imagine someone invents a new form of
surgery this new technique promises to
treat a condition previously considered
untreatable now along with it they also
create new tools that enable one to
perform this new procedure then they
publish a research paper about it get
mentioned in the news and tens of
thousands of people find out about this
great new innovation that promises to
improve the lives of so many around the
world
however they keep the new tools that
they created to themselves in a case
like this how useful would you say this
new technique is everybody finds out
about it but very few people are
actually able to use it or benefit from
it now I think you can appreciate the
fact that being protective of tools that
enable new techniques will not only
limit the impact these techniques can
have but also it’ll prevent others from
trying them out and improving upon them
and I’m here to tell you that something
exactly like that is happening in
science and it’s decreasing our chances
of collaboratively solving some of the
most heart important problems we face
today I want to talk to you about the
role of openness and collaboration in
science specifically about the movement
of open science and within open science
open source just introduce what open
science is open science is the movement
dedicated at making all scientific
research and data openly and freely
accessible to everyone and one of the
six core principles of open science is
open source referring to open source
software and what is open source
software it is software for which the
source is made freely available to the
general public for use or modification
I’m currently a PhD candidate at MIT
working in computer vision and
the last six years or so I’ve had to use
or write some type of software for my
research and I want to share with you a
few things I’ve learned around open
sharing of software in scientific
research now you could say that science
has been open for centuries people
publish papers in journals and anybody
can read these journals and then they
can give credit to the original author
by citing their paper in any follow up
publications
however these journals aren’t really
open and what I mean by that
specifically is that if you ever want to
read a paper in one of these scientific
journals you have to pay often a
significant amount per paper and the
worst thing is that you even have to pay
to read a paper on research that was
funded by your tax dollars a portion of
the sixty five billion dollars in
government funding that goes to
universities and colleges every year in
the US now fortunately the open science
movement has actually made some progress
on this issue in the last few years in
fact there’s a bill in the House of
Representatives right now called the
fair access to Science and Technology
Research act or faster which would
mandate the public release of publicly
funded research which will hopefully get
passed and become law soon however this
bill says nothing about the software
that is created as part of these
publicly funded projects and why is that
a problem
in today’s day and age where computers
and software are becoming more powerful
by the day
software is beginning to drive a vast
majority of research out there we live
in a world where it is possible for a
researcher to simulate and analyze let’s
say how a nanoparticle releases drug
into the bloodstream
before any actual tests are conducted on
animals and oftentimes software exactly
like this is a key enabler of any
follow-up research in addition this
software tends to be fairly large
complex and tedious to replicate so for
example if a researcher like me had open
access to a paper about
some software-driven research I would
still have to spend several months if
not years just recreating that software
before I can replicate the published
results and as a researcher it is
critical for me to be able to reproduce
these results because then only can I
start building upon them now this really
undermines some of the key enabling
qualities of software which is one that
it’s easy to distribute to it produces
almost the exact same result on any
computer and three that it can be
quickly and easily updated and improved
not to mention the potentially huge
impact some open source software coming
out of science can have consider the
example of this machine learning library
called scikit-learn and some of you here
actually might be familiar with this the
original paper about this has been cited
over 7,000 times which is a lot in just
six years and it has enabled research
across a huge variety of domains all the
way from evolutionary biology to
astrophysics to psychology it even helps
power Spotify a service that many of us
use pretty much every day now I want to
clarify that for software that comes out
of privately funded research or software
that forms the basis of a new company
whether that’s made open-source is
purely a business decision I’m only
talking about software as it’s used in
academic research that mostly comes out
of publicly funded projects which if it
isn’t shared mostly we’ll just sit on a
hard drive somewhere before it gets
thrown out so this culture of being
protective of software that research is
based on is a stumbling block and really
slows down the innovation process in a
lot of fields and I can say this from my
personal experience my research has been
based at the intersection of two fields
which are culturally very different
computer science and mechanical
engineering now computer science on one
hand as a field has really adopted this
culture of open-source there are a lot
of researchers however in mechanical
engineering who are still protective
about their software
so they keep it to themselves access to
all of this software coming from the
computer science community has enabled
me to make very rapid progress however
to implement some of my ideas on the
mechanical engineering side I’ve had to
recreate over 10,000 lines of software
over almost three years and I’ve heard
repeatedly over the last five to six
years from many of my colleagues at MIT
over and over again that they needed
some software to advance their research
and they were not able to gain access to
it all this for software that has been
written multiple times in multiple labs
as part of multiple projects a lot of
which were taxpayer funded so perhaps we
need to start thinking about another
bill that would mandate the release of
software also that comes out of these
publicly funded projects now I want to
make a point that appeals to the
researchers here you guys might be
wondering well why is it that some
researchers in some of these fields feel
the need to be protective like this they
published papers openly but why not also
release software and the answer to this
question lies in the way the academic
community evaluates researchers
researchers get evaluated based on the
number of papers they publish and how
influential those papers are and one of
the strongest indicators of how
influential a paper is is how many
citations it has that is how many other
research papers make a reference to it
so researchers sometimes feel if they
keep the software they create to
themselves then they will be the only
ones who can publish more papers based
on that software and then more papers
will translate to a better academic
profile now what if I were to tell you
that it might actually be in a
researchers best interest to open-source
the software that they write there are
literally hundreds of examples of papers
out there with which software was
released which are also the all-time
single most cited paper of their authors
often by a significant margin now
citation data is not easily accessible
even a preliminary survey tends to show
that open sourcing your software can
actually get you a lot of citations and
if other people find your software
useful then you become more visible in
your community and that further pushes
your reputation plus it’s easier than
ever these days to publish a paper
solely about a software package through
new journals like the Journal of open
source software now some of you might be
thinking well making software shareable
and usable it takes a lot of extra time
and effort and I totally agree it does
but it seems like the more effort you
put into making your software usable the
higher the chances that it will come
back and get you a bunch of citations
and consider the fact for a moment that
maybe your papers aren’t really complete
without the software anyway because that
is preventing other researchers from
rapidly reproducing your results as I
close I’d like to ask you all to take
some action to help this cause of open
source in science as researchers and the
entire academic community you should
adopt the culture of open source and
released the software that you write if
nothing then for your own academic
profile this doesn’t always happen but I
believe that open source is one of those
few things that ensures everybody wins
and let’s try and change this culture
slowly because that type of change can
only come from within the community and
as citizens you should expect and demand
openness from taxpayer-funded research
this will ensure that appropriate
legislation such as faster gets drafted
and passed even if a single researcher
does not have to spend months
replicating software all of your tax
dollars going into research will go much
further and most importantly this will
ensure that you have the choice to
access any of these publications if you
ever feel the need to so just like that
surgeon who knows about a new technique
but does not have access to the required
tools researchers in a lot of fields
these days don’t have access to the
tools they need to advance their
research
and these days oftentimes that tool is
some type of software so whenever
possible let’s put the tools that
surgeons and researchers need back in
their hands and let’s all help advance
our society to become a place where not
just new techniques and discoveries but
also the tools that enable them are
openly shared so that we can not just
solve problems we face but also make
amazing new discoveries together thank
you very much
[Applause]
surgery this new technique promises to
treat a condition previously considered
untreatable now along with it they also
create new tools that enable one to
perform this new procedure then they
publish a research paper about it get
mentioned in the news and tens of
thousands of people find out about this
great new innovation that promises to
improve the lives of so many around the
world
however they keep the new tools that
they created to themselves in a case
like this how useful would you say this
new technique is everybody finds out
about it but very few people are
actually able to use it or benefit from
it now I think you can appreciate the
fact that being protective of tools that
enable new techniques will not only
limit the impact these techniques can
have but also it’ll prevent others from
trying them out and improving upon them
and I’m here to tell you that something
exactly like that is happening in
science and it’s decreasing our chances
of collaboratively solving some of the
most heart important problems we face
today I want to talk to you about the
role of openness and collaboration in
science specifically about the movement
of open science and within open science
open source just introduce what open
science is open science is the movement
dedicated at making all scientific
research and data openly and freely
accessible to everyone and one of the
six core principles of open science is
open source referring to open source
software and what is open source
software it is software for which the
source is made freely available to the
general public for use or modification
I’m currently a PhD candidate at MIT
working in computer vision and
the last six years or so I’ve had to use
or write some type of software for my
research and I want to share with you a
few things I’ve learned around open
sharing of software in scientific
research now you could say that science
has been open for centuries people
publish papers in journals and anybody
can read these journals and then they
can give credit to the original author
by citing their paper in any follow up
publications
however these journals aren’t really
open and what I mean by that
specifically is that if you ever want to
read a paper in one of these scientific
journals you have to pay often a
significant amount per paper and the
worst thing is that you even have to pay
to read a paper on research that was
funded by your tax dollars a portion of
the sixty five billion dollars in
government funding that goes to
universities and colleges every year in
the US now fortunately the open science
movement has actually made some progress
on this issue in the last few years in
fact there’s a bill in the House of
Representatives right now called the
fair access to Science and Technology
Research act or faster which would
mandate the public release of publicly
funded research which will hopefully get
passed and become law soon however this
bill says nothing about the software
that is created as part of these
publicly funded projects and why is that
a problem
in today’s day and age where computers
and software are becoming more powerful
by the day
software is beginning to drive a vast
majority of research out there we live
in a world where it is possible for a
researcher to simulate and analyze let’s
say how a nanoparticle releases drug
into the bloodstream
before any actual tests are conducted on
animals and oftentimes software exactly
like this is a key enabler of any
follow-up research in addition this
software tends to be fairly large
complex and tedious to replicate so for
example if a researcher like me had open
access to a paper about
some software-driven research I would
still have to spend several months if
not years just recreating that software
before I can replicate the published
results and as a researcher it is
critical for me to be able to reproduce
these results because then only can I
start building upon them now this really
undermines some of the key enabling
qualities of software which is one that
it’s easy to distribute to it produces
almost the exact same result on any
computer and three that it can be
quickly and easily updated and improved
not to mention the potentially huge
impact some open source software coming
out of science can have consider the
example of this machine learning library
called scikit-learn and some of you here
actually might be familiar with this the
original paper about this has been cited
over 7,000 times which is a lot in just
six years and it has enabled research
across a huge variety of domains all the
way from evolutionary biology to
astrophysics to psychology it even helps
power Spotify a service that many of us
use pretty much every day now I want to
clarify that for software that comes out
of privately funded research or software
that forms the basis of a new company
whether that’s made open-source is
purely a business decision I’m only
talking about software as it’s used in
academic research that mostly comes out
of publicly funded projects which if it
isn’t shared mostly we’ll just sit on a
hard drive somewhere before it gets
thrown out so this culture of being
protective of software that research is
based on is a stumbling block and really
slows down the innovation process in a
lot of fields and I can say this from my
personal experience my research has been
based at the intersection of two fields
which are culturally very different
computer science and mechanical
engineering now computer science on one
hand as a field has really adopted this
culture of open-source there are a lot
of researchers however in mechanical
engineering who are still protective
about their software
so they keep it to themselves access to
all of this software coming from the
computer science community has enabled
me to make very rapid progress however
to implement some of my ideas on the
mechanical engineering side I’ve had to
recreate over 10,000 lines of software
over almost three years and I’ve heard
repeatedly over the last five to six
years from many of my colleagues at MIT
over and over again that they needed
some software to advance their research
and they were not able to gain access to
it all this for software that has been
written multiple times in multiple labs
as part of multiple projects a lot of
which were taxpayer funded so perhaps we
need to start thinking about another
bill that would mandate the release of
software also that comes out of these
publicly funded projects now I want to
make a point that appeals to the
researchers here you guys might be
wondering well why is it that some
researchers in some of these fields feel
the need to be protective like this they
published papers openly but why not also
release software and the answer to this
question lies in the way the academic
community evaluates researchers
researchers get evaluated based on the
number of papers they publish and how
influential those papers are and one of
the strongest indicators of how
influential a paper is is how many
citations it has that is how many other
research papers make a reference to it
so researchers sometimes feel if they
keep the software they create to
themselves then they will be the only
ones who can publish more papers based
on that software and then more papers
will translate to a better academic
profile now what if I were to tell you
that it might actually be in a
researchers best interest to open-source
the software that they write there are
literally hundreds of examples of papers
out there with which software was
released which are also the all-time
single most cited paper of their authors
often by a significant margin now
citation data is not easily accessible
even a preliminary survey tends to show
that open sourcing your software can
actually get you a lot of citations and
if other people find your software
useful then you become more visible in
your community and that further pushes
your reputation plus it’s easier than
ever these days to publish a paper
solely about a software package through
new journals like the Journal of open
source software now some of you might be
thinking well making software shareable
and usable it takes a lot of extra time
and effort and I totally agree it does
but it seems like the more effort you
put into making your software usable the
higher the chances that it will come
back and get you a bunch of citations
and consider the fact for a moment that
maybe your papers aren’t really complete
without the software anyway because that
is preventing other researchers from
rapidly reproducing your results as I
close I’d like to ask you all to take
some action to help this cause of open
source in science as researchers and the
entire academic community you should
adopt the culture of open source and
released the software that you write if
nothing then for your own academic
profile this doesn’t always happen but I
believe that open source is one of those
few things that ensures everybody wins
and let’s try and change this culture
slowly because that type of change can
only come from within the community and
as citizens you should expect and demand
openness from taxpayer-funded research
this will ensure that appropriate
legislation such as faster gets drafted
and passed even if a single researcher
does not have to spend months
replicating software all of your tax
dollars going into research will go much
further and most importantly this will
ensure that you have the choice to
access any of these publications if you
ever feel the need to so just like that
surgeon who knows about a new technique
but does not have access to the required
tools researchers in a lot of fields
these days don’t have access to the
tools they need to advance their
research
and these days oftentimes that tool is
some type of software so whenever
possible let’s put the tools that
surgeons and researchers need back in
their hands and let’s all help advance
our society to become a place where not
just new techniques and discoveries but
also the tools that enable them are
openly shared so that we can not just
solve problems we face but also make
amazing new discoveries together thank
you very much
[Applause]
Please follow and like us: