldn’t be recorded. So I moved the pilot to home cooking: a domain thick with tacit knowledge passed down socially, where the signature-dish task could actually be performed and filmed in someone’s own kitchen.
Five cooks chosen for maximum variation — so convergence would mean something structural, not shared history.
THE DECISION
Changing domains was a deliberate trade, not a retreat. Cooking cost me direct enterprise relevance but bought the one thing the method needed to be tested at all — an unrestricted view of the real task. I found and interviewed the five cooks myself, recruiting for variation over volume — different dishes, sources and motivations — so any pattern that still recurred couldn’t be an artefact of one background.
DEFINE
03
Phase 1 — the best spoken account first
Each session opened with comic-based prompts showing everyday cooking tensions — estimating quantities without measuring, tasting and adjusting mid-prep, recovering from a mistake — and I recorded how the person said they handle each. This is the declarative baseline: the fullest verbal account they can give under good conditions.
Comic prompts instead of direct questions — a neutral scene pulls a truer account than “how do you cook this?”
THE DECISION
The baseline isn’t the data — it’s the control. A behaviour later counts as a tacit cue only if it never appeared here. Capturing the strongest possible verbal account first is what makes the later gap meaningful. I used comics rather than direct questions because a neutral scenario draws out how they’d actually act instead of a rehearsed self-description.
DEVLOP
04
Phase 1.5 — watch, don’t ask
Then the person cooked their signature dish in their own kitchen while I filmed and stayed out of the way. The only questions allowed were present-tense — “what are you checking right now?”, never “why did you do that?” Against the Phase 1 baseline the gaps were immediate: the cook who’d called his process “random” was running a precise sensory system — oil sizzle, the clarity of the dal, spoon resistance for texture.
Present-tense questions only. “Why did you” rewrites the past; “what now” catches the live judgment.
THE DECISION
Present-tense questioning is the whole discipline of this phase. “Why did you” invites a story invented after the fact; “what now” catches the judgment as it happens. I kept observation and questioning separate — unlike contextual inquiry — so I could see exactly where the spoken account stops and the doing continues.
DEVLOP
05
Phase 2 — the allegory
Observation shows what people do, not what it means or where it came from. So the final phase handed each cook a single allegory — a field medic working under pressure — with a codebook mapping cooking to the medic’s world. They retold their practice as the medic, translated it back to plain language, then linked it to something I’d just watched them do. Through the medic’s inherited kit, every participant named the person they’d learned from and how: “I just watched her hands for years.”
A field medic, not a chef — enough structural echo to transfer, enough distance to break the rehearsed answer.
THE DECISION
I assigned one fixed allegory rather than letting people choose — an earlier version proved free choice just adds cognitive load. Field medicine was picked for structural parallel (fast assessment, judgment under pressure, improvised recovery) but no literal overlap with cooking, so it displaces people from their rehearsed story without contaminating it.
DEVLOP
06
Analysis — AI drafts, the researcher decides
Every phase was transcribed and pooled with the written allegory narratives. A custom AI agent generated a first pass of thematic clusters across the corpus; I then checked each cluster back against the source, rewrote the code definitions and ran affinity analysis by hand. Nothing entered the final analysis without my direct verification.
AI clusters the first pass; the researcher verifies every code. The automation is declared preliminary, not load-bearing.
THE DECISION
The AI is a speed layer on the first pass, never the judge. Letting it cluster raw transcripts saved the slow mechanical sort; gating every code behind researcher review kept interpretation with a person. I state plainly in the paper that this pipeline is preliminary clustering only and isn’t independently validated.
DELIVER
07
What the three layers surfaced together
Across five independently run interviews, seven tacit patterns recurred — lowering heat as the first recovery move, smell as the readiness gate, sensory timing replacing the clock — five of them in all five cooks, none of whom had any contact with each other. The allegory added a layer neither talking nor watching reached: how the skill was transmitted, the emotional weight it carried, the philosophy under it. One cook: “waiting is not empty time — waiting is the method.” Another reframed his own approach as “calculated jugaad, not random jugaad.” Every participant was surprised by a skill they’d never counted as skill.
Seven patterns; five reached full convergence across strangers — evidence the structure lives in the practice, not the individual.
THE DECISION
Convergence across strangers is the real evidence — if independent cooks with different teachers land on the same procedural structure, the pattern belongs to the practice, not the person. I reported the weaker results too (one pattern held in only three of five) rather than trim to a clean story.
THE ORIGIN STORYBOARD · PRESENTED AT THE JURY, 30 MAY 2025
THE ORIGIN STORYBOARD · PRESENTED AT THE JURY, 30 MAY 2025
Knowledge as an ocean you can’t grasp by force — the framing story for why tacit knowledge escapes conventional research, and what a method that respects it would do.
Knowledge as an ocean you can’t grasp by force — the framing story for why tacit knowledge escapes conventional research, and what a method that respects it would do.
THE COMICS
The artifacts behind the method
The artifacts behind the method
Two comics did real work here — one as the elicitation instrument inside the method, one as the story that pitched it.
Two comics did real work here — one as the elicitation instrument inside the method, one as the story that pitched it.
THE FIELD-MEDIC ALLEGORY · PHASE 2 INSTRUMENT
THE FIELD-MEDIC ALLEGORY · PHASE 2 INSTRUMENT
Each cook retold their craft as a field medic under pressure — same judgment, a domain with no literal overlap so the rehearsed answer breaks. Three cooking scenes, then the five medic scenes they map onto.
Each cook retold their craft as a field medic under pressure — same judgment, a domain with no literal overlap so the rehearsed answer breaks. Three cooking scenes, then the five medic scenes they map onto.
P1 · The Eyeball
P1 · The Taste Test
P1 · The Save
P2 · The Assessment
P2 · The Pulse
P2 · The Wrong Bottle
P2 · The Three Cots
P2 · The Old Kit
THE CONTRIBUTION
What transfers beyond the kitchen
What transfers beyond the kitchen
The transferable claim isn’t “use allegories.” It’s that tacit knowledge behaves as at least three interdependent layers - declarative, procedural, relational - and each needs a different instrument, arranged so each one can be read against the last. Remove any phase and the others lose what made them interpretable: the baseline gives the observation meaning, the observation gives the allegory something to anchor to. No prior method combined behavioural observation with structured allegorical displacement - that combination is the contribution. For anyone researching expertise that won’t verbalize - clinicians, senior engineers, craftspeople - the layered structure transfers even where the cooking domain doesn’t.
The transferable claim isn’t “use allegories.” It’s that tacit knowledge behaves as at least three interdependent layers - declarative, procedural, relational - and each needs a different instrument, arranged so each one can be read against the last. Remove any phase and the others lose what made them interpretable: the baseline gives the observation meaning, the observation gives the allegory something to anchor to. No prior method combined behavioural observation with structured allegorical displacement - that combination is the contribution. For anyone researching expertise that won’t verbalize - clinicians, senior engineers, craftspeople - the layered structure transfers even where the cooking domain doesn’t.
The transferable claim isn’t “use allegories.” It’s that tacit knowledge behaves as at least three interdependent layers - declarative, procedural, relational - and each needs a different instrument, arranged so each one can be read against the last. Remove any phase and the others lose what made them interpretable: the baseline gives the observation meaning, the observation gives the allegory something to anchor to. No prior method combined behavioural observation with structured allegorical displacement - that combination is the contribution. For anyone researching expertise that won’t verbalize - clinicians, senior engineers, craftspeople - the layered structure transfers even where the cooking domain doesn’t.
OUTCOME
Where it stands
Where it stands
M3 was accepted at ICoRD 2027 (Springer) as a first-author paper - a peer-reviewed, reproducible method for surfacing tacit knowledge across three layers. The pilot stands as evidence of feasibility, not generalizability, and I’ve been explicit about that: five participants, one domain, a single coder, an AI stage that’s still preliminary. What changed for me is how I read a thin interview answer - “you just get a feel for it” now reads as a pointer to a structured system I haven’t reached yet, not the end of the data. The next work is testing the method where you can’t watch the task at all.
M3 was accepted at ICoRD 2027 (Springer) as a first-author paper - a peer-reviewed, reproducible method for surfacing tacit knowledge across three layers. The pilot stands as evidence of feasibility, not generalizability, and I’ve been explicit about that: five participants, one domain, a single coder, an AI stage that’s still preliminary. What changed for me is how I read a thin interview answer - “you just get a feel for it” now reads as a pointer to a structured system I haven’t reached yet, not the end of the data. The next work is testing the method where you can’t watch the task at all.
THE RESULT
The Undocumentable, Documented.
The Undocumentable, Documented.
The Undocumentable, Documented.
Seven tacit patterns, convergent across five strangers — a reproducible, peer-reviewed method, accepted at ICoRD 2027 (Springer).
Seven tacit patterns, convergent across five strangers — a reproducible, peer-reviewed method, accepted at ICoRD 2027 (Springer).
Seven tacit patterns, convergent across five strangers — a reproducible, peer-reviewed method, accepted at ICoRD 2027 (Springer).
NEXT PROJECT
ARiA — XR assistant →
The ICWIHI-published research → product piece.