[This is an extension of my last post, Epochs of open science.]
A comparative analysis of the renaissance and web3 today would reveal a set of common elements: increased access to capital, greater diversity, and the behavioral norms of knowledge and resource sharing.
Without diving into the lit, we can reason that each of these has an independent effect on innovation.
Capital should be obvious; labor and materials do not come free.
Knowledge and resource sharing seem fundamental; you can't scale innovation by keeping ideas to yourself. And if there was ever evidence for the utility of idea sharing in innovation, the open source movement would be exhibit A.
Diversity may be less obvious, but it's quite intuitive with the following thought experiment:
If you could plot the flow of every conversation in a single geographical region; you'd get a path diagram d. You could then compare d across geographies by computing differences in language patterns1.
Now if you brought two geographies together [i.e., adding diversity to the conversation], you might get d2 OR you might get dd — where the range of possible paths expands by an order of magnitude relative to the level of difference between geography A and geography B.
Metcalfe's Law suggests the latter might be true. Which begs the question: what happens when the entire world comes together?
Personally, I don't consider it a coincidence that this precise set of elements was present throughout two of the great rebirths of modern human history.
Moreover, [and I'm not sure if it's a part of the equation or a product of it] art seems to be a factor that we shouldnt ignore.
Presumably, if you took a highly diverse group of ~1000 or so people, loaded them up with capital and told them to build something cool, a bunch of them will hand you art, another bunch will hand you dog shit, and a select few will hand you some wild display of human ingenuity that just might change the world.
- I use geography here for it for the sake of simplicity. Diversity certainly comprises many factors.
Contributing to DAOs has to be one of the richest experiences I've had in my career. I've learned so much so quickly. In a year it feels like I've gained half a decade of experience.
Being part-time with another full-time job is tough though. I got into a good groove of things for a while but I've recently had to go back into the office full-time, which threw a wrench in my daily routine. The adjustment back is... well, an adjustment.
It won't stop me from working on the things I'm passionate about though.
Speaking of, I've been diving deep into language models with Hugging Face lately. I've been working with my good friends at LabDAO to build a system called LION: linguistic informatics on organizational networks.
I'm going to write much more about LION soon. In simple terms, we aim to demonstrate the use LLMs as replacements for classical approaches to the psychological analysis of language data. Further, by applying new insights to organizational network analysis we hope to make novel discoveries for understanding organizational networks.
Another hope of mine is that this tool actually reduces the need for surveying in organizations. If LLMs can effectively answer questions about the organization whose data it is trained on, we may be able to ask the same survey questions we typically would an organization's workforce, and ideally get a correlated representation of those answers.
The potential for applying language transformers in psychological science is astronomically high and severely under explored. I'd like to change that.