longer delay time of feeback, slower clockspeed, from physical led to market for tech or idea? pharmacetuicals has much more intricate decsion tree tools co-specialize in investment (sol, idea) but at the same time bm uncertainty? mvp being the result -> prototype and staged (how excited i ); experiment has externalits asymmetry -> successful experiment reveals less about the value of ideas than failed experiment learning infomation is not the sole purpose of experiment consensus-oriented process ⭐️you should find someone who hates you in the org when designing pilot (low and high bar) decision making (isolated) -> integral element of learning "Jana discussed the how some characteristics in deep tech affects the goal of experiments. Jana mentioned phsical. I'm just curious, how do you explain the causal relationships between the four factors: 1. physical component of deep tech, 2. slower feedback cycles, 3. specialized expertise, 4. existence of idea/tech markets? How do these factors influence each other, and which do you see as the primary driver in shaping the innovation process in deep tech domains?" This question: 1. Addresses all four elements (A, B, C, D) you mentioned 2. Asks for clarification on their causal relationships 3. Prompts the speaker to explain which factor they see as the primary influencer 4. Allows for a discussion of how these elements interact to shape deep tech innovation processes I think physical component makes deeptech industry clockspeed slower and make, and the different forms of experimentation that emerge? cost of entry and how long scott stayed in market? three layer of persuasion: - provide new data (don't look at the sample but look at other sample - cvg to same bm (drive)) - provide new model - provide theory based persuasion (selection) - instead of looking at others ~ adding new data, enhance model, change the diagnostics or model selection criteria Q1. hart "we suffer from over precision, we may want to be confidence biased, estimation bias" -> did you mean this for sustaining the business or persuasion or for learning? if duo is doing startup, the likelihood of saying yes (go) gets lower, so it's better to be more optimistic; bias (given, conditional on wha A Q2. principled way to set the threshold to increase or decrease one layer of prior? there is a tradeoff as computational cost is larger e.g. when estimating, "collapsing prior over prior in one prior" is more efficient as estimator's variance becomes lower, AER (princept auto levi ) -> collecting info (drama); Q3. Jana's talk explained some character of deep tech affects the goal of experimentation I'm just curious, how do you sequence the causal loops between the four factors? 1. physical component of deep tech, 2. slower feedback cycles, 3. specialized expertise, 4. existence of idea/tech markets? maturity state of technology (what level we're in); state2action ratio (geared to product) - If you only had over-precision, you might not adapt your business plan even when faced with evidence that fewer people like ice cream than you thought. - But because you're also confidence biased about your ice cream's quality, you still believe you can succeed. - This confidence pushes you to keep trying and experimenting, even though you're resistant to changing your initial beliefs about the market.