🗄️🧠scott - amoon.world🌙

[[scott_stern]] --- - [[#Table 1: Definition and example of Four Axioms of Entrepreneurial Strategy|Table 1: Definition and example of Four Axioms of Entrepreneurial Strategy]] - [[#Table 2: Entrepreneurial Strategy Compass with Detailed Explanations and Examples|Table 2: Entrepreneurial Strategy Compass with Detailed Explanations and Examples]] - [[#Table 3: Categorization and Definition of Key Choices in Entrepreneurial Strategy|Table 3: Categorization and Definition of Key Choices in Entrepreneurial Strategy]] - [[#Table 4: 4es x 3evol|Table 4: 4es x 3evol]] - [[#Table 5.|Table 5.]] - [[#Table 6.|Table 6.]] - [[♻️world/culture/🏫university/MIT/TIES|TIES]] [[scott_stern]] with backgroud/belief/expertise in [[scott23🛠️_econ_idea_innov_ent.pdf]], [[scott24👓_Bayesian_Entrepreneurship.pdf]] have value hypothesis as [[val(scott).png]] [[📜stern06_tech_org_econ_exp]] --- mobility ventures from scott's book analyzed via [gpt](https://chatgpt.com/c/681391f4-3b88-8002-ba68-83b9ea9b3d5d) ![[Pasted image 20250514112204.png|400]] ## Table 1: Definition and example of Four Axioms of Entrepreneurial Strategy | Axiom | Definition | Combined Examples | |-------|------------|-------------------| | Freedom | There is more than one potential path to create and capture value from an idea | - Starbucks: Howard Schultz's cafe idea vs. original founders' coffee bean business - Amazon: Jeff Bezos exploring multiple verticals before choosing books - 23andMe: Considering consumer-focused ancestry research vs. enhancing prenatal genetic screening | | Constraint | An entrepreneur cannot pursue all these paths at the same time | - The Body Shop: Anita Roddick choosing ethical products over animal testing - RapidSOS: Focusing on enhancing 911 services rather than creating a standalone app - Pillpack: Choosing to build their own pharmacy instead of licensing their design to established pharmacies | | Uncertainty | Entrepreneurs are uncertain about the value of particular strategies but also the range of value that might result from their idea | - Amazon: Jeff Bezos's systematic search across multiple industries before settling on books - Biobot Analytics: Uncertainty about which data from sewers would be most valuable to measure and for whom - Gimlet: Uncertainty about the best way to revolutionize the podcast industry | | Noisy Learning | Learning is noisy and ongoing so finding out more about one route often allows a reassessment of other alternatives | - ThirdLove: Discovery of the need for half-size bras through customer feedback and experimentation - 23andMe: Iterative learning about consumer preferences in genetic testing and health information - RapidSOS: Ongoing learning about emergency response needs leading to pivots in their technology approach | [[📜gans20_choose(tech)]] ## Table 2: Entrepreneurial Strategy Compass with Detailed Explanations and Examples resource, capabilities, component/integrated innovations are abstracted meaning the process and supply chain | Strategy | 🦠Intellectual Property (scott) | 🐅Disruptor (angie) | 🐬Value Chain (charlie) | 🐘Architectural (vikash) | angie agree/disagree | | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------- | | Nickname | THE THINKERS | THE HUSTLERS | THE PARTNERS | THE BUILDERS | | | Tagline | "Ideas Factory" | "Creative Destruction" | "Core Competency" | "Zero to One" | | | Customer Focus | DELIVER value for EXISTING users | DISCOVER value for NEW users | DISCOVER value for EXISTING users | DELIVER value for NEW users | | | Innovation Type | Develop GENERAL COMPONENT innovations | Develop SPECIALIZED SYSTEM innovations | Develop SPECIALIZED COMPONENT innovations | Develop GENERAL SYSTEM innovations | | | Orientation towards | 👥📦COLLABORATION by 🧠⚙️CONTROLLING | 🥷🗄️COMPETITION by 🤜💪EXECUTING | 👥📦COLLABORATION by 🤜💪EXECUTING | 🥷🗄️COMPETITION by 🧠⚙️CONTROLLING | | | (⚙️Resources, 💪Capabilities) X (🗄️integrated, 📦functional) | leverage(ent, 📦FUNCTIONAL RESOURCES) | build(ent, 🗄️INTEGRATED 💪CAPABILITIES) | build(ent, 📦FUNCTIONAL 💪CAPABILITIES) | leverage(ent, 🗄️INTEGRATED ⚙️RESOURCE) | | | | | | | | | | Value Creation Hypothesis | deliver(🧠, 📦COMPONENT INNOV. resource partners, 👴EXISTING customers) | discover(🤜, 🌏SYSTEM INNOV. resource partners, 👶NEW customers) | discover(🤜, 📦COMPONENT INNOV. resource partners_👴EXISTING customers) | deliver(🧠, 🌏SYSTEM INNOV. resource partners_👶NEW customers) | disagree with its customer centered | | Value Capture Hypothesis | strat: 📦FUNCTIONAL ⚙️RESOURCE ops: 🧠CONTROLLING | strat: 🗄️INTEGRATED 💪CAPABILITIES ops: 🤜EXECUTING | strat: 📦FUNCTIONAL 💪CAPABILITIES ops: 🤜EXECUTING | strat: 🗄️INTEGRATED ⚙️RESOURCE ops: 🧠CONTROLLING | | | | | | | | | | Examples | Harry Potter, Getty Images, Xerox, DOLBY, INTELLECTUAL VENTURES, Genentech | NETFLIX, Zipcar, Salesforce, Amazon, Skype, oDesk | Foxconn, PayPal, Madaket, Mattermark, DRIZLY, STRATACOM | Facebook, AngelList, eBay, Ford, Etsy, Dell | | | high bar (collaborate) vs low bar (compete) | high bar | low bar | high bar | low bar | | | 🧠market viability testing vs 🤜go to market testing | market viability testing | go to market testing | go to market testing | market viability testing | | | 2025 version | emphasizes idea generation, retains control of a startup’s product, and aims to achieve growth by creating value for existing customers | creates value for a niche set of customers by redefining existing value chains and competing with incumbents that are poorly serving their customers. | builds and executes specialized products and services within a partner organization’s existing value chain, in a model similar to a consulting relationship. | builds and controls a novel value chain for new customers and succeeds by meeting a specific need for each stakeholder in the value chain. | | | media version 2025-06-06 | emphasizes idea generation, retaining control over a startup's product, and aims to achieve growth by creating value for existing customers "of a partner" | starts with a niche set of customers often those customers poorly served by the incumbents - and then scales to serve mainstream customers. | builds and executes specialized products and services within a partner organization's existing value chain, in a model similar to consulting. | builds and controls an "entirely" novel value chain for new customers and succeeds by meeting a specific need for each stakeholder within that new value chain. | | | [[mng(platform)]] 3 Value Types from [[Jose Lopez]] - **Standalone Value**: Intrinsic value of the product/service independent of network effects - **Same-Side Value**: Value from having more participants on the same side (supplier-supplier or customer-customer) - **Cross-Side Value**: Value from connecting different sides (suppliers and customers) | - **Primary**: Standalone Value - **Value Creation Mechanism**: These ventures develop innovations with inherent value independent of network effects (like Dolby's audio technology or Getty Images' content library) - **Value Expression**: "Our COMPONENT INNOVATIONS deliver value for EXISTING users through the inherent quality and capability of our offering, regardless of how many others adopt it" | - **Primary Value Type**: Hybrid of Standalone and Cross-Side Value - **Value Creation Mechanism**: These ventures create new systems that initially provide standalone value but grow through orchestrating new cross-side interactions (like Netflix disrupting entertainment by connecting viewers with content in new ways) - **Value Expression**: "Our SYSTEM INNOVATIONS discover value for NEW users by first providing compelling standalone benefits, then creating powerful cross-side connections as we scale" | - **Primary Value **: Same-Side Value - **Value Creation Mechanism**: These ventures enhance coordination among existing players in a value chain, creating efficiencies within supplier networks or customer communities (like PayPal improving payment processes among merchants) - **Value Expression**: "Our COMPONENT INNOVATIONS discover value for EXISTING users by enhancing coordination among similar stakeholders to improve established processes" | - **Primary**: Cross-Side Value - **Value Creation Mechanism**: These ventures build platforms and marketplaces that create value by connecting different user groups (like Facebook connecting users with developers, or eBay connecting buyers with sellers) - **Value Expression**: "Our SYSTEM INNOVATIONS deliver value for NEW users by orchestrating connections between previously disconnected stakeholder groups" | | 2025-06-06 🧠learn then act VS 🤜act then learn (go to market strategy??) ip strategy - market viability testing; disruptor strategy - go to market strategy 2025-06-17 connecting with b_r, b_c as dual problem's solution idea, i.e. [dual optimization in newsvendor model cld](https://claude.ai/chat/db810104-b534-4577-b5ad-9f2700dc9cb2) ![[🗄️🧠scott 2025-06-17-13.svg]] %%[[🗄️🧠scott 2025-06-17-13|🖋 Edit in Excalidraw]]%% ## Table 3: Categorization and Definition of Key Choices in Entrepreneurial Strategy | Error Type | Value Domain | Uncertainty Components | Definitions | |------------|--------------|------------------------|-------------| | Statistical Error | Value Creation | Customer | Any person, group or organization who will pay the startup money for any of its goods or services. Uncertainty arises from unpredictable customer preferences and behaviors. | | | | Technology | The tools, techniques, designs and knowledge used by a business to create practical value for consumers. Technological uncertainty arises from inherent architectural differences across different technologies that lead to inherently different ability to attain quality levels, innovation levels, and cost structures. | | Approximation Error | Value Capture | Organization | A startup's initial choice of capabilities it will nurture and resources that it will draw upon, determined by the entrepreneur's initial team and culture. Organizational uncertainty arises because of how agile or sluggish a firm may be in allocating resources to quality vs. innovation. | | | | Competition | Firms that provide similar products or services (or otherwise solve the same or similar customer needs) to a startup's chosen customer. Uncertainty arises from unpredictable competitive actions and reactions. | | | | Regulatory | Uncertainty arising from potential regulatory actions, including scope and timing of regulations, as well as the impact of antitrust regulations on industry structure. | | | | Industry | Uncertainty arising from the actions of firms, regulators, and the nature of technology. Factors such as technological architecture's influence on firm's ability to integrate or modularize, or antitrust regulation's influence on firm's accumulation of market power introduces variation in the industry structure. | | Optimization Error | Value Delivery | Market | Uncertainty arising from the actions of firms and consumers. On the supply side, firm's decisions affect switching costs, network effects, and integration/modularization. On the demand side, consumers' preferences for quality, innovation, price, and compatibility introduce uncertainty. | | | | Product | The set of potential product or service designs, features, and attributes. Uncertainty arises from the unpredictable success of different product designs and features in meeting customer needs and preferences. | [[⭐️thesis]] ## Table 4: four entrepreneurial strategy x three evolution types using [conv(jb, scott|charlie) cld](https://claude.ai/chat/9784e8f4-8480-4825-bc95-142c911151d4) 1. Speed ratio = 👆/👓 (implementation vs testing speed) 2. Feedback ratio = 💨/(👆+👓) (feedback vs total execution time) 3. Learning ratio = 🙈/🤯 (understanding vs theory decay) ![[Pasted image 20241203105756.png|1600]] | Ent.Strategy (Scott speaks) | Hypothesis (Scott speaks) | Example (Scott speaks) | Evolution (JB speaks) | Key Parameters (Charlie speaks) 👆/👓, 🔄/(👆+👓), 🍃/🗑️ | | ---------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | -------------------------------- | ------------------------------------------------------------------------------------------------------------------------------ | | Value Chain "The Partners" | CREATES VALUE as its COMPONENT INNOVATIONS discover value for EXISTING users CAPTURES VALUE as it EXECUTES on key FUNCTIONAL CAPABILITIES | Foxconn, PayPal, Madaket, Mattermark, Drizly, Stratacom | Adaptation (F1→F1+) | - High 👆implement time/👓test time - Low 🔄feedback time/👆+👓total time - High 🍃rejecting rate/🗑️retiring rate | | IP Strategy "The Thinkers" | CREATES VALUE as its COMPONENT INNOVATIONS deliver value for EXISTING users CAPTURES VALUE as it CONTROLS a key FUNCTIONAL RESOURCE | DOLBY, Harry Potter, Xerox, Intellectual Venture, Genentech, Getty Images, | Co-opted Adaptation (F1→F2) | - Low 👆implement time/👓test time - Medium 🔄feedback time/👆+👓total time - Medium 🍃rejecting rate/🗑️retiring rate | | Disruptor "The Hustlers" | CREATES VALUE as its SYSTEM INNOVATIONS discover value for NEW users CAPTURES VALUE as it EXECUTES on key INTEGRATED CAPABILITIES | NETFLIX, Zipcar, Salesforce, Amazon, Skype, oDesk | Co-opted Nonadaptation (None→F1) | - Very high 👆implement time/👓test time - High 🔄feedback time/👆+👓total time - High 🍃rejecting rate/🗑️retiring rate | | Architectural "The Builders" | CREATES VALUE as its SYSTEM INNOVATIONS deliver value for NEW users CAPTURES VALUE as it CONTROLS a key INTEGRATED RESOURCE | Facebook, AngelList, eBay, Ford, Etsy, Dell | Co-opted Nonadaptation (None→F1) | - High 👆implement time/👓test time - Medium 🔄feedback time/👆+👓total time - Medium 🍃rejecting rate/🗑️retiring rate | ## Table 5. epistemic vs aleatoric uncertainty | Business Risk | Fundamental Uncertainty | | --------------------------------------------------------------------- | -------------------------------------------------------- | | - Probability can be calculated exactly (e.g., 1/6 for fair die roll) | - Cannot assign accurate probabilities to outcomes | | - Based on historical data (e.g., franchise failure rates) | - No historical data to base predictions on | | - Applies to established businesses expanding to new countries | - Applies to entirely new business ideas/concepts | | - Examples: gambling odds, weather forecasting, insurance pricing | - Examples: new product launches, unfair die rolls | | - Can do clear-cut risk calculations | - No structure to understand scenarios and probabilities | | - Data-driven assessment possible | - Cannot realistically conceptualize scenarios | | - Familiar business scenarios with known variables | - Uncertainty about underlying quality of idea | | - Decision-making process is straightforward | - No clear framework for decision-making | ## Table 6. test2choose1 examples | Key Message | Example | Lesson | | -------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- | | Test Multiple Paths | 🎯 PillPack testing three customer segments: - Elderly patients - Post-incident middle-aged - Health-conscious supplement users | Different market segments reveal different value propositions and challenges | | Commit After Testing | 🍲 Soup seasoning analogy | You must eventually commit to a strategy, but only after proper testing | | Low-Cost Initial Testing | 📚 Amazon's early book category testing 👗 Vera Wang's small Madison Avenue salon | Start with controlled experiments before major investment | | Danger of Single-Path Commitment | 🧃 Juicero's $400 juicer failure 🏝️ Fyre Festival disaster | Lack of testing multiple strategies can lead to expensive failures | | Balance Learning & Operations | 🥾 L.L. Bean Boot: - Good: Market validation - Bad: Initial manufacturing issues | Success requires both market fit AND operational excellence | | Uncertainty Management | 🎲 Unfair dice analogy | You can't eliminate uncertainty, but you can manage it through systematic testing | | High-Fidelity, Low-Cost Testing | 👗 Vera Wang testing designs alongside established competitors | Find ways to get meaningful feedback without major resource commitment | ## Table 7. why test2choose1 | Strategic Element | Core Concept | Key Examples | Implementation Principles | | --------------------- | --------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------- | | ⭐️Innovation Quality | Create robust solutions through systematic exploration | Vera Wang's revolutionary designs; L.L. Bean's waterproof boot development | - Think deeply about value creation - Solve problems creatively - Build on validated insights | | Risk Reduction | Multiple viable strategies reduce dependency on single approaches | Fyre Festival failure due to single-path strategy; Amazon's early category testing | - Test multiple paths before commitment - Validate core assumptions - Maintain alternative options | | Resource Optimization | Identify most efficient route through comparative testing | Jeff Bezos testing books alongside other categories; OpenTable's focused market entry | - Compare cost/benefit ratios - Start small, scale proven approaches - Minimize sunk costs | | Market Learning | Generate comprehensive customer insights through different approaches | L.L. Bean's boot testing with hunters; Vera Wang's bridal salon experiment | - Gather diverse customer feedback - Test in real market conditions - Validate customer needs | | Strategic Flexibility | Maintain ability to pivot while gathering critical information | Vera Wang's limited initial investment; OpenTable's city-by-city expansion | - Keep initial commitments limited - Build in adaptation options - Preserve pivot capability | | Competitive Advantage | Build stronger market position through informed testing | OpenTable's focused San Francisco testing; Amazon's category selection | - Understand market alternatives - Position strategically - Build scaling foundation | | Failure Prevention | Identify critical flaws before major investment | Juicero's $400 juicer failure; L.L. Bean's prototype testing | - Test core assumptions early - Identify potential pitfalls - Validate before scaling | examples from [[Scott_school.pdf]] p.1: 🎲fundaments uncertainty on knowing unfair dice p.2💊PillPack’s idea: a roll of bags of medications presorted by the time of day of each dose for online pharmacy Three executing strategy (customer segments): - elderly patients with multiple chronic conditions - middle-aged patients after their first major medical incident (e.g. heart attack) - healthy consumers with highly individualized and complex vitamin and supplement routines p.3: 🍲To get the best tasting soup, you have to commit to how you season it (and maybe then see what Gordon Ramsey thinks of it). p.4: 📚 Amazon’s lowbar testing, 🧃 Juicero and Fyre Festival’s failure from not testing (let alone parallel) “Had Evans considered multiple strategies, before committing to the technology and model, he may have better understood the potential flaws in his assumptions and in the idea itself” p.5: 🥾bean boot mixed learning (bad operations, good market) i.e. option creation p.6: 👗vera wang’s testing of critical hypothesis with high-fidelity and low opportunity costs #scott [[val(scott).png ]] - role: evaluate, segment - hypothesis: best segment for persuasion - marginnote3app://note/FDFC2128-3EB3-4F93-A4DD-7D895BB4CA77 - evaluates [[4🔴metrics]], [[T5_💜mob_innov_msr]], [[T4_💙ENT_DSL]] [[2025-11-19]] 테스트는 아이디어·전략의 결합된 신호이므로 “둘 다에 대한 분리”를 점진적으로 만들어냅니다. 구현은 “전략층 분리”를 확정하지만, 기회비용 k 때문에 다른 경로를 소거할 위험이 있습니다. 이 때문에 선(先)테스트 → 필요하면 추가테스트 → 하나 선택이라는 순서가 나옵니다. [[Front/On/love(cs)/thesis_v1/📝product/00_TOC|00_TOC]] ---- ## Table 8. Uncertainty in Start-Up Ideas Does the type of start-up idea and the nature of the uncertainty it faces affect the predictability of that idea's success? To explore these questions, researchers empirically studied the evaluations of 537 early-stage start-ups founded by MIT students and alumni in high-growth industries. These evaluations, performed by 251 experienced entrepreneurs, investors, and executives, were never disclosed to the entrepreneurs, and the assessments themselves did not affect the start-up (e.g., by possibly influencing access to mentoring or funding). Yet, without having met the founding team and with only the information contained in brief, half-page summaries, these evaluators successfully gauged the quality and subsequent success of the early-stage ventures. The explicit start-up and entrepreneur characteristics (such as start-up sector, start-up progress, or founding team experience) contained in the written summaries were not necessary to make their judgments—the strength of the idea spoke for itself. However, the researchers found something else. They found that the experts could only effectively evaluate start-ups in sectors such as hardware, energy, life sciences, and medical devices. They were not successful in markets such as consumer products, consumer web and mobile, and enterprise software sectors. The researchers suggested that the information needed to assess start-ups depends on the nature of the uncertainty they face. For instance, in settings where the technological uncertainty associated with the start-up idea is high relative to the market uncertainty (e.g., a new insulin treatment), one can compare the idea to successful commercialization of products using similar technologies. However, where market uncertainty is relatively high (e.g., a new service targeted at enterprise customers), it may be more difficult to predict consumer behavior." # Table 9. test idea or performance | Variable | Description | **Coffee Context (Sec. 2.1)** | **EV Context (Sec. 2.2)** | | --------- | ------------------------------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------- | ---------------------------------------------------------------------------- | | a_φ, b_φ | Beta distribution parameters for idea quality prior | Prior belief about consumer preference for premium coffee | N/A | | a_θ, b_θ | **Beta distribution parameters for implementation effectiveness** (used in **Beta-Bernoulli model** in Sec. 2.1 and **Normal-Beta model** in Sec. 2.2) | Prior belief about implementation success (Starbucks vs. Peet’s) | Prior belief about implementation success (battery efficiency, scaling) | | μ_φ | Mean of normal distribution for idea quality | N/A | Expected market demand for EVs ($/year) | | φ (phi) | Fundamental market potential (population parameter) | Market demand for premium coffee ($/year) | Market demand for EVs ($/year) | | θ (theta) | Implementation capture rate (unit-level parameter) | Success of implementation strategy (0 or 1) | Success of implementation strategy (continuous, e.g., sales conversion rate) | | y | **Realized profitability** | Profitability of chosen implementation strategy ($, binary outcome) | Profitability of EV model ($/year, continuous outcome) | | c_φ | Idea validation cost | Cost of blind taste tests, market research ($) | Cost of prototype testing, surveys ($) | | c_φθ | Implemented idea test cost | Cost of opening stores, launching sales ($) | Cost of pilot production, small-scale launches ($) | [[📜tenanbaum14_1sample(1decide)]] ## Table 10. | Prior Reality \ Signal | 😫 Bad News | 😆 Good News | | ---------------------- | --------------------- | --------------------- | | 💩 Bad Idea | λ₀ (True Negative) | 1-λ₀ (False Positive) | | 💡 Good Idea | 1-λ₁ (False Negative) | λ₁ (True Positive) | [[cohesive]] low and high bar test false positive (1-lambda0; good news failure), false negative (1-lambda1; bad news for success) mu upper bar (any positive new - you're in), mu lower bar (move a lot) - mu uppr bar (); mu underlying experimental technology comes with type1,2 error (repeated draws - precise signal) - proportionally spend effort to figure out min type 1 or type2 or both (budget of 2) - signal would be perfect (min type 1 error; lambda0=1) lambda0=1 -> no false postivite; lambda1 = 1 -> no false negative (given the capital lambda) if the signal is less then have of the time, you flip (you get 1 for free) for those not at mutilde, put all your effort (if you're i'm in -> choose low bar; choose experiment only fails (high fp, low fn)) - best foot forward (zappos) -> skeptic (choose high bar; get rid of false positive (only way this works is it actually works)) only reason it sells is bc it is good zappos (only situation they failed is they; even though they believed in it); high bar: hard to pass tight prior ; uniform prior ![[Pasted image 20250212141732.png]] [[📜gans23_choose(ent, exp)]]] | Approach | Low Bar (Lenient Threshold) | High Bar (Strict Threshold) | | ------------------------------------- | -------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------- | | 🧠Market Viability Learning (Control) | Quick customer surveys about product features Focus groups for new product concepts | Blind taste tests for premium coffee with rigorous statistical significance Early Tesla battery performance tests in controlled lab conditions | | 🤜Go to Market Learning (Execute) | Zappos' initial shoe sales experiment Airbnb's initial photo service for NYC listings | Commonwealth Fusion Systems' full reactor test SpaceX's complete rocket launches | 2. Compete vs Collaborate and Bar Height: This is an intriguing question. Let me think through it: For 🥷 Compete: - Entrants often use low bar testing because: - Need to move fast before incumbent response - Can iterate and improve rapidly - Example: Netflix's initial DVD-by-mail was a relatively low bar test compared to Blockbuster's infrastructure - Can't afford long validation periods For 👥 Collaborate: - Entrants often use high bar testing because: - Need to prove credibility to potential partners - Must demonstrate clear value proposition - Example: Genentech needed rigorous scientific validation before pharmaceutical partnerships - Have more time for thorough validation due to collaborative rather than competitive dynamics ## bayes.ent book The greatest myth of entrepreneurship may be that entrepreneurs are born, not made. Those who subscribe to this view argue that it is the innate gifts of the individual—a heroic tolerance of risk, extreme self-confidence, omniscient vision, and ultimate powers of persuasion—that determine whether someone becomes a successful entrepreneur. Accordingly, efforts to teach entrepreneurship are futile and, by extension, so are efforts to conduct scholarly research on the topic beyond hagiographies. The myth of the entrepreneur has been repeated so often that it can be easy to overlook its flawed logic. All professions benefit from talent — the innate skills and traits that make some more capable than others. But they also benefit from training — from lawyers to engineers, scientists to athletes. Certainly, gifted brain surgeons are endowed with a combination of exceptional intellectual and emotional capacities and manual dexterity. And yet, no one argues that “brain surgeons are born, not made”—nor (thankfully) has anyone been persuaded to stop training surgeons, lawyers, engineers, actors, musicians, dancers, accountants, economists, scientists, or athletes all of whom benefit from their innate talent when combined with rigorous and systematic training. No different from any other professional, an entrepreneur should engage in a profession of entrepreneuring. We intentionally use the word “entrepreneuring” instead of entrepreneurship. Entrepreneurship, in its literal meaning, suggests a state of being. Someone who is an entrepreneur, however, is not just in a state of being an “entrepreneur.” They are engaged in a specific set of activities – entrepreneuring – just as engineers are engaged in engineering. The word entrepreneurship over-emphasizes the qualities of the individual while entrepreneuring recognizes the activities, behaviors and practices that — through training — allow people to become better entrepreneurs. A systemic body of knowledge and principles is essential to effective professional education. And yet, unlike other fields within management, entrepreneurship has lacked a unifying framework defining the central questions, key skills and offering a conceptually coherent means to address them. It is hampered by the lack of what Michael Porter’s “Five Forces” framework provided to the field of strategy, or what the efficient market hypothesis and capital asset pricing model contributed to the field of finance. Entrepreneurship has been a phenomenon without a central theory or even a set of theories. The collection of papers in this volume of Bayesian Entrepreneurship is an attempt to (re)orient the field of entrepreneurship around the focal challenge facing the entrepreneur: decision- ii iii making under uncertainty and the processes of experimentation, learning, and adaptation. By structuring the field around a core activity that is relevant to all entrepreneurs regardless of the market opportunity they are addressing or the technological solutions they are generating, this approach makes a potentially large contribution to the study and practice of entrepreneurship. Bayesian Entrepreneurship is an attempt to orient the field around entrepreneurs’ prior beliefs about their opportunity space and how those beliefs evolve over the course of their entrepreneurial journey. If we take a Bayesian perspective, then entrepreneurship is more about establishing hypotheses rather than having the right “vision”, and running the right experiments rather than writing one-hundred-page business plans. In fact, a Bayesian view of entrepreneuring accepts that entrepreneurs rarely have their initial vision right given that their priors are based on incomplete information, and thus are necessarily flawed. What they can do is systematically learn through experimentation. Whatever their starting point, entrepreneurs will, by definition, face varying degrees of uncertainty and risk. For those who follow the model developed at Flagship Pioneering, and who are engaged in completely unprecedented science, technology or markets, the challenge is one of uncertainty in the sense Frank Knight described it more than a hundred years ago. In this “Knightian” world entrepreneurs are akin to pioneers who venture into new, uncharted lands in order to discover (and settle) the most attractive places and they need to constantly experiment in order to chart their territory. For others solving more familiar problems in well charted terrain, their job will be to experiment to resolve specific risks. Regardless, Bayesian entrepreneurs are experimenters par excellence. An entrepreneur earns rents — for themselves, their team and their investors — by discovering among the vast pool of uncertain opportunities those which offer the most attractive risk-return profiles. The process by which this occurs is anything but straightforward. In most cases, the Knightian entrepreneur, like the pioneer, has only a vague notion about what might lie “over the horizon.” Other entrepreneurs will be focused on which path up the mountain is more likely to yield a rapid ascent given the risks inherent in both. They may be able to make probability estimates of technical or market risks, but outcomes can vary dramatically. Both probe, both experiment, and both update their plans and visions (their priors) based on what they learn. Bayesian perspectives are useful for understanding the approaches of entrepreneurs along the full spectrum of risk and uncertainty. If we take a Bayesian perspective, then entrepreneurial strategies emerge through an evolutionary process of exploration (search), selection (experimentation and evaluation), and adaptation (learning). Emergent processes generate outcomes that are not fully predictable at the outset, yet might appear perfectly rational in hindsight. There are serious implications of this concept for the study and teaching of entrepreneurship. Too often, case studies of entrepreneurship draw lessons that are distilled based on a particular company’s successful strategy. This has a number of flaws for training a more effective Bayesian entrepreneur: could they repeat such a strategy (e.g. how could you create the next Amazon), was it luck or a judgment based on information gathering and experimentation, what were the alternative paths explored but not selected? While stories of failure often teach us more, they are often pushed out of the typical iv entrepreneurship classroom in favor of heroic, idiosyncratic successes. An emergent (Bayesian) perspective warns us of the fallacy in teaching only from the simplified, linear success stories. Entrepreneurs change their mind all the time (“update their priors” in Bayesian terms). Steve Jobs was famously adamant that he didn’t want 3rd parties developing apps for the iPhone—until he changed his mind. Amazon Web Services was not the product of a grand vision of Jeff Bezos. It emerged from a small experiment to open Amazon’s application interfaces to third-party developers in order to drive more traffic to Amazon’s retail site from “affiliate” sellers. It was only after seeing the results of this experiment that Amazon leaders began to realize the greater potential of web services. Netflix initially used traditional per-rental pricing of videos (similar to incumbents like Blockbuster) but later changed its strategy to a monthly subscription pricing model. Moderna did not set out to create a new vaccine technology but rather to enable a patient’s body to make any needed biotherapeutic drug. After building its unprecedented mRNA platform, the use of this new therapeutic modality to make a prophylactic vaccine for healthy subjects (not patients) emerged as the preferred application. When viewed as the result of an emergent process of search and selection, a successful strategy is just one outcome along a path of many alternative evolutionary branches. Teaching entrepreneurship through a Bayesian lens means following these paths and focusing on the (many) choices along the path—and learning lessons about the search and selection processes itself—rather than simply the final selected outcome which we observe as a success. It also means learning about the paths not taken, and the cases where even when entrepreneurs searched widely and designed, executed and evaluated their experiments effectively, the outcome was still disappointing. It is also an opportunity to teach about when to stop along a particular branch — in a way that is more systematic and dispassionate — so as to reallocate resources (including the entrepreneur’s own time and energy) to other activities. The papers in this volume help us understand the economics behind this Bayesian process. They give analytical rigor to activities that skillful, reflective entrepreneurs and venture builders do on a daily basis. The economics of Bayesian entrepreneurship are a reminder that experimentation is, after all, not free. Entrepreneurs do not have unlimited resources. They do not have unlimited time, money or talent. Allocation of resources to one experiment means not allocating it to another. It is one thing to believe that entrepreneurs should experiment and update their beliefs based on new data—it is quite another to understand what the best experimental strategies may be under different circumstances. The Bayesian approach developed in this volume begins to help us understand how to make choices among different experimental strategies and how those choices may impact venture performance. And, how those pathways might be more or less successful in different regions and countries depending upon the resources of the ecosystems in which entrepreneurs build their ventures. It also highlights an important organizational aspect of the entrepreneurial process—that individuals and their teams must be able to adapt their priors (and behavioral economic research suggests this may not be so easy in practice!), but so too must their stakeholders (employees, directors, investors, partners, etc.). The entrepreneurs’ burden is not just to run the right experiment, and keep their minds open to updating, but also to convince others as well. v The notion that entrepreneuring revolves around experimentation also invites inquiry into questions of organizational process. Many writings highlight, even celebrate, the seeming chaotic nature of entrepreneurial ventures. They attribute their dynamism and adaptability to their lack of processes and structure (which are, in turn, equated with “bureaucratic” corporate enterprises). Indeed in this perspective, the words entrepreneur and process do not belong in the same sentence. A Bayesian perspective — which puts experimentation at the heart of the entrepreneurial process — challenges this perspective. Experimentation is a process that requires careful consideration of design, disciplined execution, rigorous collection and analysis of data, and thoughtful evaluation. Experiments can produce surprises (they often do), but the process of experimentation itself should not be a surprise. And by the same logic, the process of experimentation can be shaped and structured to be more systematic and undertaken in parallel and at scale, as has been the focus of decades of work at Flagship Pioneering bringing rigor to entrepreneuring. From a Bayesian perspective, there is nothing ad hoc about the entrepreneurial process. It is a rigorous, careful process of experimentation, iterative learning, and adaptation. Preface The field of entrepreneurship has long been characterized by debates over whether success is a product of innate talent or can be systematically nurtured and taught. In this book, we introduce the concept of Bayesian Entrepreneurship—a novel, synthetic approach that reconceptualizes entrepreneurial decision-making as a dynamic process of forming, testing, and updating subjective beliefs in the face of uncertainty. Drawing on rigorous Bayesian learning principles, the book integrates insights from decision theory, economics, and behavioral studies to provide a coherent framework for understanding how entrepreneurs identify opportunities, experiment with strategies, and ultimately persuade stakeholders. The aim is to offer researchers and graduate students new tools and pathways to analyze and teach entrepreneurial phenomena. A Bayesian approach to entrepreneurial decision-making offers a framework to assess how entrepreneurs form beliefs about the prospects for a given opportunity, how these beliefs evolve over time through active experimentation and learning, and the consequences of such beliefs for entrepreneurial strategy and performance. The goal of this book is to shape distinctive implications and empirical approaches to the study of entrepreneurship guided by three founding premises. First, the entrepreneur must hold stronger positive beliefs about the opportunity relative to others. This involves a distinct theory that translates into a different perspective on the opportunity’s prospects. Second, this systematic divergence in beliefs impacts how an entrepreneur will undertake learning about an opportunity. Notably, the demand for “experiments” is fundamentally influenced by beliefs about the opportunity. For example, relative to a disinterested agent, a Bayesian entrepreneur will conduct experiments that are more likely to allow for “false positives” than “false negatives.” Finally, entrepreneurs are more likely to convince those who share their idiosyncratically positive beliefs about an opportunity (with implications for homophily and firm culture), yet will also engage in choosing experiments that cater to those with different (more negative) beliefs than they themselves hold. Chapter 1, by Ajay Agrawal, Arnaldo Camuffo, Alfonso Gambardella, Joshua S. Gans, Erin L. Scott, and Scott Stern, lays the theoretical groundwork for the Bayesian approach to entrepreneurship. The chapter argues that entrepreneurial decision-making can be rigorously understood as a process in which individuals hold heterogeneous and, often, strongly optimistic prior beliefs about opportunities. These priors are systematically updated through experimentation and learning via Bayes’ rule. By synthesizing insights from traditional economic theory and emerging behavioral research, the authors demonstrate how the Bayesian framework provides a structured method for evaluating risk, testing hypotheses, and ultimately guiding strategic xiii xiv CONTENTS choices in uncertain environments. In Chapter 2, Alfonso Gambardella, Luca Gius, and Scott Stern shift the focus to the nature of the entrepreneurial agent—termed Homo Entrepreneuricus. The contributing scholars examine how individual entrepreneurs differ from other economic agents by virtue of their idiosyncratic, often contrarian, beliefs and cognitive predispositions. By combining empirical observations with theoretical modeling, the chapter illustrates how these unique traits shape opportunity recognition and decision-making. The authors argue that understanding these distinct features is essential to explaining why entrepreneurs consistently pursue ventures that defy conventional wisdom, thereby setting the stage for a Bayesian analysis of their behavior. Joshua Gans in chapter 3 critically contrasts the classical notion of Knightian uncertainty with the Bayesian perspective on entrepreneurial decision-making. Through a detailed exploration of uncertainty types, the authors demonstrate that while entrepreneurs undoubtedly face profound unknowns, the systematic application of Bayesian updating allows for the quantification and management of such uncertainty. By formalizing the conditions under which probabilistic reasoning remains valid even in the presence of incomplete information, this chapter provides a robust argument for abandoning fatalistic views of uncertainty in favor of a model that embraces learning and adaptability. In chapter 4, Ashish Arora brings theory into practice by presenting a user-oriented examination of Bayesian Entrepreneurship. The authors, drawing from both academic research and real-world case studies, illustrate how entrepreneurs apply Bayesian principles to navigate complex market dynamics and secure necessary resources. The chapter emphasizes the role of experimentation—not only as a means for internal learning but also as a tool for persuading external stakeholders such as investors and partners. This practical orientation bridges the gap between abstract theory and the tangible challenges of entrepreneurial execution. In Chapter 5, Thomas Åstebro, Frank M. Fossen, and Cédric Gutierrez turn to the behavioral aspects of decision-making in entrepreneurship. The contributing authors investigate whether the apparent biases and heuristics observed in entrepreneurial behavior are simply flaws or if they can be reinterpreted as rational strategies under a Bayesian framework. By juxtaposing traditional interpretations of overconfidence and optimism with models of systematic belief updating, the chapter challenges prevailing notions and highlights how even seemingly irrational behavior may have adaptive value when viewed through the lens of Bayesian learning. Having explored what Bayesian entrepreneurship is, the book now considers where priors come from. Ehrig, Teppo Felin, and Todd Zenger (Chapter 6) examine the foundational role of causal reasoning in shaping entrepreneurial priors. Authored by experts in both entrepreneurship and decision science, the chapter argues that the process of constructing a causal model of opportunity is central to forming robust prior beliefs. Through a blend of theoretical discussion and empirical evidence, the authors show how entrepreneurs use logical reasoning and scientific principles to develop, test, and refine their hypotheses about value creation. In Chapter 7, Elena Novelli presents a detailed exposition of how a scientific approach to decision-making naturally aligns with Bayesian learning. By conceptualizing entrepreneurial action as a series of experiments that test underlying hypotheses, the chapter elucidates the CONTENTS xv mechanisms by which systematic data collection and analysis improve decision quality. The discussion highlights the importance of rigorous experimentation in reducing uncertainty and accelerating the convergence of beliefs, thereby enhancing overall venture performance. Chapter 8 from Ramana Nanda explores the intricate interplay between initial beliefs, experimental design, learning, and persuasive communication. The authors demonstrate that the design of experiments not only informs an entrepreneur’s self-assessment but also plays a critical role in convincing external stakeholders to invest in the venture. By analyzing how different experimental strategies can yield varied signals and updates, this chapter provides a nuanced understanding of how Bayesian persuasion can be effectively employed to align divergent beliefs among partners, investors, and other decision-makers. Part III of the book focuses on how priors are updated through experimentation and communication. Hyunjin Kim in chapter 9 focuses on the centrality of data in the entrepreneurial learning process. Through an in-depth analysis of various data collection and interpretation methodologies, the authors argue that accurate, timely, and well-structured information is vital for updating prior beliefs. The chapter outlines practical techniques for minimizing noise and bias in experimental data, thereby ensuring that learning is both meaningful and actionable for entrepreneurial decision-making. In Chapter 10, Josh Krieger deepens the discussion by exploring how continuous Bayesian updating drives adaptive learning in entrepreneurial ventures. By providing formal models and empirical evidence, this chapter illustrates how iterative belief revision—prompted by sequential experiments—leads to improved strategic decisions. The authors emphasize that this dynamic process is essential not only for refining current ventures but also for identifying when to pivot or terminate a project in light of new evidence. Susan Cohen and Rem Koning (Chapter 11) delve into the role of advice and external information in shaping entrepreneurial choices. The contributing scholars examine how the dissemination and reception of advice can be understood as a Bayesian process, where the credibility and prior beliefs of both the advisor and the entrepreneur interact to produce a persuasive outcome. The chapter provides insights into how entrepreneurs can leverage external expertise to update their beliefs, thereby enhancing decision quality and strategic alignment with key stakeholders. The final part considers the pedagogy and practice of entrepreneurs and how these are impacted on by a Bayesian approach. Erin Scott (Chapter 12) investigates the interrelationship between learning processes, strategic commitment, and decision-making in entrepreneurial ventures. The authors articulate how early-stage learning influences long-term commitment decisions and how Bayesian updating can mitigate the risks associated with premature commitment. By examining both theoretical models and case studies, this chapter offers a comprehensive framework for understanding how initial choices set the trajectory for future venture performance. In Chapter 13, Chiara Spina focuses on the determinants of effective learning in entrepreneurial contexts. The contributing authors identify and analyze a range of factors—from individual cognitive capabilities to external environmental influences—that facilitate or impede the learning process. The chapter emphasizes that systematic experimentation, when properly designed and executed, not only accelerates learning but also enhances the entrepreneur’s ability to adapt to xvi CONTENTS evolving market conditions. The final chapter (14) from Andrea Coali1, Claudia Frosi1, Diego Jannace1, Saeid Kazemi1, and Abhinav Pandey integrates the theoretical and practical dimensions of the book by advocating for the incorporation of causal reasoning into entrepreneurial education. The authors argue that by teaching future entrepreneurs to construct and critically evaluate causal models, academic programs can better prepare them for the uncertainties of real-world ventures. This chapter outlines practical pedagogical strategies and offers a roadmap for embedding Bayesian methods into the curriculum, ensuring that the next generation of entrepreneurs is equipped with the tools to make sound, data-driven decisions. In sum, this book is designed for researchers in entrepreneurship and graduate students seeking innovative analytical tools and theoretical frameworks. By weaving together rigorous Bayesian methods with practical insights from entrepreneurial practice, the volume not only challenges conventional perspectives but also provides a comprehensive guide for systematically exploring and understanding the complex dynamics of entrepreneurial phenomena. [[25_TEPEI_empirical]] --- \section*{Working Interpretation of My Advisors' Mental Models} \subsection*{Professor Scott Stern: Entrepreneurial Strategy as Bayesian Choice Under Uncertainty} I understand your mental model of entrepreneurial strategy as starting from four axioms: (1) \emph{freedom} -- every idea admits multiple distinct paths for creating and capturing value; (2) \emph{constraint} -- an entrepreneur cannot pursue all of these paths in parallel and must choose among them; (3) \emph{uncertainty} -- both the level and the range of value associated with each path are uncertain; and (4) \emph{noisy learning} -- experimentation generates information in a way that is ongoing, imperfect, and often reshapes beliefs about other paths as well. Together, these axioms frame entrepreneurial strategy not as an optimization problem with a unique best plan, but as a sequence of high-stakes choices made under deep uncertainty, in which the entrepreneur must choose \emph{which} environment to place the venture into rather than simply adjusting to a given environment. Building on these axioms, I interpret your Entrepreneurial Strategy Compass as specifying four canonical commercialization strategies -- Intellectual Property, Value Chain, Disruption, and Architectural -- that arise from the endogenous choice of two fundamental dimensions: the cost of accessing specialized complementary assets and the strength of the appropriability regime. For a given idea, each quadrant corresponds to a distinct value-creation and value-capture hypothesis (for example, component versus system innovation, and existing versus new users), along with specific complementarities between collaboration versus competition and control versus execution. In my reading, the compass encourages the entrepreneur first to articulate multiple coherent strategy hypotheses for the same idea, then to recognize that committing to any one of them \emph{creates} the strategic environment in which the venture will subsequently be evaluated. I see your taxonomy of uncertainty and error types as refining this compass into a design space for experimentation. Along the value-creation dimension, uncertainty about customers and technology is modeled as statistical error: entrepreneurs can, in principle, learn about preferences and technological performance through repeated sampling, even though the sampling process is noisy. Along the value-capture dimension, uncertainty about organization, competition, regulation, and industry structure is captured as approximation error: here the challenge is to choose and refine a tractable representation of a complex strategic environment, knowing that any representation will be imperfect. Along the value-delivery dimension, uncertainty about markets and products is treated as optimization error: even with reasonably accurate beliefs about demand and technology, entrepreneurs can still mis-allocate effort across pricing, product design, and channel choices. This three-by-three structure (value creation, capture, delivery crossed with statistical, approximation, and optimization errors) gives me a way to locate where a particular strategic question sits and to choose correspondingly appropriate experimental designs. Within this structure, I interpret your ``Test Two, Choose One'' principle as a normative rule for experimental strategy in the face of deep uncertainty and high opportunity cost. Rather than searching for a single ``best'' plan, the entrepreneur should deliberately surface at least two viable strategic alternatives (for example, distinct customer segments, commercialization partners, or go-to-market architectures), design low-cost but high-fidelity tests that differentially inform their relative attractiveness, and only then commit to one path. The PillPack example illustrates testing multiple customer segments in parallel to reveal different value propositions; Amazon's early category experiments show how low-bar tests across many product categories can identify a dominant wedge; and Juicero and Fyre Festival illustrate the downside of committing to a single untested path. In my Bayesian language, ``Test Two, Choose One'' is a stopping rule: we update our priors about each strategic alternative using experimental data until the posterior difference between the best and the next-best alternative is large enough to justify irreversible commitment. I also understand your emphasis on segmentation and evaluation as a micro-foundation of this process. In my notes I summarize your role as focusing on identifying the ``best segment for persuasion'' across customers, investors, and other stakeholders, and on evaluating ventures along multiple outcome metrics that align with the four axioms and the compass rather than a single financial metric. In my research, I translate this into a Bayesian model in which each segment has its own prior over idea quality and strategic fit, and experiments are designed to update these segment-specific beliefs. Your distinction between epistemic uncertainty (fundamental uncertainty about the underlying quality of the idea) and business risk (quantifiable variation in outcomes under a known model) further shapes my modeling choices: early-stage entrepreneurial problems live primarily on the epistemic side, where learning is about the model itself rather than about parameters within a fixed model, which justifies my focus on formalizing belief revision and model comparison rather than only on variance reduction. Finally, I see your integration of the compass with my operations-oriented metrics as a guidance system for pacing experimentation and commitment. When I translate the four entrepreneurial strategies into ``speed ratio'' (implementation time divided by testing time), ``feedback ratio'' (feedback time divided by total cycle time), and ``learning ratio'' (cumulative understanding divided by theory-decay), I understand you as suggesting that different quadrants of the compass require different balances between acting then learning versus learning then acting. For example, I interpret Intellectual Property strategy as emphasizing ``learn then act'' with market-viability testing under strong appropriability, while Disruption strategy emphasizes ``act then learn'' with rapid go-to-market execution under weaker appropriability. This mapping helps me design experimental policies and operational architectures that are coherent with the chosen strategic quadrant instead of treating experimentation, operations, and strategy as separable decisions.