AI Research NotesKnowledge Management · Tacit Knowledge · MBSE · Digital Twin · SECI · Technology Transfer · Reverse Engineering
Chapter 15 · Knowledge Management and Technology Transfer

기술을 소유한다는 것은 파일을 보관하는 일이 아니라 사람의 암묵지와 디지털 산출물을 다시 실행 가능한 능력으로 유지하는 일이다

Knowledge Management and Technology Transfer — From Tacit Knowledge and MBSE to F/A-18 Transfer and Reverse Engineering

Olivier L. de Weck · Springer · 2022 · Chapter 15

HUMANTACIT KNOW-HOWEXPLICIT ARTIFACTSCAD · CODE · REPORTBOM · SCHEMATICMODEL · PATENTMBSEDIGITAL TWINsingle sourceof truthTECHNOLOGYOWNER · ATECHNOLOGYRECIPIENT · BLEARNING BY DOINGINTERNALIZATIONSKILLS · ARTIFACTS · LICENSES · PEOPLEKNOWLEDGE IS TRANSFERRED ONLY WHEN CAPABILITY IS RECREATED

Chapter 15는 technology roadmapping의 후반부에서 자주 생기는 착각을 정면으로 다룬다. 기술 프로젝트가 끝났고 보고서·CAD·software가 repository에 들어갔다고 해서 그 기술이 조직에 남은 것은 아니다. 기술지식은 explicit artifacttacit know-how에 동시에 존재하고, 사람·문서·모델·물리적 부품·training·joint R&D를 통해 이동한다. 그래서 knowledge management의 목표는 정보를 많이 저장하는 것이 아니라 필요한 사람이 필요한 시점에 기술을 이해하고 재현하고 수정할 수 있는 capability를 보존하는 것이다. 이 장은 apprenticeship에서 MBSE/digital twin으로, knowledge repository에서 SECI spiral로, 내부 기술이전에서 국가 간 F/A-18 프로그램으로, 마지막에는 reverse engineering까지 이어지며 ‘기술지식이 어떻게 조직의 능력이 되는가’를 한 흐름으로 설명한다.

SOURCE BASIS · Technology Roadmapping and Development, Chapter 15 “Knowledge Management and Technology Transfer” · Springer Nature Switzerland AG · 2022 · DOI 10.1007/978-3-030-88346-1_15 · Chapter의 역사적 사례, 조직명, 비용과 정치·군사 관련 서술은 원서의 표현과 시점을 보존하며 현재 사실로 재검증하거나 조용히 수정하지 않는다.
Part I · §15.1

What Is Technological Knowledge? — 기술은 머릿속의 know-how와 외부 artifact 사이에 걸쳐 존재한다

Explicit–tacit와 embedded–embodied 구분은 지식관리와 기술이전의 설계를 바꾸는 출발점이다.

Explicit vs. Tacit

원서는 technological knowledge가 centuries 또는 millennia에 걸쳐 생성·curation·management·transfer되어 왔다고 전제한다. 그러나 ‘기술지식’의 실체는 단순하지 않다. Explicit knowledge는 drawings, schematics, software 같은 외부 artifact로 codified되어 보이는 지식을 뜻하고, Tacit knowledge는 사람의 mind에만 존재하는 implicit know-how를 뜻한다.

KM 학계에는 이 구분 자체에 대한 철학적 논쟁도 있다. 일부는 knowledge가 존재하려면 인간의 mind에 있어야 하므로 ‘explicit technical knowledge’가 모순된 표현이라고 본다. 다른 관점은 DNA나 technological artifact에 knowledge가 embedded될 수 있으므로 인간 밖에도 존재할 수 있다고 본다. Chapter는 두 입장을 모두 방어가능한 semantic/philosophical debate로 남겨둔다.

Embedded vs. Embodied

유사한 구분으로 embedded knowledge는 machine·artifact 안에 구조적으로 담긴 지식이고, embodied knowledge는 인간의 brain과 숙련 안에 구현된 지식이다. 같은 technology라도 artifact만 전달하면 되는 부분과 사람의 경험을 함께 전달해야 하는 부분이 다르다.

Apprenticeship: 중세의 knowledge-management system

Tacit knowledge는 apprenticeship model과 강하게 연결된다. Master는 풍부한 implicit knowledge를 갖고 있고 apprentice는 수년 동안 함께 일하며 imitation과 반복수행으로 그 know-how를 체득한다. 기본 apprenticeship 이후 다른 도시·국가를 여행하며 기술을 넓히는 tradition도 있었다. Chapter는 이를 프랑스어 savoir faire, 즉 ‘어떻게 하는지 아는 능력’과 연결한다.

독일의 journeyman model은 source 시점에 주로 남성이며 약 10%가 여성이라고 서술되고, 고향을 떠나 약 3년간 세계를 다니며 craft를 배운다고 소개된다. 인도의 guru–shishya oral tradition도 footnote에서 analogous transmission example로 언급된다.

그러나 imitation은 underlying physical law나 ‘왜 그렇게 해야 하는가’를 자동으로 설명하지 않는다. 암묵지는 강력하지만 재현성과 원리 이해를 보장하지 않는다. 이 한계 때문에 explicit representation이 중요해진다.

Eight explicit technological representations

REPRESENTATIONWHAT IT CAPTURESSOURCE EXAMPLE / NOTE
CAD model3D geometry, dimensions, tolerances, assembliesOSIRIS-REx spacecraft early design. 전통적 2D paper/mylar drawing에서 all-digital design으로 이동.
CONOPS diagramOperational events와 mission sequencePlanetary exploration mission. Timeline과 vehicles/operations를 함께 보여주며 requirements elicitation에 유용.
Software source codeFunctions, data structures, algorithmsComments와 reusable/open-source libraries가 comprehension과 reuse를 돕는다.
Physical parts / BOMProduct를 구성하는 physical artifacts 자체와 complete parts inventoryBatteries, motors, connectors. BOM은 모든 components/assemblies를 포괄적으로 기록.
Circuit diagramElectrical/electronic structure와 implementationPCB/semiconductor layout은 기업에서 매우 가치 있는 trade secret가 될 수 있다.
Conceptual modelMain function, form, architecture 관계Object Process Methodology(OPM). Source는 OPM을 ISO 19450(2015)과 연결한다.
Detailed modelParts, attributes, ports/interfaces, states, state transitionsHierarchical SysML model. 필요한 abstraction level을 적절히 선택하는 것이 중요.
Executable modelNominal dynamics와 failure modes를 포함한 behavior predictionMatlab Simscape, Modelica가 예로 제시된다.

Source는 Boeing 777을 mid-1990s의 first all-digitally designed commercial aircraft로 다시 언급한다. Explicit artifact의 범위에는 전통적인 technical reports/Design Record Bookspatents도 포함된다. Report는 detailed technical data package(TDP)의 핵심이었고, patent는 purpose, novelty, mode of operation, inventor claims를 formal하게 남긴다.

Part II · §15.1.1

From Documents to Models — MBSE는 knowledge artifact를 정적인 문서에서 실행 가능한 single source of truth로 옮긴다

Digital twin은 문서를 없애는 유행어가 아니라 element·function·attribute·state·interface를 한 모델 안에서 유지하려는 지식보존 전략이다.

MBSE의 지식관리적 의미

Model-Based Systems Engineering(MBSE)은 sophisticated/executable models를 single source of truth로 만들려는 science/technology trend로 소개된다. Ultimate goal은 physical system의 essential elements, functions, attributes, states, internal/external interfaces를 포함하는 digital twin이다.

첫 단계는 PDF나 word processor로 작성한 전통적 document-centric representation을 model-centric representation으로 대체하는 것이다. 핵심은 문서를 디지털화하는 것이 아니라 knowledge representation 자체를 executable하게 만드는 것이다.

Saab Skeldar V150, 2009

Source는 Saab Aerosystems를 Europe의 early MBSE adopter로 소개한다. 2009년 Skeldar V150 UAV project에서 MBSE를 도입했고 largely positive outcomes가 보고되었다고 서술한다. Flight software를 포함한 subsystems에 UML과 SysML을 사용해 software와 physical hardware를 함께 specify했다.

Figure 15.4는 use case, logical view, hardware view를 연결한 project representation을 보여준다. MBSE capability를 구축할 때 experienced “super users”가 workforce training과 mentoring을 주도하는 것이 중요하다고 설명한다. 이는 apprenticeship model을 digital engineering으로 번역한 형태다.

원서는 미래에는 digital twin representation만으로 technology를 이해·관리·transfer하는 것이 가능해질 수 있다고 전망하지만, 상세한 MBSE 논의는 chapter 범위를 넘어선다고 선을 긋는다.

DOCUMENT-CENTRICreports · drawings · PDFslocal files · sign-offsMODEL-CENTRICUML · SysML · OPMexecutable behaviorDIGITAL TWINelements · functions · statesinterfaces · attributesTRANSFERsimulatelearn · reuse
Part III · §15.2

Knowledge Management Architecture — repository의 크기보다 capture·index·search·governance가 연결되는 architecture가 중요하다

KM은 early 1990s에 formal academic field로 자리잡았고 creation, sharing, storing, transfer를 조직 프로세스로 다룬다.

Knowledge as a strategic asset

KM은 early 1990s부터 formal academic field로 발전했으며 individuals와 organizations 내부·사이에서 knowledge/information을 create, share, store, transfer하는 process를 다룬다. 실제 기업에서는 business management, engineering, HR, IT 가운데 하나 또는 여러 부서가 관여한다. 구현방식은 조직마다 크게 다르지만 목표는 organizational learning과 performance improvement다.

KM mechanisms

Training & mentoringFormal training, mentoring, expert interviews/recordings로 사람의 know-how를 전파한다.
RepositoriesKnowledge databases, R&D reports, standards, work instructions, domain repositories를 보존한다.
People networksSMEs와 communities of practice가 질문과 tacit context를 연결한다.
Work environmentsIntranet, concurrent design facilities(CDF), cooperative environments가 collaboration을 구조화한다.
Decision support / expert systemsStored knowledge를 실제 판단과 problem solving에 연결한다.

Figure 15.5: source → access → repository

Large aerospace corporation의 KM architecture는 knowledge sourcing과 access를 분리해 설계한다. Source side에는 portals, email sharing, metadata management, incentives, autocrawling이 있다. Autocrawling은 internal databases/document repositories뿐 아니라 Internet 같은 external sources에서도 information을 자동수집하는 방식이다.

Central access layer에는 suggested search, semantic search, visualization, topic modeling, NLP, machine learning, Wolfram과 같은 knowledge engine이 들어갈 수 있다. 핵심은 images, audio, video, documents를 consistent하게 tag/label하는 indexing engine과 ontology다. 검색기능의 고도화보다 먼저 필요한 것은 서로 다른 digital assets를 같은 언어로 분류할 수 있는 semantic discipline이다.

Governance and metrics

AREASOURCE EXAMPLESWHY IT MATTERS
Search qualityLatency, fraction of irrelevant resultsKM이 느리거나 noise가 많으면 직원은 외부 Web을 먼저 찾는다.
ParticipationContributions uploaded by community membersSourcing side의 지속성을 측정한다.
UseNumber of hits, user ratingsRepository가 실제 work에 쓰이는지 본다.
Access controlNeed-to-know restrictionsClassified/confidential knowledge를 보호한다.
Openness vs secrecyStrategic asset와 generally accessible knowledge 구분공유와 differential advantage 사이의 균형을 만든다.

What goes into the repository?

Existing repositories 위에 KM layer를 올릴 수 있다. Examples는 SME roster와 contact/expertise, Chapter 14의 scouting reports, R&D project reports, approved internal processes/standards, standard work instructions, structural design·software engineering·propulsion·aerodynamics 같은 domain knowledge databases다. External database와 연결할 수도 있으며 provider는 subscription 또는 per-search fee를 받을 수 있다.

Rollout and ROI

Legacy documents를 어디까지 넣을지는 실제로 큰 비용문제다. Source는 most important projects/programs부터 staged rollout하고 점차 less important ongoing/legacy work로 확장하는 방식을 설명한다. Comprehensive knowledge database는 time-consuming하고 expensive할 수 있으며 KM ROI에 대한 academic research나 quantitative industry information은 당시 많지 않다고 지적한다.

‘모든 것을 저장하자’는 목표는 비용과 품질을 동시에 악화시킬 수 있다. 무엇을 먼저 보존할지 priority를 정해야 한다.

NASA Engineering & Safety Center

Source는 2003 Columbia accident(STS-107) 이후 만들어진 NASA Engineering & Safety Center(NESC)를 substantial KM effort의 사례로 든다. NESC는 independent safety analysis뿐 아니라 training programs와 retiring/already-retired employees의 knowledge documentation에도 투자한다. Workforce generation gap에서 knowledge loss를 줄이는 것이 목적이다.

Part IV · SECI & Failure Modes

The Knowledge Spiral — 지식은 저장소에 정지해 있지 않고 사회화·외재화·결합·내재화를 반복하며 다시 사람에게 돌아간다

Nonaka–Takeuchi의 SECI는 technology transfer를 ‘파일 복사’가 아니라 trust와 learning-by-doing을 포함한 반복순환으로 본다.

SocializationTACIT → TACITFormal/informal relationship network를 만들고 mutual trust를 형성해 경험을 공유한다.
ExternalizationTACIT → EXPLICITDialogue, reports, models, drawings, code 등 artifact로 knowledge를 꺼내어 표현한다.
CombinationEXPLICIT → EXPLICITOpenly available artifacts를 link하고 서로 다른 knowledge를 새로운 방식으로 결합한다.
InternalizationEXPLICIT → TACITLearning by doing을 통해 acquired knowledge를 실제 숙련으로 다시 체화한다.

Cycle은 반복되며 knowledge creation/sharing이 spiral처럼 증폭된다고 source는 설명한다. 이 구조에서 repository는 중간단계일 뿐이며 끝이 아니다.

Why would employees share?

KM의 가장 큰 practical issue 중 하나는 incentive다. 직원은 다른 사람이 재현할 수 있을 정도로 technical detail을 문서화하고 upload·convert·label해야 한다. 그러나 많은 정보는 개인의 mind와 local hard drive에만 남는다.

첫 번째 barrier는 time이다. KM activity가 performance target, time allocation, job description에 들어가지 않으면 항상 urgent work에 밀린다. 더 복잡한 barrier는 job security다. 자신만 가진 implicit knowledge를 explicit하게 만들면 자신의 희소성이 사라진다고 느낄 수 있다. Source는 이 문제가 KM literature에서 충분히 말해지지 않지만 실무에서는 real issue라고 강조한다.

여러 나라·지역의 facilities가 서로 경쟁하는 firm에서는 “왜 우리 site의 know-how를 다른 site에 넘겨야 하는가?”라는 조직정치가 더해진다.

Knowledge that looks obsolete can return

Technology cycle은 fashion cycle처럼 이전 knowledge를 obsolete로 보이게 만들지만 나중에 되돌아올 수 있다. Source는 supersonic flight, hydrogen fuel cells, appliance system design을 examples로 든다. 따라서 short-term relevance만으로 archival value를 판단하면 장기 capability를 잃을 수 있다.

When employees prefer Google to the internal KM

KM system이 cumbersome하거나 존재 자체를 모르면 직원은 open Internet을 먼저 검색한다. Source는 external search provider가 search keywords의 sequence까지 통계적으로 추적할 수 있으므로 work-related query가 unintentionally trade secret를 드러낼 수 있다고 경고한다.

Technical challenges

  • Open-source vs commercial KM solution tradeoff.
  • Internal documents가 여러 storage locations와 legacy systems에 분산됨.
  • Technical documents는 Web pages와 달리 hyperlink가 적어 relevant document discovery가 어렵다.
  • Standard ontology가 없으면 classification과 topic identification이 불안정해진다.
  • 잘 indexing된 institutional repository 자체가 cyberattack의 고가치 표적이 된다.
Part V · §15.3–§15.3.2

Technology Transfer — 전달의 성공은 artifact의 이동이 아니라 recipient의 technological capability가 실제로 높아졌는가로 판단한다

Owner A의 knowledge base와 Recipient B의 need/gap 사이에는 계약·사람·training·joint R&D·incentive가 놓인다.

Where technology transfer happens

Technology transfer는 한 unit에서 다른 unit으로, 또는 legal boundary를 넘어 external organization으로 knowledge가 이동할 때 발생한다. Source가 열거하는 situations는 licensing, corporate R&D unit에서 business unit으로의 transfer, business-unit 간 transfer, foreign military sales(FMS), 새 R&D/production facility 구축, M&A 이후 knowledge consolidation, international JV에서 mandated transfer, university lab 기반 startup formation이다.

Chapter가 Merz(1990)를 따라 제시하는 정의의 핵심은 technological knowledge, skills, artifacts를 다른 individual/organization으로 전송해 receiving party의 technological capability를 높이는 것이다.

Owner A ↔ Recipient B conceptual framework

Recipient B는 먼저 technological need/gap을 분명히 해야 한다. 이 gap은 roadmapping과 scouting을 통해 식별될 수 있다. Negotiation과 contractual agreement 이후 technology owner A는 knowledge base의 relevant technology를 B에 transfer하고, B는 money, commercial rights 등 incentives를 A에 제공한다.

OWNER · Atechnological baseknowledge · skills · artifactsRECIPIENT · Bneed / gapcapability to be createdTECHNOLOGYtraining · people · models · licensesINCENTIVESmoney · commercial rights · access

Transfer instruments

  • Drawings, software, CAD files, reports 등 design artifacts의 acquisition.
  • Owner A가 Recipient B 직원에게 제공하는 formal training.
  • Technology를 실제로 정착시키는 joint R&D project.
  • 정해진 기간·scope의 key technical personnel exchange.
  • Technology를 받는 joint venture 설립.
  • Startup의 대부분 또는 모든 직원 자체를 채용하는 “acqui-hiring”.

Source는 특히 personnel exchange와 well-prescribed project charter를 가진 joint R&D가 매우 효과적이라고 설명한다. Charter에는 scope, goals, schedule, budget, acknowledged risk level이 들어가야 한다. 최종적으로 technology transfer는 국가나 기업 사이의 계약처럼 보여도 실제 전송은 humans 사이에서 일어난다.

Transfer period에는 key personnel, budget, facilities, laboratories, test ranges 등 충분한 resources를 A와 B 모두 제공해야 한다. Agreement scope를 다르게 해석하면 conflict가 발생할 수 있다.

Six modes of transfer: breadth × performance

Merz(1990)의 Fig. 15.8은 y-axis에 technology performance level, x-axis에 organization의 technology breadth 또는 해당 기술에 proficient한 people 수를 둔다.

01MaintainingRetirement/departure에도 existing technological base를 잃지 않도록 procedures, drawings, models, databases를 update·organize한다.
02TrainingKnowledge를 더 많은 people/location에 넓히지만 technology performance 자체를 반드시 높이지는 않는다.
03DeepeningExisting technology가 structure/function 측면에서 어떻게 작동하는지 understanding을 깊게 한다.
04AdvancingExisting footprint와 state of the art 범위 안에서 R&D/specialized training으로 performance level을 높인다.
05DevelopingOutside best-in-class state of the art를 조직 안으로 들여와 product/service에 infusion해 current SOA에 가까워진다.
06ResearchingOriginal research를 통해 state of the art 자체를 끌어올리며 conference/partnership 등 boundary-spanning transfer가 중요해진다.

Internal technology transfer

Internal transfer는 같은 legal entity 안에서 이루어져 contractual/legal/financial barrier가 상대적으로 작다. Large company의 business units가 technology reuse로 synergy를 만들 때 특히 중요하다. Formal system과 함께 informal personal network도 매우 중요하다고 source는 강조한다.

Instruments에는 R&D newsletters/online reports, internal conferences·seminars·science fairs, R&D databases, decision-maker technology-transfer conferences, technology portfolio/roadmap analysis, workshops/concurrent engineering sessions, transfer contact persons, central innovation-planning group 또는 CTO-level staff role, job rotation·temporary assignment·missions abroad, internal Technical and Leadership University, training, workflows/checklists, internal transfer agreements, special financing/regulations가 포함된다.

성공조건은 high feedback, 충분한 time/money/incentive, technology complexity와 compatibility에 맞는 information detail/carrier, promotions·bonus·challenging project 같은 incentives, 그리고 redundancy를 확보할 만큼 충분한 people을 뽑되 senior와 junior를 healthy하게 섞는 것이다.

External technology transfer

External transfer는 organizational/legal boundaries를 넘는다. Bell Labs의 1983 study를 인용하며 developer와 recipient가 같은 report/model/documentation semantics에 익숙할수록 impedance mismatch가 줄어들 수 있다고 설명한다. Actual transfer 전에 pre-training이 필요할 수 있다.

CEO, CTO, board 수준의 top-management support도 필요하다. Hurdles는 job-loss fear/disinterest, time allocation·technical education 부족, recipient의 not-invented-here(NIH), geographical distance와 personal contact 부족, contractual data/physical-access restrictions, classification/confidentiality로 인한 incomplete transfer다.

Support mechanisms에는 publications, continuing education, R&D databases, personnel exchange, formal contracts/TAAs, licensing, contracted R&D, joint ventures/joint R&D, technology-company acquisition, technology parks, business incubators가 포함된다. Social, environmental, economic, technological, political context도 고려해야 한다. FMS는 ITAR 같은 rules를 따를 수 있고 international JV는 해당 commercial/IP terms에 의해 규율된다.

Source framing

Chapter footnote는 1990s 이후 중국이 market access·tariff·FDI와 연계해 foreign firms의 technology transfer를 요구했고 high-speed rail·nuclear reactors 등에서 domestic technological base 형성에 활용했다고 서술하며 Young & Lan(1997)을 인용한다. 이는 원저자의 역사적·정책적 framing을 요약한 것이며 이 글에서 독립 검증한 결론으로 확대하지 않는다.

Part VI · §15.3.3

United States–Switzerland F/A-18 Transfer — 수백 개 문서를 넘긴 것이 아니라 4년에 걸쳐 사람·모델·시험·공동개발을 묶어 장기 정비·개량능력을 만들었다

1992–1997 사례는 기술이전의 성패가 training, artifact acquisition, joint R&D, 후속 internal transfer의 결합에 달려 있음을 보여준다.

Origin and background

Late 1980s Swiss Air Force는 aging F-5 Tiger replacement를 검토했고 lengthy flight competition/evaluation 뒤 McDonnell Douglas의 F/A-18을 F-16보다 선택했다. Source가 제시한 lifecycle reasons는 세 가지다.

MISSIONFlexibilityAir patrol, intercept, ground attack.
MAINTAINABILITY21 vs. 56 DMMH/FHSource의 background paragraph는 F-4 Phantom 대비 direct maintenance-man-hours per flight hour를 이 수치로 제시한다.
EVOLVABILITYSpare capacityLeading Edge Extension(LEX) 등의 capacity가 future changes를 수용할 여지를 제공한다.

1992 Swiss popular vote가 acquisition을 통과시킨 뒤 34 aircraft, 약 $2B then-year dollars 규모의 FMS contract가 U.S. Navy와 Swiss government 사이에 만들어지고, F+W Emmen과 McDonnell Douglas 사이에 subordinated TAA가 체결되었다. Military technology와 know-how transfer 조항 때문에 U.S. Department of Defense와 State Department approval이 필요했다.

From student project to liaison engineering

저자는 1991년 ETH Zurich Technology and Production Management student 때 “Technology Transfer Plan for the Swiss F/A-18 Aircraft”를 term project로 만들었고, 졸업 후 1993–1997 St. Louis의 liaison engineer로 계획 실행에 참여했다고 회고한다.

Technologies and capabilities to be absorbed

TECHNOLOGYWHY IT MATTERED
MIL-STD-1553 Digital Data BusMission computers와 avionics remote terminals가 common redundant data bus로 통신하게 한다.
Digital Display Indicators (DDIs)당시 brand-new technology로 설명되며 pilots가 color displays에서 vital functions를 control하게 한다.
Flight Software / Operational ProgramAir-to-air에서 air-to-ground mode로 aircraft를 빠르게 reconfigure하는 등 software importance가 크게 증가했다.
Weapons Replaceable Units (WRU)Maintainability, built-in test(BIT), troubleshooting/replacement, reduced ground-support equipment와 연결된다.
Structural ImprovementsLEX fence로 high-AOA vertical-tail buffeting을 줄이고 fatigue life를 3,000 h at 7.5 g에서 5,000 certified h at 9 g로 높이는 ASIP collaboration.
GE F404 EPE EngineEnhanced-performance engine integration, maintenance, repair, overhaul에서 이전세대보다 높은 technological understanding이 필요했다.
Maintainability numbers

Source는 section background에서 21 vs. 56 DMMH/FH, 뒤의 WRU 설명에서 “about 20” vs. “about 55”를 사용한다. 이 글은 이를 동일 비교의 precise/rounded 서술차이로 보존하며 하나의 숫자로 임의 통일하지 않는다.

Three transfer mechanisms

≈250
SWISS ENGINEERS / TECHNICIANS
Training2 weeks–6 months U.S. stays, 약 30 courses in aircraft subsystems, flight/ground testing, engineering. FAL workers/pilots training은 별도.
TDP+
ARTIFACT ACQUISITION
Models and dataDrawings, CAD models, static/dynamic FEM, dedicated wind-tunnel models, analysis software packages.
LDP
JOINT R&D
Low Drag PylonUnder-wing LDP 공동개발. Structural/electrical testing과 flight certification을 포함.

Program scale and outcome

4 years
1993–1997
DurationFormal transfer program completion period.
≈$40M
SOURCE-REPORTED TOTAL COST
InvestmentTechnology transfer program 전체의 source estimate.
≈500
INDIVIDUALS · BOTH SIDES
Human scaleAtlantic 양쪽 참여자 규모.

Customer objective는 aircraft fleet를 operate하는 데 그치지 않고 original manufacturer가 production을 끝낸 뒤에도 30 years or longer deep maintenance와 upgrades를 수행할 capability를 얻는 것이었다.

세 mechanism은 상호보완적이었지만 저자가 가장 성공적이라고 평가한 것은 multidisciplinary Low Drag Pylon joint R&D였다. Aerodynamics, structures, electrical/avionics, weapons integration, software, flight test를 함께 다뤘기 때문이다. 단점은 Swiss-side participants가 약 10–15명으로 작았다는 것이다.

The missing fourth mechanism: retention

저자는 overall program을 성공적으로 평가한다. Swiss Air Force와 F+W/RUAG가 여러 decades aircraft를 modify/upgrade할 deep understanding을 얻었다고 본다. 그러나 가장 큰 missed issue는 personnel attrition 이후의 internal technology transfer였다.

약 500 Swiss engineers/technicians 가운데 source에 따르면 약 절반이 year 2000까지 retire하거나 company를 떠났고 acquired knowledge도 함께 빠져나갔다. Follow-on program으로 new/young staff를 충분히 onboard하지 못한 것이 KM과 internal transfer의 결손이었다. 외부 기술이전은 recipient 내부의 세대간 재이전까지 설계하지 않으면 시간이 지나며 다시 소실된다.

Source-era political note

Chapter는 “2020 Swiss voters narrowly approved the acquisition of a new aircraft, the F-35 JSF”라고 서술한다. 이 문장은 원서의 표현을 source-era claim으로만 보존하며 본 글에서 Swiss procurement chronology를 외부자료로 재검증하거나 수정하지 않는다.

Part VII · §15.4 · Synthesis

Reverse Engineering — 문서가 없을 때 artifact에서 function과 design intent를 거꾸로 복원하는 것도 지식관리다

Forward design은 requirements → function → form으로 간다. Reverse engineering은 form → function → intent로 역행한다.

Forward design vs. reverse engineering

Normal engineering design은 requirements에서 시작해 functions를 정의하고 physical form으로 mapping한다. Reverse engineering은 physical product를 disassemble/decompose하거나 accessible software를 inspect한 뒤 form에서 function과 design intent를 추론한다. Archaeologist가 유적을 보고 과거 생활을 재구성하는 것과 비슷한 과정으로 설명된다.

Why reverse engineer?

  • Original designers가 retire/deceased하고 documentation이 없거나 부족한 legacy product를 rebuild/re-engineer하기 위해.
  • Competitor product를 system/component level에서 benchmark하기 위해.
  • Military intelligence에서 opponent technological capability를 파악하기 위해.

Chapter는 detailed methodology를 깊게 다루지는 않지만 reverse engineering을 KM과 technology transfer의 일부로 분명히 포함한다.

Akutan Zero, 1942 — artifact에서 tactics까지

Source가 든 유명 사례는 WWII 중 Aleutian island of Akutan에 거의 온전하게 crash-land한 Japanese Mitsubishi Zero fighter다. Recovered aircraft를 San Diego로 운반해 rebuild, reverse engineer, flight test하면서 rate of climb, service ceiling, turning rate, takeoff/landing distance 같은 FOM을 파악하고 weaknesses를 발견했다고 설명한다.

Source는 pilot armor 부족과 high-g pull-up에서 wing failure 가능성을 weaknesses로 서술한다. 또 이 knowledge가 Pacific fleet의 air-to-air tactics 변경과 dissemination에 영향을 주었고 mid-1943부터 kill ratio가 U.S.에 유리하게 바뀌었다고 주장하며, 일부 historians가 이 사건을 Atlantic theater의 Enigma decoding과 비교한다고 적는다.

Historical attribution

위 kill-ratio와 전략효과 해석은 Chapter 15가 제시하는 역사서술을 요약한 것이다. 이 글은 해당 군사사 인과관계를 외부자료로 독립 검증하지 않았다.

The chapter’s final logic

Organization은 먼저 자신의 technological-artifact inventory를 알아야 한다. Documents, source code, CAD files와 같은 explicit artifacts가 하나의 asset이고, employees의 heads에 있는 tacit knowledge는 그보다 더 큰 asset일 수도 있다. KM은 이 둘을 proactively capture/manage하는 practice다.

Knowledge를 조직 안에서 다음 세대로 넘기거나 외부와 공유하려면 technology transfer가 필요하다. Legal transfer에도 time, commitment, financial resources, senior-management support가 필요하다. Reverse engineering은 artifact로부터 missing technological knowledge를 다시 생성하는 complementary path다.

CHAPTER 15 SYNTHESIS지식관리의 실패는 파일을 잃어버리는 데서만 생기지 않는다. 필요한 사람이 떠나고, ontology가 끊기고, model이 실행되지 않으며, recipient가 learning-by-doing을 하지 못할 때도 기술은 사라진다. 반대로 성공적인 technology transfer는 recipient가 스스로 그 기술을 운영·수정·개량할 수 있게 되는 순간 완성된다.

Discussion questions in the chapter

Chapter는 독자에게 자신의 organization에서 knowledge가 어떻게 captured/retained되는지, formal KM system이 존재하는지, 그리고 그 effort가 실제로 worthwhile한지를 묻는다. 이 질문은 repository 도입여부가 아니라 institutional capability의 지속가능성을 평가하라는 요청으로 읽을 수 있다.

AI·Agentic R&D systems에 대한 해석

Inference: Chapter 15의 논리는 AI research organization에도 직접적인 설계질문을 던진다. Agent가 paper, code, experiment log, model checkpoint, patent, issue tracker를 모두 검색할 수 있어도 tacit rationale가 사라지면 “왜 이 실험을 버렸는지”, “어떤 실패조건이 있었는지”, “어떤 parameter가 실제로 fragile한지”를 재구성하지 못한다. 따라서 AI-native KM은 vector search만이 아니라 artifact provenance, ontology, executable models, expert identity, decision rationale, negative results를 연결해야 한다.

SECI 관점에서는 agent가 tacit knowledge를 자동으로 ‘읽는다’고 가정해서는 안 된다. Human interview와 demonstration을 통해 externalize하고, structured model/KG로 combine하고, 새로운 team이 experiment와 simulation을 수행하며 internalize하는 closed loop가 필요하다. F/A-18 사례가 보여주듯 external acquisition보다 더 중요한 병목은 시간이 지난 뒤의 follow-on internal transfer와 capability retention일 수 있다.

이 단락은 Chapter 15가 AI agent를 대상으로 실증한 결과가 아니라 원서의 KM/technology-transfer framework를 AI R&D infrastructure에 적용한 해석이다.

References · Chapter 15

참고문헌

Primary Source
01
Olivier L. de Weck. Technology Roadmapping and Development: A Quantitative Approach to the Management of Technology.
SPRINGER NATURE · 2022 · CHAPTER 15 “KNOWLEDGE MANAGEMENT AND TECHNOLOGY TRANSFER”
https://doi.org/10.1007/978-3-030-88346-1_15
Works Cited in Chapter 15
02
Henric Andersson, Erik Herzog, Gert Johansson, Olof Johansson. “Experience from introducing unified modeling language/systems modeling language at Saab Aerosystems.”
SYSTEMS ENGINEERING · 13(4) · 369–380 · 2010
03
E. U. Coke, M. E. Koether. “A study of the match between the stylistic difficulty of technical documents and the reading skills of technical personnel.”
THE BELL SYSTEM TECHNICAL JOURNAL · 62 · 1849–1864 · 1983
04
Michael Merz. “Technologie-Transfer.”
VORLESUNG TECHNOLOGIE-MANAGEMENT · ARBEITSPAPIER TM-8 · ETH ZURICH · 1990
05
R. W. Messler et al. Reverse Engineering: Mechanisms, Structures, Systems & Materials.
2014
06
Ikujiro Nonaka, Hirotaka Takeuchi. The Knowledge-Creating Company: How Japanese Companies Create the Dynamics of Innovation.
OXFORD UNIVERSITY PRESS · 1995
07
Stephen Young, Ping Lan. “Technology transfer to China through foreign direct investment.”
REGIONAL STUDIES · 31(7) · 669–679 · 1997
Web References Explicitly Listed in Chapter 15
08
Knowledge management overview.
WIKIPEDIA · CHAPTER REFERENCE LIST
https://en.wikipedia.org/wiki/Knowledge_management
09
Akutan Zero background source cited in the chapter.
WIKIPEDIA · CHAPTER FOOTNOTE
https://en.wikipedia.org/wiki/Akutan_Zero