Daily Balajisms - Truly reproducible research
Turning fiat science into crypto-science
Balaji Srinivasan sees truly reproducible research as the next step in open science and citizen science movements. Science is about independent replication, not prestigious citations, he says.
There is orders-of-magnitude difference between a scientific paper, that was published recently in a prestigious journal but wasn’t replicated yet, and Maxwell equations that have trillions of replications, as they underpin our electronic devices, and are reproduced every time we use a phone.
Bitcoin is a digitally-native form of money. Less obviously, blockchains are also the truth machine – establishing universal truth, where even sworn enemies can agree on who owned what amount of bitcoin at time t.
Everybody may believe in the dollar, but as Balaji says, the No. 1 rule of the rules-based order is that America makes the rules. Fiat is backed by men with guns. With bitcoin, the rules are set by math. And math is more rigorous than science.
Ossified and centralized fiat science (prestigious citations) underpins the fiat state. Public policy and regulations are set “because of science”. Mask don’t work, before they do, based on public health officials’ statements (“by fiat”).
Balaji uses a model of physical/scanner/digitally-native versions to assess levels of digitalization. Cash is physical, internet banking is a scanner version that has a bricks-and-mortar equivalent in the form of a retail bank.
Web3 protocols are forms of programmable money that disintermediated Wall Street and gave rise to Decentralized Finance (DeFi). Similarly, Balaji says, all science will become computer science – from computational archaeology to computational social science. And sooner or later, all science will be put on-chain and become Decentralized Science (DeSci).
There is a long-term effort in reproducible research, using tools like Jupyter Notebook to run code on data within scientific papers, instead of using just PDFs. The next step is truly reproducible research, where the scientific information supply chain is fully traceable, auditable and replicable on-chain and citing papers will become a function call.
Balaji mentions how even scientific papers can get cancelled in today’s social war. Therefore, even computational history is not enough. We need cryptohistory.
Crypto-oracles will verify that scientific data were collected and put on chain. The data can still have errors, but one can at least know what, when and by whom was put on chain. And the data on chain will be immutable. The problem will reduce to tracking reputation of crypto-oracles, and the data can be audited for anomalies with techniques like Benford’s Law.
Next step is using crypto-instruments that collect scientific data and put them directly on-chain, which will make controlling for issues like batch effects easier.
Balaji sees the Nobel Prize as a flawed and Eurocentric endeavor. He proposes Satoshi Prize for truly reproducible research and DeSci, that could be awarded also to pseudonymous researchers.