Prescriptive Bayes 2025 or 26 Stanconnect

[hyunjimoon]

Call for thematic sessions for Stan conference in June seems to be great opportunity to gather interest + mature our knowledge on BayesSD. Thematic session seems appropriate (although additional poster sound good as well)

Thematic sessions are an opportunity to dive deeper into a topic and comprise three talks (15 + 5 minutes each), with an additional 10 minutes which can be used for a panel discussion. Thematic sessions may be run in parallel. To submit a proposal for a contributed talk, submit a form by March 31st.

The following were recommended components of the committee:

A. Applications of Bayesian statistics using Stan in all domains
B. Software development to support or complement the Stan ecosystem
C. Methods for Bayesian modeling, relevant to a broad range of users
D. Theoretical insights on common Bayesian methods and models
E. Visualization techniques
F. Tools for teaching Bayesian modeling

Below are my blueprint

Title of session

• engineering the bridge btw inference and decision making
  Prescriptive bayes: parallel flow of upstream testing and downstream prescription
• simulation to calibrate decision making
• learn good calibrated decision

Description of session

(tbd)

Speakers and talk titles

bayes simulation libraries (Paul Burkner)

• engineering workflow for how uncertain components can be updated and orchestrated with sbc

• bad decision function, good inference -> bad decision / good decision function, bad inference -> bad decision

• min c'x (desire prior)

• sbc (Martin, Teemu prescriptive sbc), Bayesflow (Stephan, Paul), arviz (Oriol, inference datatype for sbc + hier. visualization), readsdr (Jair), mstan (Ryan)

• potential panel: Aki Vehtari, Andrew Gelman

• required component:
  -- B. Software development to support or complement the Stan ecosystem
  -- E. Visualization techniques

bayes simulation in business school (??)

• application-focused!!
• covid + econ model with analytic estimation workflow diagram
• 15.879 seminar (bringing data into dynamic modeling) with the focus of hier.bayes
• role of SD + stats as complex project simulator
  (using simulation-based structure in marketing)
• potential panel: Elea McDonnell Feit (Drexel, Marketing), Sergey Naumov (Penn state, Supply Chain Management), John Sterman, Hazhir Rahmandad(MIT Sloan, system dynamics)
• required component:
  -- A. Applications of Bayesian statistics using Stan in all domains
  -- F. Tools for teaching Bayesian modeling

data2draws2decision with hierarchical modeling (Tom Fiddaman, Angie Moon)

• a.k.a. actionable workflow, diagnostics-based prescription, end to end + hierarchy + application

• discrete (agent-based, IP) vs continuous (compartment, SD) vs hybrid

• upstream (estimation VS predictive) vs downstream (policy VS prescriptive)

• upstream: data2draws, downstream: draws2decision - how IP is blended in + large scale

• Tom is cooking a hierarchical + system dynamic model (e.g. chronic waste disease, demographic)

• computational + large scale optimization transportation engineering approach for EV infrastructure management #142

• potential panel: Vikash Mansinghka (discrete+estimation; demand prior elicitation), Alex Jacquiliat (discrete+policy; on-demand mobility, logistics decarbonization, societal application), Adam Elmachtoub (transportation and supply chain management, portfolio optimization), Garud Iyengar (systemic risk, asset management, operations management, sports analytics, and biology)

• required component:
  -- C. Methods for Bayesian modeling, relevant to a broad range of users
  -- D. Theoretical insights on common Bayesian methods and models

[hyunjimoon]

.

[hyunjimoon]

Robin Na whom I have continuously exchanged thoughts on Bayes expressed some interest! He asked some plans and to persuade him to be a co-organizer, I drafted:

• This can start from small in person MIT event (e.g. stanconnect) within one year.
• This involves planning, doing, reviewing/publicizing.
• I started a mailing list (bayes-in-business) and other than the AOM presenters here (todo), James Evans also expressed support.

[hyunjimoon]

our StanCon talk
Stancon24_Integrating Bayesian Inference and System Dynamics with Case Studies in Epidemiology.pdf
on integrating Bayesian inference with system dynamics represents an outflow of SD principles to the Bayesian community. This presentation serves as a stepping stone for the "engineering the bridge between inference and decision making" theme of the Prescriptive Bayes session. By demonstrating how SD can enhance Bayesian models in epidemiology, we're opening doors for similar integrations in other fields like marketing and supply chain management. These concrete examples of cross-pollination between SD and Bayesian methods are crucial for developing the "learning well-calibrated decisions" aspect of our proposed session, as they showcase the practical benefits of this interdisciplinary approach.

I need to learn more about SD community's initiative to invite outsiders and how it can contribute to expanding Bayes-SD communities i.e. #156 (comment)

[hyunjimoon]

AOM bayesian statistics audience analysis

based on basic info from https://cdmcd.co/yKQa3A,

Table 1: Audience Demographics Summary with Affiliations

Role	Count	Percentage	Affiliations
PhD Student/Candidate	19	34.5%	University of Gothenburg, Virginia Tech, University of Pennsylvania, University of Nottingham Ningbo China, Illinois Institute of Technology, University of Iowa, Xi'an Jiaotong-Liverpool U., Florida International University, KU Leuven, UNC Chapel Hill, The Advanced Institute of Management (Tongji University), WHU - Otto Beisheim School of Management, UVA Darden School of Business, Cornell University, Durham University Business School, Harvard Business School, University of South Australia, University Of South Carolina
Professor/Faculty	17	30.9%	Iowa State University, Brock University, University of Minnesota, EMLYON Business School, Lund University, University of Illinois Chicago, U. of Groningen, Vanderbilt University, Waseda U., California State University - San Bernardino, University of Minnesota, LMU Munich, Universidade Nove De Julho, Texas A&M University - Kingsville, University of Sussex, Sasin School of Management, University of New Mexico, Griffith University, UCL School of Management, The University of Sydney, California State University, San Marcos
Postdoctoral Researcher	3	5.5%	University of Gothenburg, University of Toronto, Ivey Business School (Western University)
Other/Unspecified	16	29.1%	The Chinese University of Hong Kong (Shenzhen), University of St.Gallen, U. North Texas, TU Dortmund University, MIT, Fudan University, Tongji University, Guanghua School of Management (Peking U.), U. of Minnesota Carlson School of Management, China Agricultural University

Total Audience: 55

Table 2: Session Sequence with Speaker Suggestions

Session	Speaker	Updated Suggestions Based on Audience
Introduction (25 min)
(1) History and Fundamental Concepts of Bayesian Statistics (15 min)	Andreas Schwab	- Provide a global perspective on Bayesian statistics development, appealing to the international audience - Use examples from diverse fields represented in the audience - Set up a collaborative online space (e.g., Slack channel) for ongoing discussions
(2) Why We All Should be Bayesians (10 min)	David Krackhardt	- Highlight cross-disciplinary applications of Bayesian methods - Discuss how Bayesian approaches can address common research challenges across different fields - Encourage audience to share specific problems in their research that Bayesian methods could solve
Executing Bayesian Management Research (40 min)
(3) Function and Value of Priors Distributions in Bayesian Analyses (10 min)	Andreas Schwab	- Explain how priors can incorporate diverse cultural and regional insights - Discuss strategies for collaborative prior development across institutions - Provide examples relevant to both early-career researchers and established faculty
(4) Step-by-Step Execution of a Bayesian Study & Comparison to Traditional Regression Analysis (15 min)	Mark Hansen	- Offer a clear comparison that accommodates different levels of statistical background - Use examples from management research that resonate with the diverse audience - Provide resources for further learning, tailored to different career stages
(5) Execution of Bayesian Analyses: Markov-Chain Monte Carlo Simulations and Available Software Solutions (15 min)	Anup Nandialath	- Discuss software options considering differences in institutional access and resources - Highlight open-source solutions to promote accessibility - Demonstrate how MCMC can be applied to management research questions
First Q&A Session (10 min)	All Speakers	- Encourage questions from diverse perspectives (PhD students, faculty, different research areas) - Address concerns about adopting Bayesian methods in various institutional contexts
BREAK (10 min)		- Encourage participants to network and form informal interest groups
PART II
Publishing Bayesian Studies in Management Journals (15 min)
(6) How to Publish Bayesian Studies in Management Journals (15 min)	Mark Hansen	- Discuss strategies for publishing in diverse management journals - Address challenges of publishing novel methods in different research cultures - Share success stories and lessons learned from published Bayesian studies
Bayesian Statistics and Methodological Change (10 min)
(7) Emerging Bayesian Communities and Resources (5 min)	Angie Moon	- Highlight both global and regional Bayesian communities and conferences - Discuss opportunities for virtual collaborations across institutions - Introduce resources suitable for different career stages and research areas
(8) Why Is Management Research Not Exploiting Bayesian Analyses More? (5 min)	Andreas Schwab	- Address adoption challenges specific to management research - Discuss how to overcome institutional and disciplinary barriers - Encourage audience to share their experiences and ideas for promoting Bayesian methods
Second Q&A Session and Final Remarks (10 min)	All Speakers	- Facilitate a panel discussion with audience members from diverse backgrounds - Develop an action plan for promoting Bayesian methods in participants' home institutions - Encourage ongoing collaboration and community building

[hyunjimoon]

actions to consider next year. angie would be more motivated with at least one acted :)

1. assign tables before (people sit behind)

2. record e.g. BDA and stan event in stankorea
3. write a mail to scott and erin - connecting bayes pdw with bayes ent so that pdw presenters could be invited to bec and scott and erin can come to bayes pdw next year (at least david's talk which i feel is good first takeaway )
4. make template for research question on strategy - data

5. count the # of picture taken for each slide
   i only measured from mark's

• interval estimate (2)

6. demonstrate gpt code

[hyunjimoon]

simulating chat with Tom and Jeff given desire "Cookbook approach of Bayes stats for management"

1. aligned vision: Cookbook approach of Bayes stats for management

2. choice of our culture

Treat any suggestion as proposal waiting for the others to react. Both the quality of proposal and acceptance function matters for our collaborative action. David Kantor's Conversational Actions, Patterns, and Structure:

3. informing history of actions (given the same belief and desire of bayes.stats. for management)

• from 2023 with Jeff | Bayes2Business
• from 2022 with Tom | BayesSD
• from 2015 with dad and now with Charlie | SD4SCM

Q. how has encoding of belief evolved?

technical feasibility

2021

• developed test quantity based simulation-based calibration (+ library)

2022

• developed Stanify: Bridging Statistics and Dynamic Modeling with Vensim, Python and Stan

2024

• using gpt to elicit latent states that matters for action
  angie moon rational meaning construction.pdf

desirability

2023

• analyzed system dynamics audience using system dynamics conference/seminar using poster FiddamanMoon23_BayesSD.pdf

2024

• analyzed business audience using AOM talk

• as in analyzed statistics audience using stancon2024 talk, I identified mixture model (between system thinkers vs non-system thinkers) measured by the enthusiasm to my talk

4. action proposal for 2025

preparing stanconnect event which can help us
-- frame the problem we'd like to solve (proposal: ## 🖼️framing)
-- identifying beachhead through interacting with speakers (pair of scholar and practitioner) from each of the four schools of thought

🖼️framing

choice frameworks from other science: precise but narrow (Operations), rich but paralyzing (Behavioral), elegant but detached (Systems), passive inference (Statistics)

5. resource

Tools

• stanify which translates structural model (ode-based) into stan file
• vensim

Theories/backbones

MackeyBarneyDotson15_CorporateDiv.pdf summarized as MackeyBarneyDotson15_CorporateDiv_MoonSum.pdf

Fiddaman25_BayesSD_Vensim.pdf

---

things to verify during first chat
q to tom

• meaning of x for create equations for priors?

q to jeff

• is jeff's structural model from Structural RBV - EMR 2014.pdf be under the umbrella of system dynamics' ode-based model?
• can vensim provide playground function (e.g. stan's playgroud: stan without installing stan)
• how'd you feel about "simulation-based calibration for entrepreneurial choice" for the title? some draft chapters

[hyunjimoon]

Proposal for a White Paper on Entrepreneurship as an Interdisciplinary Field

Introduction

Entrepreneurship is a complex, multifaceted phenomenon that benefits from multiple disciplinary lenses. Researchers and practitioners have long noted that there is no single agreed-upon definition or model of entrepreneurship, with various disciplines emphasizing different aspects ([Interdisciplinarity and Multidimensional of Entrepreneurship. Review of ...](https://ibima.org/accepted-paper/interdisciplinarity-and-multidimensional-of-entrepreneurship-review-of-the-main-definitions/#:~:text=Interdisciplinarity%20and%20Multidimensional%20of%20Entrepreneurship,of%20selected%20approaches%20to%20entrepreneurship)). This white paper will argue that combining insights from computer science, statistics, decision science, cognitive science, and traditional entrepreneurship theory and practice can provide a more holistic understanding of entrepreneurial success. By integrating these perspectives, we aim to address fundamental questions about how entrepreneurs make decisions under uncertainty, optimize with limited resources, and craft strategy in dynamic markets. The proposed white paper will serve as both a theoretical synthesis and a practical roadmap, demonstrating how interdisciplinary approaches can enhance our understanding of entrepreneurship.

Objectives

This proposal outlines a plan to produce a comprehensive white paper. The key objectives are:

1. Integrate Diverse Perspectives: Synthesize theories and methods from computer science, statistics, decision science, cognitive science, entrepreneurship research, and entrepreneurial practice into a unified framework for understanding entrepreneurship.
2. Engage Leading Experts: Structure outreach to and collaboration with domain experts – including Vikash Mansinghka (computer science), Andrew Gelman (statistics), Charlie Fine (decision science), Josh Tenenbaum (cognitive science), Scott Stern (entrepreneurship theory), and a top practitioner (e.g. Reid Hoffman, David Siegel, or Bill Aulet) – to incorporate frontier insights and ensure interdisciplinary accuracy.
3. Design an Interactive Framework: Develop an interactive, visual decision tool that demonstrates how changes in key assumptions influence the choice between two strategic paths: (a) a long-term interdisciplinary collaboration roadmap and (b) immediate actionable research steps. This will help readers see how different conditions call for different approaches, making the white paper’s insights more tangible.
4. Emphasize Resource-Constrained Optimization: Provide a theoretical synthesis centered on the idea that entrepreneurship is fundamentally about resource-constrained optimization. We will incorporate concepts of Bayesian rational decision-making, probabilistic programming, and entrepreneurial strategy frameworks (e.g. Blitzscaling, “Nail it, Scale it, Sail it,” and structured experimentation) to illustrate how entrepreneurs can optimally allocate limited resources under uncertainty.

Through these objectives, the white paper will formally bridge theory and practice, offering clear insights and tools for both scholars and founders.

Expert Engagement Plan

A critical first step is to engage experts from each of the relevant disciplines. We will structure personalized outreach emails and propose collaboration formats tailored to each expert’s background and interests. The outreach will clearly state the white paper’s goals, highlight the specific intersections of the expert’s work with entrepreneurship, and invite their contribution (through interviews, written input, or co-authorship). Below is a plan for engaging each expert and the format of collaboration envisioned:

• Vikash Mansinghka (Computer Science – Probabilistic Programming): Collaboration Format: Invite Dr. Mansinghka to advise on modeling entrepreneurial decisions using probabilistic programming. The outreach email will emphasize how his work on probabilistic computing can model uncertainty in startup environments. Mansinghka’s group has explicitly explored AI capabilities via a combination of academic research and entrepreneurship ([Vikash Mansinghka | The MIT Quest for Intelligence](https://quest.mit.edu/about/people/vikash-mansinghka#:~:text=Vikash%20Mansinghka%20is%20a%20principal,Additionally%2C%20they%20collaborate)), making his input ideal for bridging computational modeling with practical venture experimentation. We will request a meeting or interview to discuss how probabilistic programming and AI might simulate entrepreneurial decision scenarios, and possibly invite him to co-develop a prototype of the interactive framework.

• Andrew Gelman (Statistics – Bayesian Analysis): Collaboration Format: Seek Prof. Gelman’s guidance on incorporating rigorous Bayesian statistical reasoning into our framework. An outreach email will note that entrepreneurship often involves decisions with little data, where Bayesian partial pooling can shrink estimates toward prior knowledge in data-sparse settings ([Andrew Gelman on the art of asking the right questions](https://medium.com/i-love-experiments/andrew-gelman-on-the-art-of-asking-the-right-questions-5b3e30786131#:~:text=Andrew%20Gelman%20on%20the%20art,closer%20to%20your%20fitted%20model)). We will propose a structured email exchange or call to discuss how Bayesian approaches (e.g. hierarchical modeling or decision analysis) can improve how entrepreneurs update beliefs with new evidence. His feedback will ensure our theoretical model properly accounts for uncertainty, avoids common statistical fallacies, and uses data effectively even with small sample sizes (a frequent reality for startups).

• Charlie Fine (Decision Science – Operations/Strategy): Collaboration Format: Invite Prof. Fine to contribute his perspective on decision-making under uncertainty and the “Nail it, Scale it, Sail it” framework. We will structure an outreach email highlighting our interest in his stage-based approach to the entrepreneurial journey and decision-making speed vs. deliberation trade-offs. Fine’s framework guides startups from idea validation (“nailing it”) through scaling and then sustaining (“sailing”) ([Innovation, Entrepreneurship & Leadership](https://jungle-mountain-ocean.squarespace.com/#:~:text=Based%20on%3A%20Nail%20It%2C%20Scale,It%2C%20Sail%20It)). We propose a joint workshop or roundtable (virtual) where he can share insights on designing decision roadmaps for entrepreneurs. His input will shape our roadmap for interdisciplinary collaboration (Path 1) by ensuring it aligns with how decisions naturally evolve as ventures grow.

• Josh Tenenbaum (Cognitive Science – Human Learning & Decision-Making): Collaboration Format: Consult with Prof. Tenenbaum on the cognitive science of entrepreneurial thinking. Our email will frame entrepreneurship as a form of complex problem-solving and learning, areas where his work on computational cognitive models is renowned. We will ask for an interview to explore how entrepreneurs form and test hypotheses (business ideas) in a manner similar to human concept learning, and how cognitive biases or bounded rationality might affect entrepreneurial strategy. Tenenbaum’s expertise in Bayesian models of cognition can inform how we incorporate realistic decision-making processes into our optimization framework (e.g. recognizing that entrepreneurs use heuristics or “approximate” Bayesian updates given limited cognitive resources).

• Scott Stern (Entrepreneurship Theory – Strategy & Innovation): Collaboration Format: Engage Prof. Stern to integrate formal entrepreneurship theories and strategic frameworks. We will reference his work on the Entrepreneurial Strategy Compass ([A strategic framework for start-ups - BusinessWorld Online](https://www.bworldonline.com/editors-picks/2018/07/18/174019/a-strategic-framework-for-start-ups/#:~:text=Authors%20Joshua%20Gans%2C%20Erin%20Scott%2C,technologies%20and%20started%20Dolby%20Laboratories)) and recent research bridging entrepreneurial choices with strategy and innovation policy. The outreach will invite him to review our synthesis of frameworks like effectuation, strategic positioning, and experimentation. A possible format is to have Stern as an advisor who reviews our outline or an early draft, ensuring we accurately represent entrepreneurship scholarship. His feedback will lend theoretical rigor – for example, making sure we discuss core strategic choices (customers, technology, organization, competition) that every venture must address ([ENTREPRENEURIAL STRATEGY](https://www.entrepreneurial-strategy.net/#:~:text=%E2%80%9CAt%20least%20four%20domains%20of,are%20crucial%20for%20every%20venture%E2%80%9D)). We will also ask for any unpublished insights or new frameworks (such as “Theory-Based Entrepreneurial Search” or others he is developing) that could enrich our white paper.

• Reid Hoffman / David Siegel / Bill Aulet (Entrepreneurship Practice): Collaboration Format: Connect with at least one high-profile practitioner to ground the white paper in real-world practice. For instance, if reaching out to Reid Hoffman, we will emphasize our focus on Blitzscaling (a term he coined) and ask for his perspective on when prioritizing hyper-growth is appropriate versus when it fails. To Bill Aulet, we would highlight overlapping interest in structured, disciplined entrepreneurship (as he outlines in his 24-step framework) and ask him to comment on how academic insights (like probabilistic decision models) do or do not resonate in practical teaching of entrepreneurs. Given these individuals’ busy schedules, the collaboration might take the form of a concise email Q&A or a short written statement that we can incorporate as a “Practitioner’s Insight” sidebar. Their contribution will ensure our actionable recommendations are realistic and address challenges entrepreneurs actually face on the ground.

Collaboration Format & Timeline: Each expert will receive a personalized invitation email, followed by a proposed meeting (virtual or in-person) or questionnaire. We will compile their insights through recorded interviews or written correspondence. To encourage an ongoing dialogue, we may host a virtual roundtable with all experts once initial insights are gathered – this would allow cross-pollination of ideas (e.g., letting Gelman and Mansinghka discuss Bayesian methods together, or Fine and Hoffman debate scaling strategies). By engaging experts early and iteratively, the white paper will effectively blend scholarly theory with experiential knowledge, increasing its credibility and richness. This collaborative approach is methodologically justified as it ensures interdisciplinary validity – each expert can correct any field-specific nuance and contribute methods or case studies from their domain, leading to a more robust final synthesis.

Interactive Framework Design: Assumptions, Paths, and Decisions

To make the theoretical integration tangible, we will design an interactive framework that visually maps how different assumptions lead to one of two strategic paths: (1) an interdisciplinary collaboration roadmap or (2) an actionable research plan. This framework will act as a decision-support or illustration tool within the white paper (e.g., an interactive graphic in the digital version or a decision tree flowchart in print). Below we outline the concept and methodological justification for this interactive component:

• Framework Concept: The interactive tool will present users with key assumptions or conditions about an entrepreneurial problem. As the user toggles or modifies these assumptions, the model will dynamically show which path is recommended – pursuing a broad interdisciplinary collaboration or focusing on immediate experiments. For example, assumptions might include: the level of uncertainty in the market or technology, the availability of resources (time, funding, data), the magnitude of the opportunity, and the desired time horizon for results. The framework will encode logical decision rules (drawn from our theoretical synthesis) to map combinations of assumptions to the optimal path.

  • Path 1: Roadmap for Interdisciplinary Collaboration – This path would be selected under assumptions where long-term, foundational integration is key. For instance, if the assumption is that the entrepreneurial challenge is highly complex, spans multiple domains, and the cost of isolated failure is high, the framework would advise building a collaborative roadmap. That might involve setting up interdisciplinary research partnerships, pooling knowledge across fields before attempting a solution, or developing theory through joint workshops. Essentially, Path 1 is about investing in knowledge integration upfront. It is suited for scenarios where the problem is not urgent to solve immediately, but understanding it deeply from many angles will yield superior strategies (e.g., developing a novel technology venture might benefit from months of joint research between cognitive scientists and AI experts to get the product right).

  • Path 2: Actionable Research Steps – This path is recommended when assumptions favor quick, iterative action. For example, if uncertainty is extremely high but the cost of experimentation is low (a common startup scenario), the framework will suggest diving into structured experimentation immediately to gather data. Path 2 means formulating hypotheses and running lean tests (A/B tests, prototypes, market trials) to learn rapidly, rather than first convening a broad collaboration. It is appropriate when time-to-market is critical or when the entrepreneur already has access to sufficient interdisciplinary knowledge to start testing (perhaps via an advisory board). This path aligns with approaches like lean startup or effectuation, where learning by doing takes precedence.

• Visual Design: The framework will likely be represented as a decision tree or decision matrix. For example, one dimension could capture assumptions about uncertainty (low vs high) and another about resource slack (ample vs constrained). This could form a 2x2 matrix where each quadrant corresponds to a recommendation (e.g., high uncertainty & low resources might highlight Path 2 – start experimenting, whereas low uncertainty & high resources might lean to Path 1 – coordinate experts to optimize strategy). We will incorporate more nuanced decision rules as needed (possibly using a simple probabilistic model behind the scenes to weigh factors). The visual tool will clearly indicate how changing one assumption (say, increasing available funding or extending the timeline) flips the decision from one path to the other. This sensitivity analysis helps emphasize that there is no one-size-fits-all entrepreneurial strategy – rather, the optimal approach is contingent on situational factors ([Strategy | Founder Challenge](https://www.thefounderchallenge.org/strategy#:~:text=The%20key%20tenant%20of%20strategy,proceed%20with%20your%20venture%20idea)). The interactive element thus educates the reader on why a particular method (collaboration vs. rapid experimentation) is chosen by making the trade-offs explicit.

• Methodological Justification: Incorporating an interactive framework is grounded in decision science and pedagogical best practices. It allows us to visualize complex conditional logic in an accessible way, enabling readers (or workshop participants) to engage with the material actively. By “test-driving” different scenarios in the model, stakeholders can validate whether the recommended path matches their intuition or experience, thus serving as a tool for reflection and refinement. Technically, building this framework will leverage insights from our experts: for instance, decision scientist Charlie Fine’s input can ensure the decision logic aligns with real startup decision-making patterns, and Vikash Mansinghka’s probabilistic programming expertise can help implement the interactive model in a Bayesian decision network form. This approach transforms static theory into an experiential learning exercise, consistent with our mission to make cutting-edge research simple, actionable, and impactful ([Innovation, Entrepreneurship & Leadership](https://jungle-mountain-ocean.squarespace.com/#:~:text=%E2%80%9COur%20mission%20is%20to%20make,%E2%80%9D)) ([Innovation, Entrepreneurship & Leadership](https://jungle-mountain-ocean.squarespace.com/#:~:text=1.%20Research,world%20applications)). In sum, the interactive framework is not just a visual aid, but an integral part of the methodology to illustrate how interdisciplinary thinking concretely alters entrepreneurial decision outcomes.

Theoretical Synthesis: A Resource-Constrained Optimization Perspective

At the heart of the white paper will be a theoretical synthesis that reconceptualizes entrepreneurship as a resource-constrained optimization problem, informed by advances in rational decision theory (Bayesian reasoning), computational modeling, and established entrepreneurial strategies. This section will lay out the integrated framework in a scholarly yet accessible manner, bridging concepts from our various disciplines. We will organize the synthesis around the idea that entrepreneurs are essentially trying to maximize expected success (whether defined as growth, profit, impact, or learning) subject to constraints on resources and information. Below we outline the key components of this synthesis and the contributions from each domain:

Entrepreneurship as Resource-Constrained Optimization

Entrepreneurs rarely have the luxury of abundant time, unlimited capital, or complete information; instead, they must make pivotal decisions under tight constraints and uncertainty. We will formally describe how many entrepreneurial decisions can be viewed through the lens of optimization with constraints. For example, choosing a go-to-market strategy can be framed as maximizing market traction given limited funding and a short runway, or deciding whether to pivot is akin to an optimal stopping problem under uncertainty. By casting these problems in optimization terms, we create a common language to connect different disciplines:

• Economics/Decision Science: Classic entrepreneurial problems have been described in terms of opportunity cost, constrained maximization, and trade-offs. We will reference research that highlights that entrepreneurship is fundamentally an experimental process with low odds of success and unknown outcomes until tried ([Entrepreneurship as Experimentation - American Economic Association](https://www.aeaweb.org/articles?id=10.1257/jep.28.3.25#:~:text=Entrepreneurship%20research%20is%20on%20the,We)). The ability to experiment is itself constrained by cost and time, which in turn affects how startups organize and innovate ([Entrepreneurship as Experimentation - American Economic Association](https://www.aeaweb.org/articles?id=10.1257/jep.28.3.25#:~:text=are%20not%20always%20made%20in,potentially%20very%20deep%20economic%20consequences)). We will incorporate these insights to show that the structure of an optimal entrepreneurial strategy depends on the cost of learning and the cost of failing. If experiments are cheap, a rational entrepreneur runs many experiments (exploration); if capital is limited, she must carefully choose where to focus (exploitation).
• Cognitive Science: Human decision-makers do not always solve optimization problems in a vacuum – they use heuristics and have bounded rationality. We will integrate the cognitive perspective that entrepreneurs use mental models and biases to navigate complex choices. However, research in cognitive science also shows that humans can often approximate rational Bayesian updates in many contexts. We will discuss how entrepreneurs form beliefs about their market and update them with evidence (customer feedback, prototypes) in a manner consistent with Bayesian reasoning (even if done intuitively). The concept of Bayesian rationality provides a normative benchmark: it describes how an idealized entrepreneur would update probabilities of success for various strategies as new data (experiment results) come in. This helps us frame well-known entrepreneurial behaviors: for instance, the “pivot or persevere” decision after a startup experiment can be seen as a Bayesian update – if the posterior belief in the current strategy’s success becomes low, a pivot (switching hypothesis) is rational. The cognitive science input (from Tenenbaum and others) ensures our model accounts for the fact that entrepreneurs must operate with limited cognitive resources, often relying on patterns, analogies, and prior experience to make near-optimal decisions quickly.

Insights from Bayesian Rationality and Probabilistic Programming

One of the novel angles of our white paper is bringing in Bayesian rationality and probabilistic programming to model entrepreneurial decision-making. We will articulate how Bayesian decision theory can enhance entrepreneurial strategy formulation:

• Bayesian Entrepreneurship: We will highlight recent work that blends Bayesian thinking with entrepreneurship, noting that even under deep uncertainty (Knightian uncertainty), Bayesian approaches remain robust. For example, Gans (2025) demonstrates that entrepreneurs’ use of heterogeneous prior beliefs and Bayesian updating of their strategies is consistent with successful venture formation, and that this holds true even when some uncertainty cannot be quantified ([Knightian Uncertainty and Bayesian Entrepreneurship | NBER](https://www.nber.org/papers/w33507#:~:text=This%20paper%20examines%20the%20relationship,uncertainty%20models%20can%20bridge%20different)). This finding also shows that formal Bayesian models can bridge different entrepreneurial methodologies, including effectuation ([Knightian Uncertainty and Bayesian Entrepreneurship | NBER](https://www.nber.org/papers/w33507#:~:text=incomplete%20preferences%2C%20it%20demonstrates%20that,can%20bridge%20different%20entrepreneurial%20methodologies)). We will explain in plain terms that an entrepreneur’s prior beliefs (about the market or technology) and their willingness to update those beliefs with new evidence can be a source of competitive advantage – those with strong but flexible priors can learn and adapt faster. This ties into effectuation theory’s idea of leveraging what you know and who you know, while remaining adaptable as new information arises.
• Probabilistic Programming as a Tool: With Mansinghka’s guidance, we will describe how probabilistic programming systems (like Venture, Stan, PyMC, etc.) can be used to create generative models of startup outcomes. For instance, one could write a probabilistic program that models customer acquisition as a stochastic process, or financing rounds as a function of market response, allowing an entrepreneur (or researcher) to simulate different strategies and their probability distributions of outcomes. By integrating probabilistic programming, our framework gains a computational experimentation capability – essentially, we can test “what if” scenarios in silico before testing in vivo. This is analogous to how Bayesian methods allow partial pooling between data and prior knowledge: in early stages (data-sparse) the model relies more on prior and theory, in later stages (data-rich) it relies on empirical evidence ([Andrew Gelman on the art of asking the right questions](https://medium.com/i-love-experiments/andrew-gelman-on-the-art-of-asking-the-right-questions-5b3e30786131#:~:text=Andrew%20Gelman%20on%20the%20art,closer%20to%20your%20fitted%20model)). We will argue that entrepreneurs can benefit from the same principle: early on, use simulations and prior knowledge (from experts or analogous industries) to guide decisions, and as real data comes in, gradually shift to data-driven decision-making. Probabilistic programming makes it feasible to encode prior knowledge from diverse fields (e.g. cognitive models of consumer behavior, statistical distributions of market growth, etc.) into a single model that can be continuously updated. This approach is methodologically justified as it forces clarity in assumptions and yields quantitative insights that pure verbal theory might miss. (For example, it can reveal if a strategy that sounds good on paper has only a small chance of success once uncertainty is accounted for.)

In summary, the Bayesian and computational piece of our synthesis provides a rigorous backbone: it treats entrepreneurial strategy as a process of sequential Bayesian updating and optimization, which can be approximated or automated with modern AI tools. This not only advances theory but has practical potential, e.g. a startup could use our probabilistic model to evaluate whether Blitzscaling or a lean approach is more likely to succeed given their current evidence and context.

Insights from Entrepreneurial Strategy Frameworks

Equally important in our integration are the rich frameworks and heuristics that have emerged from entrepreneurship research and practice. We will synthesize several such frameworks – Blitzscaling, “Nail it, Scale it, Sail it,” Lean/structured experimentation, and effectuation – showing how each can be interpreted through our unifying lens and how they complement the rational, Bayesian view. The goal is to demonstrate that these frameworks, while originating independently, can be connected and enhanced via interdisciplinary theory:

• Blitzscaling (Reid Hoffman’s framework): Blitzscaling is the strategy of prioritizing extremely rapid growth over efficiency in order to achieve market dominance, often under conditions of high uncertainty ([What Is Blitzscaling And Why It Matters - FourWeekMBA](https://fourweekmba.com/what-is-blitzscaling/#:~:text=Origin%20of%20the%20name)). We will examine Blitzscaling as an extreme case of resource-constrained optimization where the binding constraint is time (speed to acquire market share) rather than cash. In Bayesian terms, Blitzscaling might assume a prior that being first and big in the market yields disproportionately high payoff (network effects), thus justifying a high-risk strategy of spending aggressively now for potential huge reward. We will cite the key characteristics of Blitzscaling – speed over efficiency, massive scaling, and risk acceptance ([What Is Blitzscaling And Why It Matters - FourWeekMBA](https://fourweekmba.com/what-is-blitzscaling/#:~:text=LinkedIn%20and%20venture%20capitalist%2C%20to,their%20industry%20or%20market%20segment)) ([What Is Blitzscaling And Why It Matters - FourWeekMBA](https://fourweekmba.com/what-is-blitzscaling/#:~:text=scenarios,startups%20pursuing%20blitzscaling%20raise%20significant)) – and discuss in what scenario this is the optimal policy. For instance, our framework will clarify that if an entrepreneur’s assumptions (which can be tested in our interactive tool) indicate a winner-takes-all market with strong network effects and available venture capital, then Blitzscaling is a rational choice. Conversely, if those assumptions aren’t met (e.g. limited capital, or a niche market), Blitzscaling would score poorly in our optimization model, and a different approach should be taken. By tying Blitzscaling to formal criteria, we provide guidance on when to Blitzscale and when not to, adding nuance to Hoffman’s broad principles.

• “Nail It, Scale It, Sail It” (Charlie Fine’s stage-gate framework): This framework breaks the entrepreneurial journey into three phases – Nailing (validating the concept and achieving product-market fit), Scaling (growing the venture rapidly and efficiently), and Sailing (sustaining the business and navigating mature-market challenges) ([Innovation, Entrepreneurship & Leadership](https://jungle-mountain-ocean.squarespace.com/#:~:text=Based%20on%3A%20Nail%20It%2C%20Scale,It%2C%20Sail%20It)). We will incorporate this temporal segmentation by showing how our interdisciplinary perspective can inform each stage. During the Nail It stage, cognitive science and experimentation principles dominate (the goal is to learn as much as possible about customers and solutions with minimal resource use – essentially a constrained optimization focusing on knowledge gain). During Scale It, decision science and operations come to the fore: the venture optimizes processes, supply chains, and execution strategy, often requiring different metrics (e.g. optimizing growth per dollar spent, which connects to Blitzscaling if conditions warrant). In the Sail It stage, sustaining innovation and strategic moats become key – here we connect to strategy literature (e.g. the Entrepreneurial Strategy Compass trade-offs of collaboration vs. competition and control vs. execution ([ENTREPRENEURIAL STRATEGY](https://www.entrepreneurial-strategy.net/#:~:text=Strategic%20opportunities%20for%20startups%20can,be%20categorized%20along%20two%20dimensions))). We will illustrate these connections perhaps with a small table or graphic, and emphasize that Fine’s framework provides a roadmap over time, which our interdisciplinary approach can enrich by specifying what methods or theories apply best at each phase. Methodologically, including Nail-Scale-Sail ensures our white paper addresses the dynamic aspect of entrepreneurship – strategies are not static but evolve, and a good theory must account for this evolution.

• Structured Experimentation and Lean Startup (Eric Ries & scientific entrepreneurship): Many practitioners advocate for systematic experimentation in startups – formulating hypotheses and testing them with minimal viable products (MVPs) to gather validated learning. We anchor this in academic context by referencing that entrepreneurship has been described as a process of experimentation with highly skewed outcomes and unknown probabilities until tried ([Entrepreneurship as Experimentation - American Economic Association](https://www.aeaweb.org/articles?id=10.1257/jep.28.3.25#:~:text=Entrepreneurship%20research%20is%20on%20the,We)), and that the costs and constraints on experimentation deeply influence how much and what type of innovation occurs ([Entrepreneurship as Experimentation - American Economic Association](https://www.aeaweb.org/articles?id=10.1257/jep.28.3.25#:~:text=are%20not%20always%20made%20in,potentially%20very%20deep%20economic%20consequences)). This directly supports the lean startup ethos: if the cost to experiment is low, one should experiment broadly; if the cost is high or resources limited, experiments must be chosen more carefully (perhaps using tools like our probabilistic models to pick the most informative experiments). We will present structured experimentation as the practical counterpart to Bayesian updating – each experiment is essentially a data point that updates the entrepreneur’s belief about the business. We will highlight techniques like A/B testing, “Test Two, Choose One” approaches ([ENTREPRENEURIAL STRATEGY](https://www.entrepreneurial-strategy.net/#:~:text=Starting%20with%20your%20idea%2C%20use,for%20bringing%20forward%20that%20idea)) (an approach also promoted by the Entrepreneurial Strategy team to systematically generate and evaluate strategic alternatives), and metrics-driven iteration. By synthesizing these ideas, our framework will underscore that experimentation is not random trial-and-error, but can be optimized: entrepreneurs can decide what to test (and what to ignore) by considering expected value of information – a concept from decision science that aligns with lean methodology’s idea of prioritizing high-impact experiments. The white paper will likely include a brief case or example illustrating structured experimentation (perhaps drawing from a startup that pivoted successfully by following the data), to make this concrete.

• Effectuation and Adaptive Strategy (Saras Sarasvathy’s theory): We will also integrate the effectuation perspective, which contrasts with predictive rational planning. Effectuation posits that successful entrepreneurs often start with given means and allow goals to emerge, emphasizing affordable loss over expected return, and leveraging contingencies. Interestingly, effectuation can be interpreted in our framework as an adaptive heuristic for resource-constrained environments: when formal optimization is too hard (i.e. uncertainty is so high that assigning probabilities is futile), effectual logic says “do what you can, limit downside, and update goals as you learn” – which is a robust strategy under extreme uncertainty. We will discuss how effectuation aligns with the Bayesian/evolutionary view: it’s essentially a prior that favors survival (affordable loss ensures the entrepreneur lives to fight another day) combined with a readiness to completely alter course (which is akin to having a very open prior that allows for drastic updating). Moreover, we’ll mention the connection noted earlier that Bayesian models can coexist with effectuation ([Knightian Uncertainty and Bayesian Entrepreneurship | NBER](https://www.nber.org/papers/w33507#:~:text=incomplete%20preferences%2C%20it%20demonstrates%20that,uncertainty%20models%20can%20bridge%20different)) – formalizing uncertainty doesn’t negate the value of heuristics; instead, it can highlight why those heuristics work. Including effectuation ensures our interdisciplinary take doesn’t bias only toward formal planning; it recognizes the human, improvisational side of entrepreneurship and places it within our resource-constrained optimization narrative (as a strategy for when formal analysis is intractable or when resources are so constrained that one must create opportunities from what’s at hand).

By examining these frameworks (and others as relevant) side by side, the white paper will create a synthetic view of entrepreneurial strategy. We will show, for instance, how the Entrepreneurial Strategy Compass’s four generic strategies (Intellectual Property, Disruption, Value Chain, and Architectural strategies, born from the collaborate/compete vs. control/execute trade-offs ([A strategic framework for start-ups - BusinessWorld Online](https://www.bworldonline.com/editors-picks/2018/07/18/174019/a-strategic-framework-for-start-ups/#:~:text=Authors%20Joshua%20Gans%2C%20Erin%20Scott%2C,technologies%20and%20started%20Dolby%20Laboratories)) ([A strategic framework for start-ups - BusinessWorld Online](https://www.bworldonline.com/editors-picks/2018/07/18/174019/a-strategic-framework-for-start-ups/#:~:text=technology%20to%20product%20developers%20and,known%20for%20their%20high%20rates))) can be understood through Bayesian optimization: each strategy excels under particular prior assumptions about the environment (e.g. an IP strategy is optimal if one assumes strong appropriability and benefit from collaboration, whereas a Disruption strategy is optimal if one assumes being first and fast yields the lion’s share of rewards ([A strategic framework for start-ups - BusinessWorld Online](https://www.bworldonline.com/editors-picks/2018/07/18/174019/a-strategic-framework-for-start-ups/#:~:text=consider%20two%20competitive%20trade,such%20as%20Sony%20and%20Bose)) ([A strategic framework for start-ups - BusinessWorld Online](https://www.bworldonline.com/editors-picks/2018/07/18/174019/a-strategic-framework-for-start-ups/#:~:text=The%20Disruption%20Strategy%20involves%20the,Netflix%E2%80%99s%20model%20became))). This not only unifies theory but provides actionable guidance – an entrepreneur can assess which conditions match their situation and thus which strategy type to pursue. The key tenet of strategy that choosing one path means forgoing others (“you can’t be all things to all people”) ([Strategy | Founder Challenge](https://www.thefounderchallenge.org/strategy#:~:text=The%20key%20tenant%20of%20strategy,proceed%20with%20your%20venture%20idea)) will be a recurring theme, reinforced by our interactive framework that forces a choice between paths given a set of assumptions.

Methodological Justification for Theoretical Integration

Combining these diverse ideas under one roof is a challenging task, but our approach is methodologically sound for several reasons. First, interdisciplinary integration addresses complex phenomena more fully than a single-discipline lens. Entrepreneurship involves human behavior (cognition), uncertainty (statistics, decision theory), technology and innovation (computer science), and strategic interaction (business and economics); a valid theory must account for all these facets. By explicitly drawing on each discipline’s strengths, we mitigate each one’s blind spots. For example, pure economic models might assume rational actors with infinite compute, but cognitive insights introduce realistic bounded rationality; conversely, purely behavioral descriptions can be ad-hoc, but marrying them with Bayesian decision theory gives a normative baseline to measure against. This complementary pairing increases the explanatory power of our framework.

Second, our use of resource-constrained optimization as a unifying principle is justified by prior research recognizing resource constraints as central in entrepreneurship ([Entrepreneurship as Experimentation - American Economic Association](https://www.aeaweb.org/articles?id=10.1257/jep.28.3.25#:~:text=are%20not%20always%20made%20in,potentially%20very%20deep%20economic%20consequences)). It provides a common currency (everything has a cost or a trade-off) that allows us to translate concepts between domains. For instance, a psychological bias (like overconfidence) can be seen as mis-estimating probabilities in a Bayesian model; a strategy framework like Blitzscaling can be seen as changing the optimization objective (maximize speed, not efficiency). This cross-translation is only possible because we established a shared conceptual ground (optimization under constraints).

Third, the inclusion of expert practitioners and empirical frameworks ensures that our theoretical synthesis remains grounded and testable. Each framework we include (Blitzscaling, Lean, etc.) comes with real-world case studies and data. By aligning our integrated theory with these known patterns, we can validate the theory. For instance, if our Bayesian model predicts that “when X and Y conditions hold, blitzscaling is optimal,” we can cross-check this against known blitzscaling success cases (e.g. Facebook, Amazon) and failures. Likewise, if our model suggests a staged approach, we can reference how many startups indeed go through the “nail it, then scale it” sequence. This interplay between theory and practice (a form of abductive reasoning, oscillating between deduction and induction) strengthens the credibility of the white paper’s conclusions.

Finally, the methodology of building an interactive model and engaging experts in its creation acts as a form of peer review and validation. By coding our assumptions and seeing the outcomes, we are effectively testing the internal consistency of the integrated theory. By soliciting expert feedback, we catch any conceptual errors early and ensure the model resonates across fields. This process is akin to constructing an interdisciplinary proof-of-concept for the theory.

Expected Outcomes and Next Steps

Upon completion, this white paper will deliver several valuable outcomes:

• Comprehensive White Paper (Publication): A formally structured document with clear headers, rigorous yet accessible analysis, and cross-disciplinary citations. It will serve as a reference for scholars seeking a synthesized theory of entrepreneurship and for advanced practitioners interested in the theoretical underpinnings of their craft. We expect the white paper to be suitable for publication or circulation among academic and professional communities interested in innovation and entrepreneurship.

• Practical Frameworks and Tools: The interactive decision framework developed will be made available (e.g., as an online tool or appendix). This will allow entrepreneurs, educators, and researchers to input their scenario assumptions and receive guidance on whether to prioritize interdisciplinary research or action-oriented experiments. It effectively translates our theoretical integration into a decision support tool. Additionally, any probabilistic models or templates (for example, a Bayesian startup simulator) created during the project will be shared as open-source resources so others can adapt or build on them.

• Enhanced Collaboration Network: By bringing together experts from six domains, we lay the groundwork for ongoing interdisciplinary collaboration. The relationships formed (and the communication formats tested, such as the expert roundtable) can continue beyond the white paper. One possible next step is organizing a workshop or panel series featuring these experts, to further disseminate the insights and keep the dialogue going. Another is to initiate a research consortium or working group on “Interdisciplinary Entrepreneurship Research” that periodically meets to tackle emerging questions (for example, how AI and human cognition together influence future entrepreneurial opportunities, etc.). The white paper will include a section that outlines a roadmap for future interdisciplinary collaboration (Path 1 from the interactive framework), informed directly by what we learn from the experts about how to best work across fields.

• Actionable Research Agenda: The white paper will conclude with a set of actionable research questions and steps (Path 2 outcomes) that can be pursued immediately by interested researchers or by our team. These might include, for example: designing an experiment to test Bayesian decision models vs. human entrepreneurs in startup pitch competitions, or applying probabilistic programming to analyze a large dataset of startup outcomes to identify which strategic choices correlate with success under different conditions. We will ensure these recommended steps are concrete (who could do it, what data or methods are needed) and aligned with the gaps our synthesis uncovered. The intent is to spur follow-up studies that continue to build on the integrated framework, gradually strengthening it with empirical evidence.

In conclusion, this white paper will not only bridge theories from computer science to cognitive science to entrepreneurial practice, but also provide tangible tools and pathways for moving forward. By structuring the content with clear objectives, engaging the right expertise, and employing an interactive, assumption-sensitive model, we ensure the final output is both intellectually rigorous and practically useful. Entrepreneurship, as an interdisciplinary field, stands to gain from this integration – yielding insights that no single-discipline approach could produce in isolation. We are confident that this endeavor will enhance our understanding of how entrepreneurs can better navigate uncertainty and constraint, and ultimately, how society can better foster successful entrepreneurial outcomes through a fusion of ideas and methods across fields.