🧬startup genome project #184
Replies: 6 comments 7 replies
-
history of this project:
|
Beta Was this translation helpful? Give feedback.
-
Isabella shared transformer code using in: 500 types of operations (word embeddings), 100 startup, M_t:1 mapping (established occupation would have larger heterogeneity in their operations and isabella's ratio was 20:1 (20k tasks, 1k occupation) so i set it as 5:1 out1: clustering 500 operations into 20 (following isabella's ratio (30:1) who clustered 20k tasks into 700types) out2(out1): 20 by 20 (correlation between operation cluster), (perhaps we could keep A, B, C from #184 (comment) and have 3 by 3 as well) out3(out1): 1k by 30 matrix |
Beta Was this translation helpful? Give feedback.
-
Today we discussed the problem of identifying the set of actions that were available to a startup retrospectively. I think there's a related challenge in this genetic view of organizational activity, making the analogy a little strained. In genetics, while a protein is assembling, the next letter in the sequence depends only on the next base pair on the helix. It doesn't immediately matter whether the string before is ATCGA or TTTTT. It might matter later of course. For the firm, on the other hand, each letter in an ABC sequence leads to a different state, and there's potentially feedback immediately (or at least soon). So the choice of the next letter is directly constrained by the letters that went before. Also, the appropriate choice is subject to variation in the state of the environment (competitor actions or market conditions). |
Beta Was this translation helpful? Give feedback.
-
I had chat with Jean-Baptiste Labrune on expative (this seems to explain biological exaptation well) Innovation, borrowed scenary, craft of gardens, humans situated cognition, lucille schuman's book plans and situated actions which starts from italian boat (only have desitination) vs german ship (map, gps). JB majored in computer science and evolutionary psychology, studied scientist, artist, children on how their reconfigure for creativity change of context, secondary adaptation (answer), birds have wing primary to maintain body temperature (ostrich, kiwi, penguins) but at certain moment flipped very fast which led to secondary adaptation which is flying. He explained different between co-opted adaptation. we identified common interest in startup evolution and agreed on continuous interaction in different forms like catching up on March, hiring as consultant (if we get grant). |
Beta Was this translation helpful? Give feedback.
-
applying ATGC coding to Silicon Valley: product1.csv with gpt_process1_gene, state, action, stochastic state transition algorithm as below:
product2.csv with gpt_process2_action_state_projection as below):
product3 is below gpt_process3_plot note that gpt analysis can go as detailed as below but I only asked representative state and actions.
NSSNail Phase: Season 1, Episodes 1-8. This phase covers the foundational period where the team focuses on developing their minimum viable product, establishing their business, and facing initial challenges in operational feasibility, technology development, and gaining early interest from potential customers and investors. Scale Phase: Season 2, Episodes 1-10, through Season 3, Episodes 1-10. The Scale phase encompasses the startup's efforts to expand its operations, navigate funding and legal challenges, improve technology feasibility, and address customer desirability through market engagement and product development. Sail Phase: Season 4, Episodes 1-10, through Season 6, Episodes 1-7. During the Sail phase, Pied Piper encounters new challenges and opportunities as it seeks to innovate, expand its market presence, adapt to competitive pressures, and align its operations and goals toward sustainability and growth in the face of complex industry dynamics. |
Beta Was this translation helpful? Give feedback.
-
my team mate (Bo Tu) from money for startup class is part of sculpting evolution group at mit media lab and his work on predicting protein characteristics from genotypes and the lab's philosophy in general gave me great insight e.g. https://www.sculptingevolution.org/research/accelerating-discovery. My hypothesis is "predicting protein characteristics from genotypes" technology can be re-configured to predict startup growth. Bo's recent paper: https://www.biorxiv.org/content/10.1101/2022.03.09.483646v4 |
Beta Was this translation helpful? Give feedback.
-
theory
treating$\theta$ as factors (harvard case study, pitchbook's company info (pitchbook_var.pdf), $y$ as observed success (investment, user metric, product-market fit), $\tilde{\theta}$ as gene, we are approximating $p(y, \theta)$ to $p(y, \tilde{\theta})$ assuming every recorded textual information on startup has as much as info as gene for the purpose of predicting startups's success
empirics
moving forward, will focus on transportation/mobility & healthcare sector. transportation accounts for low 5-15% of total startup (healthcare 10-20%, IT 30-40%) but considering charlie's expertise, requirement BitAtomEnergy management + ongoing collaboration with grab and 🛰️ mobility patterns and behaviors from GPS trajectory: smart-grab collaboration #183, let's start with this.
need and customer
align lessons learned from humanity's success and failures
solution and technology
tech1. statistical analysis from humane genome project to identify the effect of unit operations (A,T,G,C) in certain context (chromosome) and their mechanism (BRCA gene is related to stabilizing protein); auto-sequencer that enabled this analysis at scale ($1M to $1k); gene therapeutic treatment for Cystic fibrosis, Huntington's disease, Sickle cell anemia
tech2. navigator (online learning of driver preference, calibration under scare data in GPS #183) that recommends path (and pivots) using signal from satellite's (global positioning system)
fulfillment (:= producing implemented need-solution scale)
Beta Was this translation helpful? Give feedback.
All reactions