Downloaded 25 times
More Related Content
Similar to High Output Research and Development - Dave Litwiller - Oct. 18, 2017
More from Dave Litwiller
High Output Research and Development - Dave Litwiller - Oct. 18, 2017
- 1. HIGH OUTPUT R&D CatapultingYour R&D Organization to the Top Echelon of Productivity and Strategic Impact OCT. 18, 2017 DAVE LITWILLER Notice: All images, designs and trademarks are the property of their respective owners
- 2. R&D PRODUCTIVITY NEEDS TORISE • KW R&D labour costs are roughly half of those in Si-Valley (salary, benefits, equity) • Lower relative staff turnover rates make the costs here even more attractive • The word is out now: • The nearly risk free R&D talent arbitrage is getting more competitive • Wages are escalating well beyond inflationary rates • I believe we are at the front end of a secular trend that will likely last for years • We need to assume that KW R&D will need to get much more productive over the years ahead to remain as competitive as the region is today • => The stakes are rising to improve R&D management skill, tools and training
- 3. PRODUCTIVITY IMPACT OF HIGHOUTPUT R&D PRACTICES Empirical spread from good to great R&D performance among head-to-head competitors in a sector: Technology Sector R&D Productivity Spread Software 2* to 3* Biotech 3* to 5* Tightly Coupled Multi-Disciplinary 4* to 10* Parts of the reasons for such a wide spread intra-sector: • The best don’t just work faster, they learn faster with each project, getting better at an increasing rate vs. competitors • The best don’t decide between having the product good or having it fast; good and fast are part of the same
- 4. KEY QUESTIONS • Whataccounts for the most productive R&D enterprises at the scale-up stage? • What role do culture, leadership and organizational design play? • How should the current health of a R&D team and structure be evaluated? • What are the highest leverage improvements which can be made over the short, medium and long term?
- 5. OVERVIEW Culture • Cultural Diagnostic Leadership Competition Organization Trainingand Onboarding Bottom-Up vs. Top-Down Technologies and R&D Organizations Project Managers and Project Management Tools, Scaffolding and Documentation Suppliers Metrics of Performance
- 6. GO FOR BROKE CULTURE •Individual high energy and group electricity • People testing their personal limits • Obsession, tempered by well grounded technical reason • Intersection of personal and group goals • Objective performance measured against global best • Widespread fervency • There are few morale problems, usually none; people are buoyed by the work • Creativity • Inside the box for the fundamentals • Outside the box selectively for breakthroughs in performance, reliability and simplicity • Design duality: For each difficult design issue, simultaneous drive to remove a performance constraint while increasing reliability
- 7. GO FOR BROKE CULTURE •Competitive • Internal: Burning desire to show who can be the best and work the fastest with quality to earn the esteem of the strongest peers • People who relish being the underdog, showing they can beat the better known and better resourced • External: • Will to win • Failure is the enemy • Clear threat, to catalyze action and overcome resistance • Simultaneously with these competitive forces, a passionate belief that success will bring to the world something that will make it a better place
- 8. GO FOR BROKE CULTURE •Confidence • Good: • Optimism, grounded in technical and market analytical reality • Attract strong peers and partners to join the effort • Positive that solutions to difficult challenges will be found • Bad: • Hubris, either unfounded or carried over from an increasingly irrelevant past • Dismissive of inconvenient but objective technical and marketplace facts • Loss of ability to self-criticize, or to act on self-criticism
- 9. GO FOR BROKE CULTURE •BHAG: Big Hairy Audacious Goal • Only way to attract and retain the best people over time • Only way to get true sign-up is if the prize professionally and personally is worth the sacrifice • Must be matched with plausibility that the goal can be achieved • Complimented with reflexive, fast iteration, mutation and small-step adaptation • Technology • Products • Processes • Deciding what not to do any longer, at about the same rate as deciding what new things to do
- 10. CULTURAL DIAGNOSTIC • Individuals pushingthemselves and each other for more, better, faster • People revel in the talent of others, and not overtly fear it • They see a priceless opportunity to learn, to see how good they can be, and practice in their disciplines to make a big impact • New arrivers feel the tension to perform, pushing much harder than they would were they left to their own devices • There’s a stream of institutional conviction that pulls newcomers in to a higher sustained pace of progress and expectation for contribution and success • There is a palpable tension that no one wants to be the person who delivered his or her piece of the puzzle late, holding back the rest of the team • >10% of R&D staff self-motivated to work to extreme time/hour levels
- 11. CULTURAL DIAGNOSTIC • There isexcitement about the rate at which concrete progress is being made, and the efficiency by which people are achieving it when working all out • This becomes a kind of narcotic the high performers cite as part of their fuel which drives the R&D effort further forward • Camaraderie of a like minded drive for success; spontaneous pulling together to achieve common ends • As well, there’s camaraderie at home for what the R&D team members are doing at work, for them to sustain a high level of dedication • Collegiality that layers on top of pulsing intra-team competition and thirst for progress just below the surface • Yin-Yang of speed of execution and frugality • The reputation for unmatched speed and scope of progress becomes the magnet for further inbound top tier R&D talent, no longer as much the legacy reputations of the people already on board
- 12. CULTURAL DIAGNOSTIC • Individual successis seen through the lens of commercial success in products and services • Not: Cool engineering for its own sake, instant gratification, science in the name of science, or patents as the primary trophies • Product reliability and quality advances are seen as being equally valuable as raw functionality advances; reliability and quality are not step-children • People are judged by the quality of their work and ability to meet integrated schedules, not by personality, title or rank • Superstars pulled in from the more individualistic and exploratory world of academia get quickly in sync with team-based activities and the need to work well with other disciplines toward focused objectives and deliverables
- 13. CULTURAL DIAGNOSTIC • People jumptoward fixing problems, both within their function, and cross-functionally, to make the whole company and customers successful • Decisions and issue ownership assignments are made quickly, not languishing or bouncing around the organization looking for a home • There is a web of voluntary, mutual responsibility • There is shared optimism, despite personal differences, that collective dedication, intelligence and expertise will prevail to overcome all problems • Optimism is self-fulfilling just as is pessimism • There is mutual respect within the team for the talents of others, and that respect is objectively well deserved, not just a social nicety
- 14. CULTURAL DIAGNOSTIC • Information sharing •The only real social obligation apart from advancing the work • Underpins trust, morale, communication, and keeping talented, ambitious people working together • To keep credibility high, information must be well curated, articulately written or spoken, and edited • Related: • There needs to be comparable disclosure of positive as failed results • Otherwise, information, learning, and intellectual rigor gets lost, capacity for error correction deteriorates, the openness intrinsic to high performance culture erodes, and avoidable risk creeps into R&D • Movement and exchange of people is intrinsic to ongoing high fidelity, low latency information exchange
- 15. CULTURAL DIAGNOSTIC • The rightpeople are in the right jobs (as obvious as this sounds) • People recruit people stronger than themselves • Managers develop management and leadership in others • Regular pruning and reseeding of the misfit • There are no self-anointed experts • The day you become one, you cease to be one; stasis sets in • Knowledge is perishable; excellence has to be constantly renewed • If self-satisfaction can happen to an individual, it can happen to a team, and the company
- 16. CULTURAL DIAGNOSTIC • There iscoherent and useful localization of company vision and mission • Little Balkanization of mission and vision • People are able to reach clear yesses, noes, and not leave too many dangling maybees • The difference between the rhetoric and polled reality of mission and values is nearly zero
- 17. CULTURAL DIAGNOSTIC • Bootleg engineeringis in evidence • People are so impassioned about the technical and market opportunity to want to push the boundaries on their own time • Empirically, there is a high correlation between project quality and investment of personal time • Bootlegging is a critical link between conformity and surreptitious journey to greater independence and boundary pushing as technical staff seasons • Added benefit: Projects done outside the lines tend to be carried out hyper efficiently, and killed off quickly if an insurmountable infeasibility is reached • Leaders curate excellent ideas bubbling up from within • Not: Late stage leadership efforts to level-up underwhelming ideas
- 18. CULTURAL DIAGNOSTIC • Leaders onlyhave to intervene after senior staff and middle managers have solved most of the difficult problems • Attention to detail is a given • Within technical sections, functions, across functions, and over time • A distinctive dialect is present • Unique to the project or company, evidence of highly compressed, efficient communication and tribal gelling • Not just a mosaic of current buzz words or sycophancy
- 19. CULTURAL DIAGNOSTIC • The lastthree promotions were all people who are exemplars of the mission, values and desired model of execution • Especially elevations to the most senior positions • Related Diagnostic Trait: Open rebellion by team against any mediocrity in management • The orthodoxies that all strong cultures develop and come to cherish remain in step with competitive reality
- 20. INTERNAL COMPETITION • Internal competitionneeds to reflect the competitive intensity of the external environment • As companies mature, the right kinds of internal competition are strong antidotes to tunnel vision and bureaucracy
- 21. INTERNAL COMPETITION • Spectrum: • Usvs. x, x=the problem to be solved, giants of the past, failure… • Us vs. the competition, where derision or indifference of competitors toward the insurgent is taken as a call to arms • Team desire to be the best; us bettering ourselves of yesterday • Sub-team desire to be the best, ratcheting others forward • Multiple critical design study teams • Competitive internal development teams, and engineering audits of projects in difficulty • Competing lines of business or business units • Internal marketplace for ideas and resources; intense advocacy from individual R&D staff at all levels to get their most promising observations and ideas acted upon • Staff and management role ambiguity (intentionally)
- 22. RECONCILING INTERNAL COMPETITION • Whencompetition is necessarily such a strong part of high output R&D organizations, much effort needs to be put into keeping competitive forces from turning overly inward and becoming harmful • Using organizationally designed intra-company competition is premature if the company has not yet developed the ability to work out its big brawls of identity, strategy and priorities • Otherwise, the big brawl just tends to atomize to smaller interconnected brawls played out in localized competitive efforts
- 23. RECONCILING INTERNAL COMPETITION • Diagnosticsfor whether internal competition is tipping over to destructive levels: • If information hoarding or other resource hoarding is reaching unhealthy extremes • If agreements are being brokered between competing people or teams for détente, to not push each other so hard or be as critical • If collaborating 3rd parties or departments outside of R&D can’t figure out which competing R&D team or faction to back, and thus hold back or slow down their work
- 24. RECONCILING INTERNAL COMPETITION • Stronger:Designed co-operation • Setting up sub-team mutual co-dependency for certain ongoing tools, work items, measurements and data • Forces people to regulate their behavior toward collaboration, rather than overly individualistic pursuits • Softer: • Information sharing (i.e. open seminars), credit sharing, award sharing, co-authorship on published papers • Pre-loading commitment to future collaboration after initial period of competition
- 25. EXTERNAL COMPETITION • Even fornewer R&D staff who do not yet have a complete sense of the market, every design engineer needs to know: • Where his or her product stands relative to the external competition • Quality • Performance • Reliability • Market share • COGS • Trends for above • R&D productivity vs. competition and reference class benchmarks for products under development • Time to market • Time to volume • Providing this frame of reference reinforces commercial success as the primary determinant of technical merit
- 26. LEADERSHIP • Deeply technical •Remarkable command at a fundamental level of the underlying science and technology • Only way to deal with fast changing, complex technology • Key to positive leadership rather than passive reviewer-ship • Management systems, tools, leadership techniques, organization, processes and culture are not a replacement for technical depth and a high rate of technical learning • Ultimate source of foresight, veracity of aims and potency of ambitions • => Committee Abdication: Non-technical leaders in R&D nearly always failover to management by consensus, with all of its attendant group decision problems in slow speed and bias • => Insufficiently technical leaders in R&D often delegate decisions to more junior technical staff and managers, and then never fully own the outcome • Sees implications of new technologies before most others do • Considers possibilities that do not occur to ordinary scientists, mathematicians and engineers
- 27. LEADERSHIP • Deeply technical(cont’d) • Strong, pragmatic sense of future customer needs • Maximum speed through OODA loop, to compete on time • Speed can only come from a deep intuitive understanding of a rapidly changing environment • Orientation is the most difficult part of OODA to achieve • Not easily placated with blithe assurances about design and technical issues • Rare ability to identify what is essential in an intricate problem, and interpret what it means • Is everywhere as the battle is raging • Real-time sense of the ebb and flow; otherwise responses and decisions are too slow
- 28. LEADERSHIP • As goodat analyzing human problems as technical ones • Both individual and relationships • Tends to know what everyone is doing, all the time, with a sincere and personal interest
- 29. LEADERSHIP • Pushes teamhard, pushes self harder • Takes the problem to heart • Deeply competitive w/ outside world • Self-reliant, demanding the same of others • Leads from the front
- 30. LEADERSHIP • Socratic • Ableto stimulate and challenge; asks good questions • Able to ask the right question at the right time • The right question can change the frame of reference of what is being observed or hypothesized, to lead to more productive avenues of resolution • Emphasis is on the unknown and the undecided • Truth and reality are the supreme criteria • Highly interactive
- 31. LEADERSHIP • Paradox ofsustaining great collaboration • Able to make good decisions on the spot, but not in an arbitrary way • Decisions are made without unduly curtailing the autonomy of individual contributors • Willing to hear others out with an open mind, even if they have to be overruled, in order to maintain their ongoing commitment
- 32. LEADERSHIP • Deep generalists,not narrow specialists • Able to see connections, interesting problems, solutions through multiple disciplines, and multiple frames of reference • Do not just fixate on one design parameter to the exclusion or detriment of others; able to compromise but with high standards • Able to see merit of nonconventional approaches more than narrow specialists • Read and network extensively, outside field as well as within • Masters of back-of-the-envelope analyses • Usually brilliant at mathematical shortcuts and estimations to make fast, useful estimates and check likely veracity of longer form calculations and models
- 33. LEADERSHIP • Knows whento get original data and analyses • Rather than rely on prevailing wisdom and tiring anecdotes • Students of history • Autodidactic • Avail themselves not just the most recent or available example, to maintain perspective • Technical judgement to steer through areas far beyond those previously known • Able to fit new pieces into the puzzle very quickly • Moreover, most great R&D programs require at least one significant contrarian bet against the prevailing view
- 34. LEADERSHIP • Has insightinto how to achieve resiliency in projects and products • Able to overcome unforeseen circumstances, without blowing time, cost or technical budgets • Counters performance or schedule limiting component bottlenecks by returning to the most basic forms of how to alter the physical phenomena at issue, and then builds back up to feasible alternatives • Distinguishes between excellence and perfection • Able to let team members hear the siren call of perfection, while driving toward timely delivery
- 35. LEADERSHIP • Has confidenceof group that the leader knows True North • Able to handle multiple peak performers and their inevitable clashes • Tends to have: • A keen sense for identifying the truly important problems, and the essence of how those problems impact the development and the organization • A very good memory • Ability to see the limit of any particular experiment or technique, and to see around the next bend to know what should likely be tried next • A cool head under pressure, to stay focused on the real problem • An inclination to tackle problems as they arise, not procrastinate; at the same time having optimism that solutions will be found down the line to problems which can’t be solved immediately
- 36. LEADERSHIP • Has Integrity •Intellectual and personal integrity is the basis of technical excellence and sustainable outperformance in challenging R&D • Wants to make sure that unwelcome truths are heard and acted upon • The term “hacker” may be worn as a kind of badge of honour in certain kinds of less technically demanding efforts • The great leaders of complex technology developments are never hackers • They may be great improvisers, and usually are, but always have a deep appreciation for how abbreviations, changes and workarounds will respond under all of the usage pressures the system will face
- 37. LEADERSHIP • High qualitynetwork • Drives talent pipeline and quality; a gravitational centre to entice others to share the dream • Sign: Inner circle of elite in the industry • Calibrated ability to evaluate talent • Deep understanding based on internal and external benchmarks of what constitutes great, good, and insufficient talent, looking far beyond superficial personality factors
- 38. TEAMS The Catalytic Powerof One • Each subsystem where breakthrough performance is required necessarily must have one person who is: • Obsessed with the challenge; sees self in a competitive race and with something big to prove • Energetic and relentless; nearly inexhaustible; thrives on pressure and deadlines, will not wilt • Overpowering curiosity to want to know the answer, quickly • At the forefront of the required skills and knowledge • Views success in this specific pursuit as career defining; usually views it as powerful tool to change the world
- 39. TEAMS One Person (cont’d) •Willing to forsake nearly all else in work-life to quickly succeed in the face of many obstacles • Exacting and disciplined experimentalist, guarding against the negative potentialities of bias and hubris that often come along with big intellects and big egos • Knows how to get just enough data from an experiment or trial to inform the next wave of work, not over-specifying data requirements from any one effort • Desire to know everything; engulfed in all facets of the problem • Able to keep an open mind to alternate possibilities, even when most signs are convergent
- 40. TEAMS One Person (cont’d) •As adept at designing and improvising tools and scaffolding as devising the core technology or product • Hands-on • Feels responsibility for technology from lab to field; lifecycle owner, not just the front end of R&D • ~0.5% of all R&D staff have this disposition and capacity • The Greyhound: The most durable form of performance pacesetting vs. managerially injected carrots, sticks and exhortation (many of which have declining or negative effectivity with long-term use)
- 41. TEAMS • The Powerof Seven • Optimal team size for a system or subsystem development as measured by productivity, speed and quality • Large enough to cross-pollenate capabilities, and allow for division of responsibility to match individual aptitudes and skills • Small enough to minimize mistaken assumptions and bad hand-offs, which delay work and compromise quality • Also, there’s nowhere for low effort or weak skills to hide • Able to talk quickly to solve difficulties fast vs. larger teams • Bounded team limit provides an antidote to “biggerism” both in the product and the enterprise, the flawed belief that the solution to the problems of big is to get bigger
- 42. TEAMS • Activating thePower of Seven • Target having aggregate ability within group to both know when and how to • Do it quick with more bold assumptions, vs. when to go slower and more methodically • Have disciplined back-up plans should inspired leaps fail • The best path to a sustained presence in the fecund middle ground
- 43. TEAMS • Activating thePower of Seven (cont’d) • Aim to have everyone be better than the others at one important thing, but pretty good at almost everything • Every person needs to move the team’s discussion and deliberation forward in his or her own way • If some of the personalities on the team are prickly, also have at least one person who is a social lubricator, who keeps everyone talking to prevent friction overheating
- 44. TEAMS • The Limitof 50 • Fifty is about the largest technical project team size that has historically achieved breakthrough performance, across ICT, aerospace and biotech • Beyond fifty people in a R&D team • The law of averages and reversion to mean performance become much more common • Individual responsibility and impact diffuses • An increasing proportion of total effort goes into co-ordination and work about work, rather than the technical work itself • The size of just the council of sub-team leaders tends to go well beyond seven • A quantum leap in formal structure becomes necessary delineating between what is centralized and what is decentralized • In general, incremental advances can be sustained with larger teams; breakthroughs though become rare
- 45. TEAMS • The Limitof 50 • Loosely coupled technologies and incremental advance projects are more forgiving for larger team sizes • Tightly coupled systems and breakthrough technical performance projects are less forgiving, with the complex and multidisciplinary nature of constituent technology interactions and design change propagation
- 46. TEAMS • Viciously limitteam size to achieve high productivity • The highest output teams are 1/10 to ¼ the size of their standard productivity peers • Small team size is both a cause and effect of high performing individuals, supervisors, methods and tools • Small teams force people to self-select who thrive on seeing themselves as responsible for the outcome, rather than expecting a lot of support
- 47. TEAMS • Viciously limitteam size to achieve high productivity • Maximum output and quality requires a single person as the combined chief engineer, product manager, and program manager for a major development program • Single head to achieve maximum alignment, agility and speed • Full responsibility for technical, financial and operational performance of product and program • The alternative of fan-out of individual roles at the top propagates multiplication throughout team, with compounding co-ordination, communication, accountability and rate of learning overheads • Additional split program leadership issue: Lowest common denominator of work ethic among leaders tends to become adopted by the whole team over time
- 48. TEAMS • Implications • Level-upfirst by building depth and breadth within individuals, not bloating teams as the primary path to skill expansion • Tool: If teams are too large, analyze what would be required to cut team size in half while maintaining output • A factor of two reduction is usually enough of a reduction for the thought experiment to get at limiting assumptions to generate useful change
- 49. TECHNICAL TRAINING • Necessaryto build distinctiveness of culture and sense of being part of something special around elite technical excellence and agility • Continuing technical education • 5-7 p.m., one night per week, with tests • Builds connections across technology disciplines • Refreshes currency of core knowledge • Provides basis for greater individual and sub-group autonomy over time, to scale • Lunch and learn for briefer or less formal matters • Regular discussions about peoples’ interesting finds in the technical literature as a capturing, filtering and dissemination mechanism • To drive learning in fast • Say one, do one, teach one
- 50. ONBOARDING YOUNG R&D STAFF •Advantage: Little reprogramming difficulty • Starting points: • Test Engineering: • Appreciate how to abstract to the minimum effective test, while not losing critical behavioural information – onramp to great design sense • Develop a first hand sense of how costly late stage design changes can be to budget and schedule, when major issues surface in integration and system test • Destigmatize test engineering, counteracting a common tendency in the R&D pecking order that test is a lesser pursuit than design engineering • To do test engineering well, it is not any less intellectually or technically challenging than design
- 51. ONBOARDING YOUNG R&D STAFF •Test Engineering (cont’d): • Superior capability in test (and debugging) is intrinsic to proficiency in design • Premature or overdone division of work between design and test corrodes productivity • Significant competitive advantage flows to those who can figure out how to test in hours or minutes what takes others days or weeks • Alternative: Develop a test rig and testing protocol to reveal critical design information which is currently absent
- 52. ONBOARDING YOUNG R&D STAFF •Starting points: • Field Engineering: • Incorporate rigors and complexities of field success into early design knowledge • Counterbalance the relative linearity and controlled nature of the internal development environment • Send the signal both out and in that the company stands behind its product • Develop early, lasting sense of the voice of the customer • Best are tough, but technically astute customers • Additional benefit: Learn the art of shaping customer expectations, not merely reacting to them • Development specs can guide incremental developments; breakthroughs though require a strong sense of the customer
- 53. ONBOARDING YOUNG R&D STAFF •Starting points: • Field Engineering (cont’d): • Learn how to push the envelope of the existing system designs, as the basis for more insightful future component and system engineering work • Successful products breed additional requirements • See who has the attitude and aptitude to hit the books or find expertise to learn about the technologies that are not part of their background to date, to identify who can grow into broader system engineering roles over time • Provide sense of how to reduce required change in customers’ adjacent technologies and workflows to expedite adoption of new technology
- 54. ONBOARDING YOUNG R&D STAFF •Starting points: • Design Engineering: • Compiling and synthesizing engineering data into concise technology reports, building basis for management of technology over time and proper technical information sharing • Such technical communication training shapes perspective of what is relevant, self-development vectors, and development of others • Preparing critiques of competitive products or projects • Code reviews and design reviews, conducted alongside strong system-level reviewers to build system sense • Troubleshooting existing systems, building ability to think about unexpected, complex issues quickly and well
- 55. ONBOARDING YOUNG R&D STAFF •Starting Points: • Production Engineering • Learn how product really gets built and tested • Appreciate any differences between the development and production environments • Learn how to quickly design custom production and test equipment, building improvisation and instrumentation engineering wherewithal • Learn how to develop and refine production processes • Form a personal network of contacts in production with whom to discuss future production-related product design issues
- 56. ONBOARDING YOUNG R&D STAFF •Daily review of design work • Illuminate unexamined assumptions • Impart the right ideas about design quality and expedite the journey up the judgment learning curve • Quickly identify further individual and systemic training improvements • Challenging first assignment • Build in the right work habits right away • See who has the mettle to be a strong contributor on a strong team for mutual benefit
- 57. ONBOARDING YOUNG R&D STAFF •Above all: Make sure young R&D staff come early on to appreciate the high cost of excess complexity and specious design assumptions in design, test, production, maintainability of design and the field • It is easy to design something complicated; it is hard to design something simple • Simplicity is usually the only path to reliability
- 58. ONBOARDING YOUNG R&D STAFF •Match staff strength with supervisory strength • Put the highest potential outperformers under the supervision of the most capable senior engineers • Have design staff be present for their designs’ early demonstrations and trials with prospects and customers • Provide first hand feedback from paying users about design fitness • Explain the central importance of patenting • As well, describe the process for invention disclosure, vetting and patent application • You don’t want to lose in the courtroom what you’ve hard won in the marketplace as time goes on
- 59. ORGANIZATIONAL SKILL EVOLUTION • VisualizationTool: • Map of: • Current organizational technical skills and activities • Current industry technical skills and activities • Future industry • Future org • Gap and differentiation analysis • Importance of looking ahead one to three years: • As business grows, rapid shifts in skill mix become more difficult to achieve • Ingrained biases • Absolute volume of new hires and training required to shift balance
- 60.
- 61. BOTTOM-UP VS. TOP- DOWNDESIGN • Top-down prevalent in engineering and computer science education today • Computing and simulation centric, allows rapid iteration, feedback, and learning in a single domain or narrow range of domains • Assumes and encourages loose coupling of components, ample design margin at interfaces, and relatively stability of those interfaces as components evolve • Most applicable in many digital hardware and software environments
- 62. BOTTOM-UP VS. TOP- DOWNDESIGN • Bottom-up has become a somewhat forgotten art in some engineering fields • Bottom-up is required when components and subsystems must be operated near their maxima, with little design margin, tight coupling and change propagation, and especially with cross-disciplinary integration • Typically, reliability, cost and space demands require that each component perform a multitude of functions • Bottom-up is most applicable in biotech, aerospace, semiconductor processes, materials science and intensely space, power or unit cost constrained ICT products
- 63. BOTTOM-UP VS. TOP- DOWNDESIGN • For highest productivity, the training, tools, and management methods change considerably between top- down and bottom-up engineering • R&D organizational dynamics follow subject system development dynamics; form follows function
- 64. BOTTOM-UP VS. TOP- DOWNDESIGN • R&D Organizations Oriented for Top-down Design • Easier to add individual contributors quickly working on the core IP base • New arrivers only need to understand part of the system • The effects of change from mistakes or inappropriate assumptions are easier to isolate and fix • Short training cycles to reach productivity • Less technical supervision required • Limits: • System performance = Sum of parts • Complexity explosion at top of organization and system under development if not done right • Unexpected surprises of tight coupling can require extensive system redesign if any one component has to change
- 65. BOTTOM-UP VS. TOP- DOWNDESIGN • R&D Organizations Oriented for Bottom-up Design • Longer learning curve to appreciate and respect the cross-functional implications of design choices; more methodical technical training req’d • The essential R&D skills are to anticipate the unknown, plan for the unforeseen, and to bring together many areas which had formerly been of ignorance or only partially understood • More intensive, active and cross-functional technical supervision is needed • The magic is in the relationships between components, and the ability of components and subsystems to do double and triple duty contributing to overall system requirements • This is the path to design elegance and resiliency • There are mirror issues for people • Upside: System performance can be >> sum of parts with emergent properties and advantageous sub-system interactions, better compressibility of functionality, and easier scaling complexity • Typically: There are opportunities for much stronger IP protection potential because of cross-functional and multi-functional insights
- 66. BOTTOM-UP VS. TOP- DOWN •Tightly integrated, bottom-up developments: • The guiding technical expert is the project manager • The leader cross-pollenates insights and new findings among sub-teams and individual contributors • Daily R&D contact with manufacturing and field • T-shaped training for technical staff, between two and ten related functional disciplines typically • Personnel need to be comfortable in an environment where individual responsibility and formal authority are not fully matched • Fluid communication, co-location and personnel exchange between component design, sub-system, system, test groups and production engineering • Elevated importance of continuity of core team members • Increased importance for a regular flow of new product development, not boom-bust cycles with long fallow periods
- 67. SET-BASED ENGINEERING • People considerseveral design alternatives during the early stages of development • Start a bit slower, to finish much faster • Forces the issue of being able to do rapid design studies, developing deep technical domain expertise • Keeps options and minds open longer than with early lock-in on a single preferred solution, providing better adaptability to downstream internal and external design changes • Expands scope of solutions with possibilities for combined techniques • Provides warm alternatives if leading solution turns out to have greater difficulties as development progresses than were anticipated • Challenge • In the rush to get going, people can want to select a presumed winning approach too early • This often turns out to be a case of being penny wise but pound foolish for time to market schedule impact, overall development cost and product fitness
- 68. SET-BASED ENGINEERING • Heuristic • 5-6initial plausible concepts for rapid design studies • 2-3 for comprehensive analysis • 1-2 for full development implementation • Scenarios • Lower budget: Red and Blue team task forces to competitively evaluate top two design alternatives at least to the critical design stage • Where schedule trumps budget: If there is a choice between two methods, one of which is good and the other looks promising, then build both
- 69. TECHNICAL MERITOCRACY • Expect anddemand that everyone, regardless of age or seniority, be able to argue his or her point on technical grounds • Win or lose, everyone takes satisfaction in having been heard, the opportunity to argue directly for what (s)he feels is right, and having received serious attention • Drives out technically lax or superficial thinking • Reinforces sense of individual accountability and owning the outcome • Without technical merit as the overriding determinant, many more error prone subjective and changing decision influences inevitably take over
- 70. TECHNICAL MERITOCRACY • Instill thatthe ability to do breakthrough design comes down to being able to see both bottom-up and top-down alternatives to traditional architectures which break through prevailing performance and cost limits • A very good starting point is deep understanding of the elemental and integrated limits of current technology • Any tolerance for superficial understanding of prevailing design challenges is a formula for mediocrity in R&D productivity • The only way to come up with novel solutions is to ask the most fundamental questions and then answer them in new ways
- 71. PROJECT MANAGERS • Completeresponsibility for a project has to be given to a person, not an organization • Introducing project managers as a matrix overlaying functional departments is always a risk if it diffuses responsibility for project delivery from technical managers who control resources and resource development • The more complex the technology, the more novel the system, and the faster the development program, the more technical the project manager needs to be • Heuristic: At least as technical as a senior design engineer • There is really no other way to infuse an attitude of self-reliance, responsibility and technically astute risk taking • Non- or weakly technical administrators as project managers increase information processing batch sizes and introduce waiting
- 72. PROJECT MANAGERS • Administrativeproject managers are only suitable • In projects where extensive bulldogging of low tech contractors or customer deployment resources is required, and, • In technically light application layer software development with a lot of flexibility between scope and schedule
- 73. PROJECT MANAGERS • Indevelopment projects of greater technical depth and complexity, project managers need to be as technically aware, inquisitive and insightful as engineering managers • As previously described, they often should be one and the same, not split roles, especially in the highest speed and complexity developments • Project manager authority should flow exclusively from technical capability and breadth • If project managers are used separately from functional leaders, good ones provide full redundancy of technical oversight with engineering managers to make sure that important issues get identified early and acted upon
- 74. PROJECT MANAGERS • Ifthey’re used as separate entities from technical leadership, the focus should be on continuity, integration, and re-integration of technical and market information transfer as development moves phases from R to D to E to Production to Field to augment the formal reporting structure • Project managers should not be used for resource prioritization or the mainline of personnel evaluation
- 75. PROJECT MANAGEMENT TOOLS • Documentaryproject management tools are at best a support, and never a replacement for, informed, attentive technical project leadership • Ignore any reverent buzzwords of project management tools or models in vogue at any moment as elixirs • Ongoing face-to-face technical interaction and ownership is the mainstay of managing complex R&D programs well • R&D projects that fall behind, stay behind • Schedule recovery is a rarity after early phase delays • Infuse the team with the necessity of hitting the early internal milestones at full scope as vigorously as the more externally visible late-stage milestones • Marching army costs usually predominate total development cost. The leading way to control development cost is to hit the schedule
- 76. PROJECT MANAGEMENT TOOLS • Everythingpossible shall be done to save time • Time is the ultimate perishable commodity • Re-use existing technologies to the greatest extent possible • Outset thought experiment: Imagine the development program is in schedule or cost overrun delivery danger late in development because of each component • Would you be willing to substitute a readily available alternative component, at least as a bridge solution, to get to market? • Notional Limit: • 30% to 40% new technology in each generational advance of the overall system for tightly coupled technologies • Higher is possible for loosely coupled
- 77. PROJECT MANAGEMENT TOOLS • Red-Yellow-GreenTask Status Dashboards • Only let yellow be used to mark transitions from G ↔ R for one week max on individual work items • Forces action to recover, or acknowledge late • Otherwise, ambiguity takes over, first with tasks, then the program
- 78. PROJECT MANAGEMENT TOOLS • Co-locatedevelopment and production • This alone is usually worth a ~25% increase in R&D productivity in time to volume • Goal: Initial conception to pilot production in one place • Echo effects for the supply chain and field issues engineering • Engineers never lose responsibility for their designs, they keep learning how to make their designs and production better • Always have back-up plans for any schedule limiting component or sub-system • Delays in delivery of a critical path item usually relax the tempo of the entire development • Reigniting tempo is much harder than keeping it up continuously
- 79. PROJECT MANAGEMENT TOOLS • Rule:Any cause for delay that becomes probable has to be immediately reported to the chief engineer in writing by the person anticipating the delay • Have a fast, transparent change management process • Relatively easy to do in loosely coupled, top-down systems • More challenging in tightly coupled, bottom-up systems • Change propagation matrix or network visualization tools help avoid simple cognitive errors • War room change implication stand-up meetings, up to 3* daily at highest tempo
- 80. PROJECT MANAGEMENT TOOLS • Conditiondevelopment teams to present options to recover from setbacks requiring escalation for decisions in terms of both cost and time • What is the option that may cost more, but recovers time that would likely otherwise be lost? • What option keeps costs in check, but may take more time? • Instills dual wherewithal in speed and technical dexterity • Kill problems, don’t just wound them • Otherwise, they tend to come out of remission at inopportune moments with amplified impact to speed and quality of R&D
- 81. PROJECT MANAGEMENT TOOLS • Theengineer who screwed up reports the screw up to the chief engineer first • Quid pro quo is that the chief engineer will provide a first level heat shield for those who come clean quickly • Estimate work down to ½ day increments for each contributor • Doing so tends to make sure that all work makes it onto the WBS, avoiding dark work that otherwise can unexpectedly surface later in the development project wreaking technical and program havoc • Fine grained effort estimation errors tend to cancel whereas coarser estimation errors tend to have similar over- or under-estimation bias, with errors cumulating adversely • Peer and supervisor review effort estimates to get better at this quickly
- 82. PROJECT MANAGEMENT TOOLS • Pullthe development team’s leadership together at 10% increments through the development program • Conduct overall reviews of design trade-offs, key learnings, propagation of change, assumptions, trends, and inferences • Necessary to make sure that blind spots aren’t building up in the daily maw of tactical design decisions
- 83. PROJECT MANAGEMENT TOOLS • Getan early grip of the time acceleration factors that will be used during the test phases of the development • Much time goes by during test and a lot of cost accrues • Late performance or reliability surprises are much costlier to fix than if they can be revealed earlier • Send out daily, detailed results reports during integration and system test • Make sure everyone involved in the design knows exactly where things are at as issues emerge during test
- 84. COMPUTER MODELS AND SIMULATION •Appropriate Use • During detailed design after the basic concept has been properly defined and critical design issues fully understood based on fundamental understanding • Risk from Premature Use • The emotional suasion of the appearance of predictive information access, control, and intuitively appealing graphics • Obfuscation of incomplete or flawed critical issue understanding
- 85. COMPUTER MODELS AND SIMULATION •Sign of constructive use • First pass results within 10% of up-front back-of-envelope estimates, and ultimate implementation • Monte Carlo multi-variate analyses in tightly coupled systems • First time close, if not first time right • Problem diagnostic • If R&D staff don’t understand they underlying subject matter well enough to do back-of-envelope analysis to the above accuracy, and if implementation results are further off • If people leave out important variables, with technically lackluster explanations about the reasons why • If issues and their efficient resolution need to move fluidly across technology domains • In these cases, simulation-centric iteration will usually only achieve sub-standard local optimization, often slowly
- 86. TOOLS AND SCAFFOLDING • Sustainablyleading enterprises • Typically need to invest in at least one capital intensive area of infrastructure, which most competitors choose not to, which gives the winner deeper technical insight into a core performance issue, or the ability to iterate and optimize much faster than peers • Me-too infrastructure risk • Makes it much harder to achieve and sustain outperforming R&D productivity over time
- 87. TOOLS AND SCAFFOLDING • Toavoid support infrastructure costs without sufficient offsetting productivity: • Match the investment plan to a productivity plan for infrastructure utilization • Best: Competitively measurable speed and utilization rate impact, to drive out inefficiency • Build a shared understanding of how high impact and effectiveness will be measured, evaluated and course corrected • Typical winning candidates: • Tools that allow rapid trials at the frontier where incumbent and competitive efforts stalled out trying to achieve further progress
- 88. MODEL SYSTEMS • Mosthigh output R&D programs force the model systems for early test to conform considerably to the usage demands of the final operational systems • This is relatively simple to achieve in loosely coupled systems • It is more difficult in tightly coupled work, but more impactful to schedule and cost because of the intensity of interrelationships • Requires deep technical insight that there are no big unknown unknowns to make this bet pay off
- 89. MODEL SYSTEMS • Withmore contrived test of the system or major portions thereof, the full operational demands of making advanced technology perform at scale often get concealed until much later, slowing overall development • If high technical uncertainty requires simplified model systems, it is usually best to focus on deep component and sub-system performance understanding, to build up to larger system understanding
- 90. DOCUMENTATION • A DesignObjectives and Criteria manual should be updated at the beginning of each major development project • The design manual details how the R&D team plans to meet every performance, reliability and regulatory requirement • The document becomes a compendium of design expertise, and a how-to manual for creating a world beating product
- 91. DOCUMENTATION • Value driversfrom documenting design objectives and criteria • #1: Experience needs to be written down, studied, codified, and taught to become an institutional asset • #2: Writing and documenting activates different mental processes for designers than tactical on-the-fly design and development, to improve design quality at the source; the benefit is for the writer as much as the reader • In the mindset of The Medium is the Message • The design manual should not merely be a scattering of missives of varying quality on a number of electronic work surfaces and internal collaboration tools • To drive design quality, the design manual should be an edited, carefully prepared document for the future team, signed by its sections’ authors and editors, and dated
- 92. DOCUMENTATION • If veryhigh risk technology is on the critical path of near term product development, de-burden the entire development beyond the up front design manual from more rigorous upstream documentation • Instead, document very thoroughly and precisely on a fast following basis upon achieving success
- 93. SUPPLIERS • Use ofoutsourcing for significant developmental technologies requires more, not less, management attention than internal efforts • Training, tools and methods vary too much between organizations for many routine decisions to be made on autopilot, as they often would in a vertically integrated organization • Almost everything to be decided during subcontracted development at a supplier needs to be handled as an exception • Staffing and infrastructure economies can be realized from the use of good suppliers for contracted development, but not management
- 94. SUPPLIERS • To keepup R&D productivity with a significant supply chain doing commissioned R&D, expect to have to develop comparable subsystem and component expertise in design, development and manufacturing as exists in suppliers • The only way to make the right deals is with enough technical wherewithal • You build a more competitive and efficient market for the inputs you need with more technically skilled and business savvy view of what is being supplied; • Otherwise, over- or under-valuation of skills and assets can occur • Technical expertise in supplier oversight is the only way to achieve delegation without abdication, to be able to provide both help and credible threat of lifting work out to another vendor do police performance and timelines
- 95. SUPPLIERS • Assume: Eachmajor supplier will experience one major crisis where their technological understanding will prove insufficient to overcome a deficiency in a timely fashion. • Build sufficient domain expertise to be able to rapidly help fill the deficit • Trade-off to Accept: • Pride, technical orientation and intellectual commitments can create bias among those overseeing suppliers, but, bias is less risky than ignorance • In-depth understanding only comes from hands-on experience and fundamental technical training
- 96. SUPPLIERS • Fewer, deepersupplier relationships will usually work better than a large number of shallow relationships, for a tightly coupled system technology • Loosely coupled technologies are more forgiving for wide but shallow supplier connections
- 97. SUPPLIERS • Resident engineers,or rotating on-site engineers • Frequent visits are not enough; on-site is insight • Being present is the only way to really get to know the people, particularly the key scientists, engineers and managers • Avoids talking at rather than talking with supplier staff and management when trouble arises • However, guard against embedded staff at suppliers going native, or being too empathetic • Instill urgency to get answers fast, not to let questions or doubts languish • Reporting should be of how things really are, not glossing over difficulties or failures • The trick is to be able to guide suppliers technically with suggestions and criticism, without usurping the supplier’s responsibility by issuing direct orders about how they should do their work (except in extreme cases)
- 98. SUPPLIERS • Require directaccess to all internal supplier reports of deficiencies • Keep the focus during times of difficulty on the technical requirements and issues, and minimize the role of personalities • Define a technically centred issue communication and resolution protocol, such as the A3 report • Review supplier performance regularly with each vendor • Quarterly at a minimum, and monthly or even weekly during the highest velocity co-development programs
- 99. QUICK TESTS OFHIGH R&D PRODUCTIVITY • Can we plan and complete two weeks of work, to full scope and functionality, with an overall R&D productivity that is at least as good by $ and time on calendar as competitors and benchmark companies? • An R&D organization will never be better over long time frames than over short time frames • The two week completion test is easy to gauge quickly • The discipline in the face of shortfalls is to dispel belief of quick fixes over intermediate time horizons, without persuasive corrective efforts • Are the most capable people matched up with the most important and resource intensive projects? • There are power law dynamics on both sides • The productivity impact of matching is dramatic • Mismatches can accumulate quickly in a high speed environment with cascading series of tactical reallocations • This mapping exercise needs to be redone every ~three months in growth-stage technology enterprises
- 100. QUICK TESTS OFHIGH R&D PRODUCTIVITY • The work is the only item of primary importance • All other organizational issues and symbols are secondary • The work is deeply personal for people as evidenced by actions over time, not just words over the short term • Cynicism and burnout are low, even though the tempo is high • People are able to act on their perceptions quickly, reinforcing the institutional ability to think quickly and well • All contributors individually know and can articulate a similar sense about • 1) The sacrosanct, small overriding set of system technical success factors, and, • 2) The cost of delay
- 101. QUICK TESTS OFHIGH R&D PRODUCTIVITY • Unproven but critical theories get put to test quickly • There is a high level of imagination and resolve to build the diagnostic tools that can’t be readily and cost effectively sourced • Best sign: Devising and using several complementary diagnostic tools to corroborate results from any one
- 102. QUICK TESTS OFHIGH R&D PRODUCTIVITY • Development program documentation is clear, concise, consolidated, and accurate • Crisp technical, financial and operational documentation is not a sufficient condition for R&D success, but it is a necessary condition • Troubleshooters go to the source to get the most accurate data • There’s a lot of energetic, face-to-face discussion • Both within R&D, and at the interface between R&D and the rest of the enterprise
- 103. QUICK TESTS OFHIGH R&D PRODUCTIVITY • Danger signs at the highest tempo and risk phases of R&D, with susceptibility often highest late in a major development program • If improvisation is becoming confused • If superficial reasoning is being used to silence thoughtful alternative views • If performance demands are being shed from the more understood to less understood parts of the system • If people become lax when something first shows signs of working, rather than being wary of further challenges likely lurking
- 104. CARDINAL METRICS Story/Feature Point Developmentin Software and Network Services Platform Hardware and Embedded Systems High SKU Count Completion rate w/ full functionality Market Share Trailing 6Q new product sales Bounce backs COGS Trailing 6Q new product gross margin High severity production/field issues DiP # products to first prototype Issue resolution times and reopens Time to Market Time to first prototype Design in Progress (DiP, akin to WiP) Customer trial/sample conversion rate Time to Market # Products released Design iterations per product DiP Time to Market
- 105. SUMMARY • Highly productiveR&D organizations have intense cultures built around technical depth, ongoing learning, strong supervision and high stakes commercial success as the ultimate arbiter of merit • Individual high energy, creativity and intense pride are table stakes • Failure is anathema to the individuals and team; dedication to success is extreme
- 106. SUMMARY • Competition isthe engine of progress, but it needs to be channeled through various mechanisms for sustained best advantage • Small teams of high performers vastly outperform large teams of average performers • The most productive form of R&D organization varies widely and follows the character and function of the technology being developed
- 107. SUMMARY • The morecomplex and interrelated the technology, the more that R&D productivity comes down to high velocity management of risk, integration and learning • The top-down R&D management practices which serve many ICT efforts need to adapt considerably to match the needs of more integrated and tightly interrelated technologies • The biggest successes are always attributable to scientific ingenuity, technical proficiency, and relentless drive
- 108. UPCOMING SEMINARS Nov. 22,2017 • Middle Management Development: • Building the engine of management and leadership development within the rapid growth technology business
- 109. FURTHER DISCUSSION For arrange furtherprivate discussion of any of today’s topics or related matters: dave.litwiller@communitech.ca
- 110. LOTS OF FURTHERREADING
- 111.
- 112.
- 113.
- 114.
- 115.
- 116.
- 117.
- 118.
- 119.