7 minute read

NFT Academic Marketplace

This is an idea I have been thinking about.

Mission

The mission of this NFT marketplace is to enable academics to:

  • Have more agency in their research by enabling them to rethink mechanisms of funding they depend on.
  • Increase accessibility and explainability in their research.
  • Think more carefully about the long-term value of their work instead of short-term status gains.

This document

This document is structured by highlighting the current problems in academia, a proposed model for what an NFT marketplace for academics could look like, the problems it would attempt to solve, and finally open questions to answer.

Current Problems

Quantity over quality

In some scientific fields, specifically in machine learning research, there is a growing community of people who heavily think about the number of papers published and the citations accumulated as a metric of success.

The intended effect of this is to improve “research productivity”, yet has really only led to a natural selection of bad science that has hindered the quality of peer review, poor scientific methods, and increased false discovery rate. The increase in the number of citations has had the intended effect of influencing others (along with other bibliometrics including h-index) and to reward researchers based on their impact. However, it has only led to an outrageous citation list in papers and reviewers requesting citation through peer review.

There is the notion of the Minimal Publishable Unit (MPU), which is the smallest amount of work or research that can result in a publication. Existing incentives (e.g. funding, promotions, academic/social capital, etc.) rewards the number of publications and having a publication in the major conferences, something that can be objectively measured, not the quality of papers, something that cannot be objectively determined. This inevitably pushes researchers towards the MPU.

This list of intended effects versus actual effects in the academic community is further highlighted in this paper. In summary, the obstacles the research community faces at large today is a function of the mis-alignment of intention to outcome and the existing system being extremely rigid and non-reactionary to change.

Funders boundaries

There is a growing community of individuals at larger funding institutions who are aiming to embody the “fund risky projects” mindset [NIH]. Problems with some of these models include the fact that many of the large rigid institutions create RFPs with the intention of academics / research organizations working on the projects that these institutions believe are valuable. This is a very top-down way of facilitating research and under the assumption that these institutions guide the way forward for what is to be worked on. Research should be more free flowing, and take a bottom up approach where the academics define what they believe is important (they are the one’s dedicating their life to a specific field) and receive funding to continue to push forward what they think matters.

Additionally, these rigid institutions that make up most of the funding in academia will take a very long time to adopt a true sense of “funding risky projects” and thus sets academics to be bound by the confines of what these funding agencies are willing to have them work on.

Institutions that are currently working on transforming the funding space for science and technology include: Actuate; Focused Resource Organizations; Emergent Ventures; Astera Institute; Activate. These institutions help to alleviate problems in academia by giving more agency to the researchers and focusing more on solutions R&D. However, in the current academic model, which does not value solutions R&D nearly as much as the other incentives (namely “Publish Perish”) , there is a large body of academics that will not make the transition to working on research with these solutions R&D institutions (nor are these institutions seeking those types of academics).

Within the bounds of the existing operations of academia, there needs to be better mechanisms for funding riskier research that do not depend on centralized institutions for green lighting projects. Researchers should have full agency in their decisions to work on research they believe can push frontiers.

Time spent on the wrong things

Academics/Principal Investigators (PIs) spend a large amount of time writing grant proposals and thinking seriously about the funding of their lab while simultaneously working on producing quality research.

Part of the reason why industry researchers enjoy their job more is that they do not have to worry about the funding of their research and can focus their time working on highly risky projects with their team instead of aiming to get funding to work on “semi-risky” projects that do not shift paradigms that we are looking for to transform human progress.

In an ideal world, PIs do not need to spend time working on grant proposals nearly as much as working on pushing the frontiers of their research into the world.

Accessibility and inclusivity

Current venues for publications are highly exclusive and have a high barrier for entry, making it challenging to increase the diversity of individuals contributing to a specific domain of research. Increasing inclusivity requires that the research field is more open to those contributing to research (especially to underrepresented/ disadvantaged groups). We cannot increase inclusivity in research while also increasing the number of (non-research) skills required for a publication.

There is no mechanism today, other than time, donations, and personal social platforms, for researchers to support other researchers’ work. Every act of support is out of selflessness and there is a lack of incentive for cross collaboration other than having your name on another paper. The reward system in this community is highly dependent on your ability to make your research well known and marketed.

This marketability factor to get into top tier journals or via Twitter has created a “create what people would want to hear” mindset versus creating what we need to push a domain forward in academia. An example of this is State of the Art (SotA) hacking in the machine learning community (a form of the new p-hacking) whereby researchers know that by producing a SotA paper will have a higher chance of publication than not when really this is not what the machine learning community needs to push forward the domain.

In an ideal world, I would like to make bets on other people’s work and be part of their success by supporting each other and seeking people’s work who I think could transform the space and I should benefit by investing time, money and knowledge into those people.

Explainability, interpretability, visualization and pedagogy

Conforming to the status quo in many academic settings can lead to success at conferences and diverging from this status quo inevitably puts you at risk for not performing well in the current system.

As an example, let’s look at the machine learning community which has a general paper format of introduction, background, theoretical foundation, empirical results, and conclusion. As commented in this paper, there is nothing wrong with this format, but what is wrong is the idea that there needs to be a thorough mathematical background in every paper and incentives reviewers to accept papers that conform to this format. This incentive makes papers less accessible to a broader community, and creates obfuscatory and irrelevant information in a lot of academic work.

There is a growing community of people who think carefully about the explainability and visualization of technical work by having beautiful visual art to go along with it: Grant Sanderson; Jay Alamar; Distill.pub and many more. Explainability in deeply technical concepts is a challenge for most academics because the artwork side of this community is either undervalued or a barrier to entry.

In an ideal world, there is a value in explainability of your technical work and incentives to be more explainable and naturally increase accessibility.

What could a new system look like with NFTs?

In a new setting, there exists a marketplace for NFTs, specifically for the academic research community that is able to be filtered by domain of research.

In this marketplace, academics treat their research lab as a “generative NFT” collection, where for each research paper / patent / blog / Twitter thread they publish, they create a digital asset for it.

This digital asset is a mechanism for the general public/ research community to invest directly into academics and earn profits based on the success from the research. In an ideal world, we would measure success based on the value that research is having in the real world or anticipation it could have long-term to shift a domain.

This mechanism provides a potential funding opportunity for the research lab itself the more value their collection accumulates long-term. There is opportunity for people who contribute / purchase more of the digital asset to have more governance / contribution to the research organization itself should this research organization decide to turn itself into a Decentralized Autonomous Organization (DAO) with the funding it generates from the NFT collection itself.

My hypothesis is that having the broader community involved in academia can bridge the gap between the academic graveyard and not enough people working on non-hard-tech problems to push forward deep-tech innovations that matter. What would be solved with this solution?

Publish-Perish Model

There are NFT collections that benefit by having many pieces of digital art, but the most successful collections have been the ones that have the highest quality. Incentive researchers to produce highly impactful pieces of digital art instead of quantity of digital art. Rarity has been deemed to be a useful way of driving this thus far in the NFT space.

Accessibility

Anyone can put their work up for sale in the form of a digital asset. For researchers who truly care about increasing accessibility, they can invest in smaller research organizations’ digital assets. There is no barrier based on journal credentials or any filtering. The “free-markets” determine who is more valuable vs a controlled subset of individuals.

Agency via funding

It is unlikely that a research lab will be able to fund itself solely on the value / profits generated from an NFT collection, however, this can potentially provide some funding for additional risky projects and eventually lead to researchers being self-sufficient on their research and no-longer require finding from centralized institutions that have regulations on what they can and can not work on.

Explainability

Bringing artwork / explainability into a very theoretical space will naturally increase accessibility by incentivizing individuals to create work that is more explainable to a broader audience and get rid of technical jargon that is irrelevant.

Why a separate marketplace and not just create collections on existing marketplaces?

My hypothesis is that the noise of gaming, the metaverse, and entertainment more broadly on existing marketplaces will be a deterrent for academics to take part in. If you have an exclusive environment for academics to share their work that is not affected by the noise of other communities it will likely lead to more success and participation.

Open Questions

How do we incentivize what makes an NFT more valuable? Another hypothesis I have is that just the “name” brand researchers would benefit from the beginning which does not help increase accessibility unless these name brand researchers are also purchasing digital assets of underrepresented researchers who have anticipated value.

References

ReThinking ML Conference https://rethinkingmlpapers.github.io/papers/

In Defence of the Paper (https://openreview.net/forum?id=MUR1oxCnvNs)

Academic Research in the 21st Century: Maintaining Scientific Integrity in a Climate of Perverse Incentives and Hypercompetition (https://www.liebertpub.com/doi/10.1089/ees.2016.0223)

Updated: