The OpSci Open Web Fellowship is a 12-week program that provides a stipend for talented undergraduates, students, and postdoctoral scholars to work on open-source software development that align with Opscientia’s mission.
Jakub is an Open Web Fellow and works on simulations for science communities that coordinate value flow using abstracted representations of digital science objects, called tokens. He details his experiences in the Opscientia Open Web Fellowship below.
OpSci is a community of inspiring people who actively work on solving the challenges facing science. Looking into the future, OpSci is synonymous with a framework in which science is and should be conducted, leading to a fair playing field where researchers are rewarded for collaboration and not constrained by institutions.
I am currently an undergraduate mathematics and physics student with particular interest in neuroscience and machine learning. I found out about OpSci through my KERNEL interview and have been instantly captivated by its mission and the incredibly smart people in the community. I believe the decentralization of science will play a key role in future scientific advancements and am eternally grateful to play a small part in this journey.
What is the problem with science? This question has a number of possible answers, but the one I chose to focus on in my Open Web Fellowship is this: value flow. The current scientific value flow has a number of issues that become apparent when we take a closer look at it.
To do this, consider the following questions: Where does value start in a science ecosystem? What happens to it? Where does it end up?
Any scientific research project requires funding, so it is no surprise that all value is first concentrated in grant funding agencies (e.g. NIH), which distribute funding between researchers.
Once researchers have the funding they need, it is used to conduct research, which includes buying all the necessary resources (e.g. data, equipment, etc.), but also to publish the results once the project is finished. Note that at this stage, value is transformed from financial assets (the grant money) to knowledge assets (IP, data, algorithms, etc.). Research papers are usually published to scientific journals, which are centralized agencies with decades of trust and prestige within the scientific community, hence researchers are incentivized to pay to publish their knowledge assets into these centralized agencies.
It might be obvious at this point that the current scientific value flow is surprisingly linear. Value flows from grant funding agencies and is eventually mostly locked within centralized knowledge curators (journals), which take in value both in the form of money (you pay to get published) and in the form of the knowledge assets that are produced by the research project. Now you might think, isn’t this an oversimplification? Yes, it is, there is undoubtedly some leakage of value at all stages of the value flow, but that unfortunately doesn’t change the fact that a centralized pool of value eventually becomes another centralized pool of value. Plus, we haven’t considered another unfortunate reality; researchers are often employed by centralized agencies (e.g. universities or other private research institutions), which bind them by some contract that a portion of any intellectual property they produce will be owned by that agency.
The centralization of value can (and does) introduce a number of inefficiencies. For instance, if I am a researcher and have spent years collecting valuable data, that data probably doesn’t belong to me, so I have little to no control over what happens to it. Furthermore, with the little control I do have, I will not want to share that data since it represents my potential competitive advantage in getting future grants for additional research and for getting recognition within the scientific community when I publish papers on it. Essentially, the current flow of value does not incentivize collaboration and data sharing, which is inefficient, because what if somebody wants to do research on data that has already been collected, but isn’t available to use? It means the data needs to be collected again, which requires resources that could have been used on processing the existing data.
Putting everything together, we identify the following problems with the current science value flow:
- linear flow of value
- value is centralized
- research is dependent on centralized agencies
The aforementioned points lead me to my Open Web Fellowship research project, Simulations for Science Token Communities, which aims to identify possible solutions to these issues by developing and simulating new open science models.
The current state of science is unsustainable in the long run. Traditional storage solutions are way too costly to support the increasing supply of data, not to mention that they are all maintained by a handful of centralized entities, making it quite difficult to share data effectively. Lastly, as mentioned in the project summary, the science value flow is surprisingly linear, leaky, and there are many stages within the scientific pipeline that allow for misallocation of resources.
I chose to study science because I am excited by the unknown and its potential to improve the world, but speaking as somebody who wants to participate within the science ecosystem, I want there to be as little friction as possible so that our expansion of knowledge increases in efficiency rather than fails due to an ineffective system.
Overall, my research in science token communities (which actually falls into a wider category of meta-science) has shown me how incredibly complex systems-level engineering can be. The development of models such as the ones described here plays an important role in understanding science from a high point of view, but is not alone sufficient to solve all the problems of current science value flows.
As with many research projects, I conclude my Open Web Fellowship with more questions than when I started, which provides a perfect starting point for advancing this research further. I plan to continue working on this project by connecting it to the research done in fields such as behavioral economics and reinforcement learning to lay the groundwork for designing science systems maximizing a common objective function. TokenSPICE will serve as a useful tool for validating these new models and I look forward to extending its functionality to more general problems.
Entering the Web3 space is a very unique experience, there isn’t a step-by-step guide and I can’t say that anything else is quite analogous to suddenly finding yourself in a place where everybody is actively working on solving the world’s biggest problems. I have had an incredible time working with Shady El Damaty and Richard Blythman, who have been supporting me throughout my fellowship and from whom I have learned so much. I believe I’ve only been able to scratch the surface of what is possible in science token engineering and I am excited to continue contributing to open science.
Are you interested in diving into the details? Take a look at the Technical Excursions on Simulations for Science Token Communities to follow along with my model building experience.