Servet Çizmeli - Where reproducible research meets the blockchain: Melda.io and ARTiFACTS - #BFSCON
Watch on YouTubeVideo summary
Willie Amici, an environmental scientist specializing in ocean color analysis using satellite imagery, highlights a critical gap in modern scientific practice: the lack of reproducibility in computational research. Having spent years managing vast datasets and writing manuscripts, he observed that while science aims to build upon each other's work regardless of profit motives, current articles are often technically difficult to replicate. Since nearly all contemporary research is data-driven, involving complex steps from data collection and coding to hypothesis testing and publication, there is significant room for errors that prevent others from verifying results. To address this, Amici and his team in Turkey have developed a cloud-based, multi-tenant secure environment designed specifically as a reproducibility helper, bridging the gap between theoretical blockchain projects and actual researchers who need practical tools.
The platform functions similarly to a social media feed where public scientific projects are displayed, allowing users to instantly view, fork, and run code from other researchers' work while respecting data permissions. The core innovation lies in its ability to streamline the entire research lifecycle into a simple "one-click" process; users can upload their data, write Python code, and attach an article or research paper, with the system automatically handling execution and verification. This approach solves common struggles with legacy code and package management by ensuring that all components—data, code, and academic logic—are integrated in one secure location, enabling anyone to run another person's project without needing to manually recreate the environment.
To further enhance trust and permanence, the system integrates blockchain technology through Blockworks' proof-of-existence service to create immutable timestamps for scientific documents. In a demonstrated use case involving spectroscopy data on coffee samples, the platform generated a cryptographic hash of the project file once it was processed, effectively advertising that this specific version of the research existed at a precise moment in time. Beyond simple archiving, the team aims to introduce new document types that are not just static PDFs but executable artifacts, allowing for separate attribution of tables, figures, models, and results. This granular citation system would revolutionize how scientific contributions are credited, moving beyond traditional author lists to recognize specific components of a reproducible study.
Ultimately, the collaboration between Melda.io and ARTiFACTS seeks to bring blockchain technology providers closer to the actual needs of the research community by focusing on usability rather than just technological novelty. By targeting young researchers who are eager to adopt new technologies, the project hopes to foster an ecosystem where scientific documents are inherently executable and verifiable. The speakers conclude that if blockchain developers integrate their tools with this researcher-centric platform, they can significantly advance the field of reproducible science, ensuring that valuable research is not only published but also preserved, accessible, and capable of being run by anyone, anywhere.
Read the full video transcript
hello everybody my name is amici's
Willie and I am an environmental
scientist I used to do a lot of data
analysis and data collection in the
field in the oceans for ocean color
analysis with satellite images and
during that time I worked a lot with
data day-in day-out and I did a lot of
data management and writing manuscripts
and say saving everything for another
iteration and everything so I have come
to the conclusion that from my own from
my own experience there is a big need
for reproducibility because to my
opinion science the reason why we've
published science because we want others
to take on our work and then put on top
of it regardless off we want to make a
profit or not this is the cynic one on
of scientific world but right now it's
based on articles and articles are very
technically difficult to reproduce I'm
talking about mostly computational
science because lab work etc are a whole
little deal but when you talk about even
in lab work you end up collecting data
and you end up finding yourself in front
of the computer trying to make sense of
this analysis and that means data driven
everything is data driven we are located
in turkey south of Turkey and we are
developing reproducible research tools I
have to fast forward because initially I
have been given 15 minutes but the and I
ended up by five minutes so I
to quickly say say that the four for the
scientific method to be implemented with
data analysis we need to have go through
a lot of steps and those steps include
the ability to do practical things like
getting data and then making research
reading papers getting their data
understanding them coding them finding
other people's cause integrating them
and all kinds of stuff that I'm not
gonna do the ones who are publishing
know all these details and modeling
hypothesis testing publishing until
publishing and interacting with the
community there are lots of steps and in
those steps there is always room for
reproducibility problems so what we our
contribution is that a cloud-based
multi-tenant secure scientific code
execution environment as a
reproducibility helper this is what we
are trying to provide the user with and
yeah I'm not going to go into too much
detail but what I'm gonna tell you is
that the blockchain community since the
two these two conferences I visited you
have a difficulty in getting in touch
with actual researchers there are a lot
of beautiful projects but how do we make
sure that researchers as researchers use
these technologies we are closer to the
researchers because the scientific
document authoring system but in an
executable format cloud based is very
close to the researcher and we want the
researcher to use that so through the
integrations we would like to first
enable our system to offer our users
abilities of the blockchain we are not
warp chain people but also offer to
blockchain technology providers to get
closer to the researchers and then
through usage if there is any merit they
will understand and they will they will
use it so that's what we did also with
this is our interface I'm gonna
demonstrate very quickly in a second
with artifacts last year we met here and
thanks to this conference we were able
to develop together collab collaboration
and integration and I'm gonna switch to
to it and then quickly demonstrate or
this is our main entrance page which is
like a social media like think that all
public projects everybody knows who is
doing what and then they can just
immediately go to somebody else's
projects fork it and then use it of
course if data is included in it not
included all those permissions are
honored so if but if your data is open
and you're doing open science then you
can you can just run the code of
somebody else and this is a very big
advantage I myself have been struggling
a lot with just running legacy code
working with package management and
everything so our aim is to do
everything one click one click like get
the data code and the article itself or
research paper and another click run the
whole thing and make sure we are going
to make sure that it is going to run
without problems so let's say that I
have a use case here it's a data science
project but we want it to become a
scientific publication platform is that
I am here looking at some spectra most
spectroscopy data on coffee samples this
is a basic project that has been done
by by myself with collaboration with
Peter to ponder and then I have a like
the article view of this one and I would
like to be able to publish this in our
network this is what you can do with our
platform and here you can run codes of
our and Python and then upload your
files in a C in our security environment
you can upload your files and have
everything ready like article publishing
data code and all the academic logic in
one place and we would like to also
empower this with blockchain technology
this is why we interacted with artifacts
and using blocks Berg's proof of
existence this is like very very initial
state like we are creating a proof of
existence for this document and then I
am going to make and then the proof of
existence is kind of being prepared I
have to reload the page and here I can
see that now the status is still waiting
we are waiting it to be processed and
then now it's ready so we can get a file
hash hash of the file or we can just go
to the link of the blocks works page to
see that we are going to advertise that
at this point in time we have this
project to ourselves so the next steps I
would like to mention about the next
steps also is that attribution also
through a citation system for our
service and then we'd like to make a new
document type and we'd like to state
that that document is executable it's
not just PDF that you will pull your
hair too
or reproduce it but it's an executable
thing we would like to introduce this
new concept and then we would like to be
able to actually get attribution for
tables figures models results and all
different things separately so why we
are here because if you are a blockchain
technology developer we'd like to
integrate and we think that we have the
potential to get you a step further
closer to the actual researchers and we
are particularly aiming young people
because they are the ones who are really
thirsty for new technologies okay thank
you very much
[Applause]
you