2024 summer project

[hyunjimoon]

this is a living documentation of how belief, style, goal of startup prediction project team evolves.

Goal

Designing a module of summer research so that it can be maximally helpful for Anup, Idana, Michael's career under uncertainty (academia, entrepreneurial, self-growth, interest, network)

Strategy

• elicit each individual's beliefs, style, goals (BSG)
• aligning BSG to form communal BSG by
  -- updating individual BSG through learning
  -- find higher dimension where A(nup, ngie), I(dana), Michael's BSG can be harmonized
  

Tactics

outcome of our workshop (suggestion: 5/3 evening @ Dig) can be summarized as one question

1. It is succinct
2. The answer to it is falsifiable (don't take it literally)
3. It is important

example: Does Bayesian thinking drive entrepreneurial success?

• The question is clearly short
• It is falsifiable: in the most direct form, one can randomize training entrepreneurs on Bayesian thinking and observe their performance. Of course, one may find a clever way to answer this question without running an RCT.
• The answer to this question is critical and not obvious. It is critical because it informs us whether this whole research agenda is relevant and, if so, when/under what conditions. It is not obvious because entrepreneurship is a strange beast that comes in different shades. For example, one could imagine that being overconfident and non-Bayesian is exactly what you need for commercializing the right-tail high-potential technology-driven ideas. We have all heard of the disturbing antiportfolio of famous VCs.

Simulated collaboration

• Angie wishes to simulate different scenarios of synergies and have asked to members about their interest.
• Angie asked Anup to extract five keyword / research question from the grant (startup prediction tool) they written together, with updated prior on Angie's goal and belief on Michael: https://superurop.mit.edu/scholars/michael-sun/?scholar-cohort=&scholar-page=1
• Angie and Anup will each share this document with Michael and Idana asking them to leave a quick reply that could be helpful for strategy above

[hyunjimoon]

(technical feasibility) The answer to it is falsifiable
(desirability) It is important
(operational feasibility) why angie can do well with the support of collaborators

1. Does early stage startup pivot be realistically modeled with markov decision process?

technical feasibility

• due to the noisy and high causal density nature of learning, RL which prescribes action for each state that maximizes expected reward might be better than specific heuristic (e.g. 2 by 2)

• preparing simulator and writeup for class project (tested by Cathy Wu)

operational feasibility

• support from below three m3s teams on model development (inverse reinforcement learning) and application in singapore
  AI for Human Capital Development in Future of Work
  Integrated human-machine intelligence for economic growth
  Designing and Allocating Tasks in Human-Machine Teams

2. Does geometry of simplex algorithm's pivot, traversing feasible vertex to increase optimality, combined with real option theory help model early stage startup pivot?

• can it explain
  Kirtley20_pivot_entrep.pdf
  fine1_pivot.pdf
  Chen22_Programs of Experimentation and Pivoting for (Overconfident) Entrepreneurs.pdf

4. Does implementation of testing and routing function in entrepreneurship institution (e.g. pedagogy) drive entrepreneurial application?
   matt, abdullah, vikash