DeSci & DePIN

Scientific research occurs within a concrete (physical) network of infrastructure. This means that to succeed, Decentralized Science (DeSci) also needs a network of decentralized physical infrastructure including research labs, scientific instruments, etc. 

In web3, DePIN is the term that stands for such Decentralized Physical Infrastructure Networks. It is a term which was developed to refer to the convergence of Web3 tech and Iot, the Internet of Things. Originally, this was known as MachineFi to allude to Machine Finance.

MachineFi is the convergence of Machines and Decentralized Finance (DeFi). It points to Machines processing data that can be harnessed into a DeFi framework. Data is the new gold and thus a network of machines together with their data must be liberated from centralization. 

By mid-2022, crypto-venture fund Lattice was calling these real world networks Token Incentivized Physical Infrastructure Networks (TIPIN). This because the idea was to include token mechanisms to incentivize the decentralized deployment of physical infrastructure. 

By the close of 2022, crypto-market intelligence firm Messari conducted a Twitter poll targeting its users. The result of the survey is the term we utilize now, and thus the DePIN web3 term was created. For a deep-dive I recommend you to read the work of Lin and collaborators1 (2024).

To put it ultra simple, in DePIN, network participants deploy physical infrastructure which span services that users consume. This is possible as blockchain validators/miners process transactions linking deployment to service via token incentives.

  1. Lin, Z., Wang, T., Shi, L., Zhang, S., & Cao, B. (2024). Decentralized Physical Infrastructure Network (DePIN): Challenges and Opportunities. ArXiv. https://arxiv.org/abs/2406.02239 ↩︎

My understanding of DeSci

First assigment while being part of the DeSci EDU course offered by Molecule and Bio

DeSci, or Decentralized Science, foremost is a social movement attempting to correct several issues affecting Traditional Science (TradSci). Since the inception of the Open Science (OpenSci) movement, a lack of monetary incentives hampered its success as a first attempt at such correction. OpenSci truly never advanced past the creation of open databases and open access journals. Its most promising component, citizen science, to my judgement never completely took off. Even for the most developed aspect of OpenSci, open access journals, to publish in these the authors need to pay additional costs which are more often than not higher than in closed journals. Distributed Ledger Technology (DLT), public blockchains, and Web3 offer a new opportunity to achieve the yearned changes to the way science is (i) funded, (ii) executed, and (ii) communicated. The reason this modern opportunity is vital is since nowadays science is in a more regrettable status than it was when the OpenSci movement emerged.


Today TradSci has many calamities preventing advance progression and this renders it essentially futile when attempting to handle the genuine world issues which are collapsing our socio-economic and socio-ecological systems. Among the TradSci crises are: the lack of research reproducibility, rampant academic fraud, desperate need of mentoring, shortage of intellectual depth connecting across disciplines and a talent drain emergency. The purpose of DeSci is to redress these issues, it is a new attempt empowered by a new technology, namely DLT. Currently the space is developing fast and it is organized across four verticals: (i) funding and IP, (ii) research communities/DAOs, (iii) publishing & reproducibility, and (iv) identity and reputation. Different players within the space focus on developing products and services around these aspects but as one of the most successful applications of DLT is Decentralized Finance (DeFi), it is not surprising that the decentralization of funding has been the most developed so far. This has been done by interfacing DeFi with IP by means of tokenization which has the limitation of only being applicable to applied science which is most likely to produce IP. For this reason, research communities organized as DAOs have governance mechanisms which are usually linked to research funding and IP markets. The execution of research however, where the issues of reproducibility and a shortage of intellectual depth exist, has not been entirely decentralized or encouraged. Typically, only the ownership of data is discussed. Nevertheless, What exactly is data? Data holds meaning only within the context of a foundational theoretical framework (narrative); without this, they are merely sequences of symbols and numbers. A noisy stream of pseudo-data can jam any database conceivable, whether centralized or decentralized. Certain initiatives driven by AI-agent optimism are emerging, including Proof of Knowledge (PoK) and/or Proof of Invention (PoI); nonetheless, I believe an essential element is missing. In the scientific process, we engage in both “day science” and “night science” along with numerous playful lab activities/experiences that resemble art and gaming more than the rigorous implementation of scheduled experiments, data gathering, or discourse on proposal/papers.


Night science1 is an essential part of science. To achieve this, we require secure environments that foster it. Historically, prior to the crisis, this safe space was a supporting lab operated by an experienced
researcher/academic whose main goal was to guide young scientists, as both good and bad science are pursuits that span generations. It is in these environments that fresh concepts are formulated and understood. Science is an evolutionary process2 in which ideas gather evidence, gain acceptance, build prestige and are eventually expressed and translated into concrete technology (products). This is a cycle, in which normal, frontier and revolutionary science, function as wheels of advancement that produce technology3. This is the wheel that is currently damaged. To address the issue, we require not only DAOs centered on particular inquires, narratives, and products (day science) but also to enhance these initiatives with a Decentralized Public Infrastructure Network (DePIN) of laboratory environments where playful exploration, curiosity, natural history and unstructured hacker/maker activities can provide a chance for night science to complement the practice of day science. In my opinion, this is the most overlooked subject in DeSci.

  1. Yanai, I., Lercher, M. (2019) Night science. Genome Biol 20, 179. https://doi.org/10.1186/s13059-019-1800-6 ↩︎
  2. Hull D. (1988) Science as a Process: An Evolutionary Account of the Social and Conceptual Development of Science. University of Chicago Press: Chicago. ISBN 0-226-35060-4 ↩︎
  3. Kuhn, T. S. (1962). The Structure of Scientific Revolutions. University of Chicago Press: Chicago. ISBN 9780226458113 ↩︎

DeSci Bangkok 2024

After an amazing week in Bangkok, while meeting with the DeSci and Earth commons community I finally thought of an answer to Gerard ‘t Hooft‘s question to 零一学生.

When thinking about technological evolution and intelligent machines, Gerard asked us to think about the possibility of AGI-capable robots conducting scientific research/experiments. What would such science look like?

If we cannot understand the robots narratives (models) nor their chatting about measurements (data), how can we tell they are doing research? Can we think of a narrative-agnostic operational definition of scientific research?

When robots/machines use instrumentation to gather data from an experimental set-up and then generate new narratives by feeding the data to algorithms. Wouldn’t that be science? Notice, that as expected, such robots would be capable of construction. This means, such robots will ‘secrete’ technologies as a by-product of their evolution.

An operational definition of scientific research of such sort would also help us build research tools for a decentralized autonomous swarm of intelligent agents which can also instantiate such automated algorithmic research activity. All research narratives (questions too “important” and all other questions about Nature robots might have the interest to ask) are just that, narratives. Technological production on the other hand is a typical outcome of science. If robots produce new technologies, they do it by doing science.

Such a swarm of agents can also be human as it would be in a decentralized group of citizen scientists. These scientists, spanning from all generations, cultures, and walks of life might have non-overlapping narratives. The only common factor is Nature on planet Earth. Its patterns of land use over the surface, and the ecosystem services it provides over space, time, and levels of biological organization to render living conditions over such surface.

DeSci can meet ReFi only at the local level. At the human scale. Such a scale, is where communities assemble (eco-social system) and ecosystems make it possible (services). It is the loci where habitats are generated and re-generated by Nature’s constructor (the Arm). If members of these communities want to conduct scientific research, good tools need to make no assumptions about the reasons and interests driving the science. The tools should be relevant regardless.

未来研究的游戏化世界  | The Gamified World of Future Research

Can we do scientific research by playing games with microbes?

Looking forward to our summer school track on biotic games. Inspired by the works of pioneers like Raphael Kim, Roland van Dierendonck and Riedel-Kruse among others we will explore mutating Open Science Hardware into biotic game consoles.

In the mosaic image above, the “game over” subpanel is meant as a homage to these people (see https://www.opencell.bio/news/biotic-gaming-month for image credits).

For this summer X-Camp we plan to explore life over land and under water while using DIY microscopes and computer vision to learn from microbes about collective intelligence.

Kevin’s Thesis manuscript/paper accepted!

Finally! After a lot of work Kevin’s paper Spatial biology of Ising-like synthetic genetic network in BMC Biology!

 

Bi-stable (2-state) synthetic genetic networks running on round E. coli cells are coupled to other neighboring cells in a quasi-2D colony growing on agar using quorum sensing auto-inducer molecules. Left, you see bacterial colonies and on the right we see the Contact Process Ising Model (CPIM), a model of our exsperimental system. Top, Ferromagnetic interactions and Bottom Anti-Ferromagnetic .