Concerns About Blockchain for Science
This is a cross-post from Medium.
I know, I know: I wrote about blockchain for science just last summer — this blog will explain why I now consider implementing blockchain to “improve” science a mistake.
I am sharing why I changed my mind because the narrative is increasingly concerning me: The potential of blockchain for science is stressed (e.g., Nature, Physics Today) instead of what science needs. I admit to that fault too and hope to correct myself here. Regardless, I also might be making a mistake here because I am preempting a lot of actual implementations of the blockchain — so read critically and evaluate for yourself. :-)
What does science need? Sustainability. My main motivation in my daily work is: How do we do science for the benefit of science and keep science progressive for the next generations?
In my opinion, all problems science is facing right now are threats to its sustainability — as such I don’t disentangle nor rank them in importance. Access is just as important as data, which are both just as important as inclusion. And yes, keeping science sustainable for the next generations also means rigorous science should be fun, respectful, inclusive of everyone (diversity promotes progress), and personally rewarding. If science is a horrid thing to do, it will not instill passion in as many people (serious play does).
Some of the biggest threats to science’s sustainability that come to my mind (non-encompassing):
- Selective publication
- Misrepresentation of the research process (e.g., adjusting predictions after data are available)
- Limited, obfuscated, and unequal access to results
- Messy and restricted preservation
- Exclusivity of (participating in) scientific research
- Inequality in research
- Toxic research evaluations
Blockchain by itself does little to address these problems although it could solve some or even all of these issues. But is blockchain necessary to solve these issues, is the question I asked myself this last half year.
What bugs me about blockchain is that it amplifies inequalities in the current system by increasing artificial scarcity and makes the “free market” fundamental to anything it touches. As such, I think it will entrench power structures even further, despite (potentially) shifting them to a new set of people.
Think about how the richest people in the world now also have the most power to buy Bitcoins or other cryptocurrencies. It seems antithetical to me that a commons would be based on free market principles if the entrance fee to that market has already been unequally distributed (note that Decentralized Autonomous Organizations, based on smart contracts, have the same problem).
Another property of the blockchain that scratches my brain is consensus. The blockchain network operates on a consensus principle, which makes entries to a logbook valid if the nodes reach >50% agreement on that entry (and as time passes older entries are harder to change; also called proof-of-work). But! There is always the possibility for a so-called 51% attack where the network is hijacked by a set of actors and starts adjusting the log.
Now there is much to say about why consensus is a good thing for application in finance and other domains, but why do we need consensus for a scientific log? If I produce a theory, I don’t see a need nor want for agreement of the network to have a theory entered into the log nor should the network have any theoretical possibility of adjusting my theory. Similarly, selectivity for confirming ideas might creep in by embedding consensus into the infrastructure. Consensus about heliocentrism in a geocentric world would be easy, but is not beneficial to progress.
But regardless of my personal uncomfortability with the blockchain; what are worthwhile properties of the blockchain and do we need blockchain for those? These worthwhile properties are immutability (although we saw consensus is a flaw there) and its decentralized nature (although the artificial scarcity restricts what can be stored in a decentralized manner). So technically, the properties are worthwhile but blockchain implements them in a way that is not suitable for a scholarly commons because it introduces more inequality instead of less.
The great thing is that immutability and decentralization are implemented elsewhere and are not restricted by consensus or artificial scarcity: The Dat protocol is one of those. The Dat protocol does not have any artificial scarcity (it scales to theoretically infinite storage) and it does not look for consensus but historic accuracy (content integrity). ScienceFair is already using the Dat protocol to rethink how we consume and share scholarly articles. I could explain more of the technical details in a blog sometime if people are interested in that (but in the meantime see the Dat white paper).
We can have an immutable and decentralized logbook for science (and any other domain, honestly) that does not require the blockchain at all. That is, if we care about historically accurate representation and do not deal with consensus, contracts and what not. As such, it seems that the detrimental properties of blockchain are an unnecessary aspect that run counter to what a scholarly commons aims to achieve (at least from my perspective): A widely distributed and extensive collection of knowledge that is true to its historical origins. How the Dat protocol can be implemented to address the threats to the sustainability of science, is something for another time (because this blog is already long enough).