Introduction of Balaji and the Network State
Balaji Srinivasan, is the former CTO of Coinbase, former general partner at Andreessen Horowitz, twitter.com/balajis and Balajis.com
The era of traditional academia is ending, and the new era of decentralized academia, decentralized media, decentralized science is beginning. The way that the nation state justifies itself is it says that everything it's doing because of “science”, This intersection is captured in the book, The Network State, and it's at thenetworkstate.com.
There's a huge difference between science in the sense of Maxwell's equations, which are independently reproduced, "science" in the sense of just peer reviewed publications, which are often not independently reproduced or even reproducible, and it mimics the form. The journal publication, the citation, and the prestige and so on, without the substance, which is the ability to do an independent replication. And this is this core difference between, "science" and actual science.
00:00 - Intro
01:30 - Introduction of Balaji & The Network State
06:45 - Key factors holding back scientific progress
14:17 - How to change the research & funding structure
21:00 - Commonalities between New York Times and the nature of the Scientific Journals
38:00 - How does scientific progress unfold?
44:50 - Can AI be a dominant driver here?
49:10 - How to make predictions in technology
54:40 - The cyclical approach of history
01:02:48 - Advice for builders in the DeSci space
01:04:29 - Last words, book advice & outro
There's actually only one thing that's more prestigious than science, you know, that is? Math. It may sound almost trivial, but we don't normally think of them as opposed, right? You know, the king and the queen of you know, like stem, right, that are to get our STEM science. One way of thinking about it is, crypto economics is based fundamentally on math.
It is based on the fact that we can go and download the Bitcoin blockchain and run a bunch of mathematical computations, to verify it to verify all the past transactions, same with the Ethereum blockchain.
And that's different from academic economics, which claims to be based on "science", where it has gotten to this realm where you cannot verify everything.
What do you think are the key factors holding back scientific progress globally?
The entire Vannevar Bush thing of centralizing research around and after the time of World War Two, has run its course. And so, the fundamental thing holding back scientific progress globally is the choke point. That is the US government.
So all of these funds that were appropriated and centralized for the federal government, all the research that was centralized there, there's a certain school of thought that comes where it's, you know, grants, and it's papers, there's choke points that happened.
How would you change that legacy research and funding structure
The NIH budget is actually pretty large, It's several dollars a year. Giving full control would essentially shatter the entire thing into many funding sources.
More effort should be on bringing down cost, increasing reproducibility and increasing automation.
Fund independent investigators and fund relatively low cost you know maybe it's like one to three years and unlike the traditional academia you don't have to write a grant where basically everything is done beforehand.
Give people like perhaps one to three years of funding okay, and it's usually like one person or maybe like a very small group, and they're trying to do the basic science and prove it out and then if it works that technological fire you know catches then they start a company, or they turn that into a company okay and if it doesn't then you know like they either find somebody who wants to continue the funding or they shut it down.
Commonalities between the New York Times and Nature, the scientific journal for us?
Absolutely. Academia moves slower than legacy media. The pre-internet power centers of the US, like: 1) DC, which is regulation
2) Harvard, as a mechanism for academia
3) Hollywood, which is film
4) New York, which is media.
DC, and academia and Hollywood even take years to multiple years to turn something around, a paper or a film, or especially a rule or regulation that can take multiple years to turn something around, media had a 24/7 metabolism, CNN and others, like media is used to shipping a newspaper every single day. So they were the only legacy institution that had the same metabolism and speed of people who are deploying on the internet.
Legacy media, they're attacking everybody they're attacking every country every movement within the U.S. everything that possibly could contest the US establishment whether it's France or Hungary or china or India or Russia or conservatives or centrist liberals or tech people or crypto people or whatever right there's like a new enemy every single day.
How does scientific progress unfold?
Replication is massively underestimated as a driver of the use of science.
Just as an example like you guys have used tons of GitHub you know based open source projects what do you'll download it you'll run it if it doesn't run on your computer it's an amateurish project. And you know like how many times do you quote replicate their claims you're literally executing the software and if it doesn't replicate or work on your computer it's useless.
Can AI be a common driver here?
It's complicated. Crypto can be sort of corrupted or repurposed for centralization with the CBDC's and so on, but AI can also of course be decentralized with federated machine learning or if you're seeing, you know there are knockoffs of dolly 2 and other things that happen very quickly you know they're not as good.
But they very quickly get out there and moreover these models are difficult to train but as you know relatively cheap to evaluate, and so they're actually just bags of coefficients, and probably in the fullness of time they'll either get reverse engineered or hacked or something like that, once it's known that it's possible for people to catch, and we are not sure of how defensible those things are.
How to make predictions in technology?
There's different there's to pull from history, and one of the things to consider an unsolved problem is to determine whether a historical analogy is accurate.
It's like it's not just an analogy, it's a mathematical analogy where you have a second or differential equation. Mathematical analogies are the best kind because there is an isomorphism between the objects here and the objects here, and manipulating these gives you insight into these and vice versa.
The cyclical approach to history
The cycle theory is a parametric curve where you're just going around the loop, and we just come back to where we are.
If you watch the collapse of civilization's podcast, sobering to realize that in a sense we're like the guys who got to the last level of a video game and were made the most successful up to this point.
Previous human civilizations to our knowledge had gotten to like space travel and so on, but that may just mean that like our final flame out is like really chaotic, so it's actually good to decentralize to have eggs in multiple baskets to get to other planets, and you know that's what a lawn talks about backing up humanity
One of the things Steve Jobs did with an apple is he built um he pushed secrecy why do you push secrecy, so you could have multiple teams that didn't know about each other that were working on the same thing, but they were innovative since they couldn't copy from each other right, and they didn't feel competitive with each other because they didn't know about each other.
Advice for people in Desci
Build a demonstration of truly reproducible research, take two statistics papers where one cites the other okay and maybe actually uses a function from the other, have them both be on chain
maybe with the data and stuff on our weave.
Just show how you did that that would be a huge breakthrough just focus on something very specific then we've got you know that proof of technological fire where we can show here's the paper here's another one it's like arpanet you know sending like the first packet right that first on-chain citation as a function call where you have composable science will be a milestone.
Use some kind of decentralized platform to show that you can get composable science that can get composable finance.
Show-notes of the episode:
A huge thanks again to Balaji for taking the time and for the valuable input - We hope you all can take something away from this episode on your DeSci journey.
Feel free to skip around the timestamps to find your favorite topics! Also, if you prefer listening to podcasts on audio apps, feel free to search for "The DeSci Podcast" on Spotify, Google Podcasts & Apple Podcasts.
If you want to join us on the show as a speaker or have guest ideas, please shoot us a mail at email@example.com
You can also find all links to our Spotify, Discord and Twitter here.
Special thanks to: