ImpactStory Proposal - Publishable
ImpactStory Proposal - Publishable
ImpactStory Proposal - Publishable
18, 2013
bringing web-native scholarship into the academic reward system. We request two years of
● Scholars can create and publish diverse products beyond the traditional article: data,
● Scholars can publish these products on diverse new platforms like PeerJ, F1000 Research,
● Research can engage diverse people beyond the academy, through blog commentary,
The short-term result will be more open, more efficient scholarly products. In the long term, the
move to web-native scholarship will topple the paper-native model we still use today—
essentially unchanged since its 17th century birth (Cassella & Calvi, 2010). In its place, web-
native scholarship will instantly publish diverse products throughout the research process,
collectively curating them in active online communities (Priem, 2013; Smith, 1999).
The technologies exist to support this transition now, but scholars have been slow to adopt them.
Researchers face a recurring dilemma: even in situations where scholarship may be best served
by a publishing web-native product (a dataset, blog post, software repo, etc), one’s own career is
often better served by instead putting that effort into traditional article-writing. If we want to
move to a more efficient, web-native science, we must make that dilemma disappear: what is
good for scholarship must become good for the scholar. Instead of assessing only paper-native
articles, books, and proceedings, we must build a new system where all types of scholarly
The key to this new reward system will be altmetrics: a broad suite of online impact indicators
that goes beyond traditional citations to measure impacts of diverse products, in diverse
platforms, on diverse groups of people (Priem, Taraborelli, Groth, & Neylon, 2010). Altmetrics
leverage the increasing centrality of the Web in scholarly communication, mining evidence of
2
These and other altmetrics promise to bridge the gap between the potential of web-native
scholarship and the limitations of the paper-native scholarly reward system. A growing body of
research supports the validity and potential usefulness of altmetrics (Eysenbach, 2012; Haustein
& Siebenlist, 2011; Li, Thelwall, & Giustini, 2011; Nielsen, 2007; for more see the altmetrics
Mendeley collection1). Eventually, these new metrics may power not only research evaluation,
but also web-native filtering and recommendation tools (Neylon & Wu, 2009; Priem &
However, this vision of efficient, altmetrics-powered, and web-native scholarship will not occur
accidentally. It will require advocacy to promote the value of altmetrics and web-native
approach today, and an open data infrastructure to support developers as they create a new,
web-native scholarly ecosystem. ImpactStory aims to provide all three, kickstarting a transition
A number of products provide impact-tracking tools, but these are limited by the early stage of
Citation-based profiles
Thomson Research InCites, Elsevier SciVal, and Symplectic Elements are publisher-owned
tools that facilitate impact analysis for institutional subscribers. They offer limited altmetrics
1
http://www.mendeley.com/groups/586171/altmetrics/
3
There are a few free profile-based citation tools. Google Scholar Profiles provides article-level
citation data. Unfortunately, its impact data is completely closed, with no API or bulk download
access, so it cannot be built upon or enhanced. Microsoft Academic Search Academic Profiles
provides citation (not altmetrics) data and does provide an API, but its data remains too sparse
The PLOS article-level metrics (ALM) application is an open-source system used by PLOS to
display altmetrics on their own article pages. Data is open, but only available for PLOS journals
(other journals can also run ALM, if they install, host, and run their own copies, but to date none
have). Altmetric.com is a for-profit startup tool that, like ImpactStory, gathers and displays
altmetrics. However, it is aimed at publishers rather than researchers, and full access to data
Plum Analytics, another for-profit altmetrics startup, is beta-testing “PlumX,” a product that,
like ImpactStory, lets individual researchers see their altmetrics. However, Plum’s pricing levels,
source code and data are all closed. If a researcher’s institution does not subscribe (and currently
only two do), she cannot access her metrics, and neither can anyone else. Academia.edu,
ResearchGate, and Mendeley offer profiles for individual consumers. These for-profit tools
4
Synthesis
Except for PLOS ALM (which is scoped only to traditional journals), this space is dominated by
investor-backed, commercial organizations. This is not inherently bad; indeed, it illustrates the
growing demand for altmetrics tools. However, these market-driven organizations are building
on closed data and closed code, with limited incentive to invest in shared infrastructure. This is
unfortunate, because the full potential of altmetrics lies in a world of diverse services built atop
an open data infrastructure—a data commons to support not just a few altmetrics providers, but
ImpactStory was incorporated as a non-profit on December 18, 2012 by Jason Priem and Heather
It is the mission, duty, and purpose of the Corporation to create and distribute educational
tools, data and research so that the community can learn about the full online impact of
diverse research products. [...] Innovative publishers, institutions, funders, and scholars
recognize the limitations of the current approach. They want to track the full online
impact of their diverse research products, but lack tools and data. Other, more
conservative members of the community are interested in this approach but waiting to
learn more, or to be convinced of the value. The purpose of the Corporation is to meet
this societal need.
The initial Board of Directors consists of Priem, Piwowar, Cameron Neylon (Public Library of
Science), and John Wilbanks (Sage Bionetworks, Ewing Marion Kauffman Foundation). Tax-
exempt paperwork was filed in January 2013; notification of 501(c)3 status is expected in July
2013.
5
Since our origins in an all-night hackathon, ImpactStory has had the same two leaders. Heather
Piwowar published one of the first papers measuring the association between open research data
and citation rate, and has continued to publish on and work towards incentives for open science.
She has expertise in statistical analysis and ten years of software development and support
experience in small tech companies. Jason Priem coined the term altmetrics and has helped
define the research field growing up around it, organizing the first two altmetrics workshops, as
well as publishing and speaking widely on the topic. He has built several successful open-source
software applications, and has practical experience and formal training in art, education, design,
A vibrant, altmetrics-powered world of web-native scholarship will require early guidance and
Existing for-profit providers have approached altmetrics data as a commodity to be sold. This
of tightly guarded silos containing mutually incompatible data (an outcome we have already seen
The real value of altmetrics data, like other Big Data streams, is not in the numbers themselves;
the value is in what the community can build on top of the data: a new generation of web-native
assessment, filtering, and evaluation tools (Wilbanks, 2011). In this way, open altmetrics data is
6
quite like the network of open protocols and providers behind the World Wide Web: it is
to build and sustain this infrastructure—to create an altmetrics data commons. Our work will
take a three-pronged approach, focusing on advocacy, a free altmetrics webapp, and an open
Our advocacy will continue along the path set in our first grant. We will give talks at universities
and conferences, lead workshops, publish articles in high-profile venues like Nature, and pursue
more research describing and validating altmetrics. In doing so, we will bring growing attention
to the emerging altmetrics movement. More importantly, though, our advocacy will continue to
help shape the movement, keeping openness and interoperability central to emerging
conversations.
replacement for their online CVs. This “altmetrics CV” will support broad, grass-roots adoption
experimenting with web-native products like blogs, open-source software, and public datasets.
Hearing about open altmetrics is one thing— but seeing one’s own altmetrics, with the ability to
Finally, an improved ImpactStory API will form the hub of an open data infrastructure
connecting dozens of diverse data providers (like Mendeley, Twitter, or Dryad) with a
institutional repository usage widgets, literature search tools, enhanced citation indexes, faculty
7
profile collections, funding databases, institutional and regional impact assessments, expert
fact, we’ve had requests for data from projects in each of these categories already. As we
improve its scalability, our open API will support an ecosystem in which impact data flows like
water among these and other diverse applications, with ImpactStory as the “village well”
Achieving this long-term mission will require a stable revenue stream. Here we describe our plan
to secure this. Borrowing from the “Lean Canvas,” a common startup business plan approach
(Maurya, 2012; Ries, 2011), we will examine our product, markets, unique advantages, and
Product
Our product will be based on the existing ImpactStory application:2 users identify their scholarly
products, then ImpactStory gathers and displays classified and normalized impact metrics.
2
http://impactstory.org
8
Figure
1:
A
sample
article
displayed
in
ImpactStory
This functionality will remain available to anyone, for free. To earn revenue, we will also roll
out premium features for easy account creation, exploration, and automatic updating via VIVO,
ORCID, and other databases and search tools (see section 6).
Markets
Our primary market will be academic libraries buying premium ImpactStory subscriptions for
1 It improves researchers’ productivity by saving them time spent updating CVs—and does
so much more cheaply than Research Information Management (RIM) systems like
2 It empowers those institutional faculty who are at the cutting edge of scholarly
3 More subtly, but perhaps most important, it helps establish the library as more than just a
place for books and periodicals, but a forward-thinking, driving force behind scholarly
communication on campus. We hear again and again how keen libraries are to make this
9
transition, despite growing pressure on library budgets (See Appendix D). Indeed, budget
pressure is often driving interest in novel, forward-looking roles for libraries, as librarians
seek to maintain and expand their institutions’ roles in this time of change.
We have seen significant evidence of library interest in the last year: we have received 20
travel reimbursement), and 3 webinars (see Appendix B for details). This interest is particularly
encouraging given our focus on investing in the product over investing in sales to date.
Furthermore, university libraries see value in altmetrics for researchers: “we do plan to have
every single researcher at the University of Pittsburgh able to look at their own [altmetrics]
The large number of research universities combined with our low projected expenses ($302k/yr)
gives us room to reach sustainability relatively quickly in this market, even at a low initial
penetration rate. There are 657 PhD-granting institutions with over 2000 FTEs in the United
States: they have approximately 8,088,302 FTEs combined4. To speed adoption, we will aim at a
(compare to around $15k for other research productivity tools like RefWorks or Mendeley
3
http://www.thedigitalshift.com/2013/02/research/as-university-of-pittsburgh-wraps-up-altmetrics-pilot-
plum-analytics-announces-launch-of-plum-x/
4
http://nces.ed.gov/ipeds/datacenter
10
Innovation diffusion theory predicts several adoption tiers (Moore, 2006; Rogers, 2003). The
table below shows (in bold) market penetration needed to meet our projected expenses at various
Innovators,
Innovators, early adopters,
Innovators, early adopters, early majority,
Aggregate Innovators early adopters early majority late majority
Price per FTE FTEs (2.5%) (16%) (50%) (84%)
After launching we will measure adoption and adjust our prices accordingly, as recommended by
software pricing advisors (Davidson, 2009). We anticipate high market penetration with
Innovators, Early Adopters, and the Early Majority university libraries in the next three years.
Unique advantages
From a strictly market perspective, we offer potential library customers several unique
advantages. Heather and Jason are leaders in the altmetrics and open science communities (see
section 8, and attached CVs), boosting our tool’s visibility and credibility. As the only
researcher needs. Most importantly, our open data infrastructure, backed by an open-source
11
scholarly community can trust and rely on. Open data is not a side benefit for us—it’s our raison
This business plan, like all plans, carries with it risk. Below we describe three key risks, along
● Pursue secondary revenue streams, including: research centers and funders (several
have expressed interest; see Appendix B) and scholarly publishers (many of these have
● Use the unanticipated revenues from strong demand to outsource coding to contractors;
12
Summary
Our plans sound a lot like those of a startup, and not by accident—we value the agility,
responsiveness, and market thinking of startup culture, while simultaneously emphasizing our
unique value as a mission-driven non-profit in this field. Our goal is to drive mission and
sustainability simultaneously through earned income based on a product that promotes broad
This grant will promote altmetrics and web-native scholarship via continued advocacy, an
altmetrics webapp, and an open data API, all managed by a self-sustaining organization.
Continued advocacy
We will continue to provide leadership in the areas of web-native scholarship and altmetrics. In
● Work with PLOS, NISO, and others to draft open altmetrics standards,
● Embed ImpactStory data in over 200 IRs, online journals, and other tools.
Altmetrics webapp
Building on our current platform, we will add new features to streamline the account creation
process, improve value to researchers, and increase engagement. Existing features and some new
13
ones will remain free to all; many of the new features, however, will form the premium service
sold to libraries (we will also offer individual subscriptions, though we expect limited uptake of
● Account creation: We will streamline the profile-creation process: users will just enter
their name and email, then import their authored products in one click via Google Scholar,
ORCID, VIVO, or by uploading an old-fashioned PDF or Word CV. Imported data will
be parsed, saved, and enriched with altmetrics data in less than a minute.
that leads to a profile page including image, links to other online profiles, and a
comprehensive list of products (traditional and web-native) with citation and altmetrics
data built in. This profile will be complete enough to be used in place of an online CV—
analytics panel will let researchers filter, explore, and visualize their entire research
histories to surface their most powerful stories—for instance when filling out the
“broader impacts” section of an NSF grant, a researcher could sort by “public impact” in
ImpactStory to uncover the products best demonstrating her impacts on the broader
public.
● Engagement: When an author publishes a new scholarly product online, ImpactStory will
automatically add it to her online profile. New impacts will also be automatically
detected and added. Users will receive weekly summaries of their new products and
cheaper alternative to a full-fledged RIM system), since it will remove the burden of
manual updating. It will also keep researchers engaged with the application over time,
14
turning them from one-time to repeat users. For these users, altmetrics and web-native
products become parts of their scholarly identities, driving grassroots pressure to value
We will maintain our current API, but improve our ability to scale from hundreds of daily
requests to tens of thousands. This represents more than just a quantitative change, but a
qualitative one. We have had to reject at least a dozen data or API requests from researchers,
tool-builders, journals, and others simply due to scale constraints. Scaling up is a straightforward
programming task, but takes time; by getting that time, we gain the opportunity to throw the
doors open to a developer community already eager for ImpactStory data (see Appendices B and
D). We will also make it easier for external developers to add new data providers to the core
Self-sustaining organization
We will cover expenses and contracting with revenues earned from premium subscriptions to
institutions, supplemented with donations and individual subscriptions (see attached budget for
detailed projections).
We anticipate becoming revenue-neutral in three years. Beginning in July 2016, we estimate our
steady-state revenue will be $305k/year, 82% from product income and 18% from donations and
15
We estimate our ongoing operating costs at $302k/year. Most of this (56%) will pay for salaries
for the two ImpactStory employees (during the two years of the grant, we will also spend $125k
one-time on external contractors to accelerate progress toward scalability). Data sources will
cost about $72k/year at our anticipated scale, 24% of our expenses. About 12% of expenses will
go to continued outreach and market building, particularly travel and workshop hosting, with 8%
We seek $500k from the Alfred P. Sloan Foundation to help with this transition: $250k per year
for two years. We will use this investment runway to ramp up our earned income and donation
program, as shown in the chart below. We expect the combination of our earned income in the
second year and the Sloan funding to exceed our expenses: we will bank the surplus earned
income to supplement our revenue in year three when the Sloan grant has ended. By year four
16
7. What other sources of support does the proposer have
We have no other sources of support currently in hand or submitted, but we have received
several invitations to apply for supplemental grant funding with various community partners (e.g.
PKP, CDL, NSF software community, HubZero, Dryad), and are following up on these.
We are on track to meet or exceed most of our milestones by our current grant’s end. Specifically,
We have filed incorporation paperwork for a nonprofit corporation in North Carolina, established
● Goal: 60 GitHub watchers, 20 forks. So far: we have 25 stars (the new name for
watchers) and 5 forks on our new codebase,5 plus 37 stars and 5 forks on the old
codebase.6
● Goal: A dozen new information sources, particularly data repositories. So far: 6 new
active information sources (now 21 total), with several more near completion.
5
https://github.com/total-impact
6
https://github.com/mhahnel/Total-Impact
17
We are on track to meet most of our use goals:
● Goal: 50k website visits, 30k uniques. So far: we’ve had 52,365 visits, 31,070 uniques.
than 10 journals), but many more in the pipeline. We have signed a MOU with Highwire
● Goal: 5 research studies using ImpactStory data. So far: we know of at least 3, plus one
Perhaps most importantly, our code and API have been used by the community. A demo7 for
a new commercial altmetrics product (Plum Analytics) was based on our code, a hackathon
project8 was based on our API, an R interface has been written,9 a Javascript library has been
contributed,10 an R report generator11 has been written, and a faculty profile system12 enriched
with altmetrics data. We have collaborated with several other Sloan-funded projects, including
the Beyond The PDF 2 workshop and PLOS ALM, and have begun collaboration discussions
with representatives of other projects including ICPSR, Sage Bionetworks, and Hypothes.is. We
have are now planning with VIVO to embed ImpactStory metrics next to Web of Science
7
http://www.youtube.com/watch?v=pRnU8aJQQ0U
8
http://rmimr.wordpress.com/2012/12/08/rerank-it-rerank-pubmed-search-results-based-on-impact/
9
http://ropensci.github.io/rImpactStory/
10
https://github.com/highwire/opensource-js-ImpactStory
11
https://github.com/ropensci/ImpactReport
12
http://research.mblwhoilibrary.org/works/35002
18
Our efforts at altmetrics advocacy and education have been successful; we believe these are
vital for fostering the institutional change needed for widespread use of altmetrics and web-
● Given talks at over three dozen events in North America and Europe,
● Written 45 blog posts, which have together been viewed 3,330 times since September,
● Written several articles including two well-received15 Nature pieces and an MIT Press
There is also one area where we have not met projections: so far we know of only five scholars
embedding ImpactStory data on CVs. We also have evidence of three ImpactStory reports
included in annual review or tenure/promotion packages, plus one grant submission. This has
been supplemented by many reports of ImpactStory data being used less formally, thought no
less meaningfully: to provoke conversations about broader impacts at high administrative levels
(see Appendix D). Moreover, these numbers are certainly undercounts, since we know many
users do not report their ImpactStory use. However, these still fall short of hoped-for uptake. We
19
First, for most of the grant our tool lacked several important technical features, including stable
profile URLs, easy product input, and editable profiles (these have since been added or are
underway). Second, it took until late in the grant to gain significant traction for the idea that
altmetrics could be beneficial on a CV. Our most recent evidence (see Appendix D) suggests that
early-adopting researchers are increasingly interested in employing altmetrics, but this change
took longer than we expected, and is still very much in progress. We believe that continued
altmetrics advocacy is the best approach to address this, as researchers indicate is having a
powerful—if gradual—effect.
Overall, we are confident that continued advocacy, an improved webapp, a scalable open API,
and a practical, high-growth business strategy will nurture ImpactStory into a powerful and
sustainable force supporting open, interoperable altmetrics. These in turn can be the foundation
of a new system of web-native scholarship—one that rewards scholars for creating diverse,
powerful, engaged products. Building on an open altmetrics infrastructure, we can build a new,
20
Appendix A: references
Cassella, M., & Calvi, L. (2010). New journal models and publishing perspectives in the
doi:10.1177/0340035209359559
Davidson, N. (2009). Don’t Just Roll The Dice - A usefully short guide to software pricing. Red
gate books.
Eysenbach, G. (2012). Can tweets predict citations? Metrics of social impact based on Twitter
and correlation with traditional metrics of scientific impact. Journal of Medical Internet
Haustein, S., & Siebenlist, T. (2011). Applying social bookmarking data to evaluate journal
Li, X., Thelwall, M., & Giustini, D. (2011). Validating online reference managers for scholarly
Maurya, A. (2012). Running Lean: Iterate from Plan A to a Plan That Works (Second Edition.).
O’Reilly Media.
Moore, G. A. (2006). Crossing the Chasm: Marketing and Selling Disruptive Products to
Neylon, C., & Wu, S. (2009). Article-Level Metrics and the Evolution of Scientific Impact. PLoS
21
Priem, J. (2013). Scholarship: Beyond the paper. Nature, 495(7442), 437–440.
doi:10.1038/495437a
Priem, J., & Hemminger, B. H. (2012). Decoupling the scholarly journal. Frontiers in
http://www.frontiersin.org/computational_neuroscience/abstract/14455
Priem, J., Taraborelli, D., Groth, P., & Neylon, C. (2010). alt-metrics: a manifesto. Retrieved
Ries, E. (2011). The Lean Startup: How Today’s Entrepreneurs Use Continuous Innovation to
Rogers, E. M. (2003). Diffusion of Innovations, 5th Edition (5th ed.). Free Press.
Smith, J. (1999). The deconstructed journal-a new model for academic publishing. Learned
Taraborelli, D. (2008). Soft peer review: Social software and distributed scientific evaluation. In
doi:10.1186/1758-2946-3-36
22
Appendix B: Evidence of interest from libraries, centers, and
publishers
If only our Faculty Annual Reporting system worked more like @impactstory.
—Stacy Konkiel, Indiana University librarian [reproduced here as it is a public tweet]
Publishers embedding ImpactStory data (important because it raises our visibility among
researcher):
● Already using our product: pensoft, ubiquity press, peerev, CITAR, eLife, PeerJ
● Signed MOU: HighWire
● Have been in discussions with: Wiley, BioMed Central, Open Medicine
● Invited to trade conferences, including SSP in 2012 and 2013, the Highwire Spring
Conference 2012, Charleston 2012, STM conference, Council of Science Editors meeting
(teaching a class), and Emerging Trends in Scholarly Publishing workshop.
23
Appendix C: Description of ImpactStory
Our data is open, released under as open a license as our data providers allow. The data can be
downloaded as JSON (to support building other applications), CSV (to support analysis and
visualization), or JavaScript widget code (for easy embedding). The URL for a collection is a
permaurl which is openly viewable by anyone, so can be tweeted or included on grant
applications.
ImpactStory’s data is accessible to all types of users: any researcher can explore collections of
their work via our web interface to surface data-driven stories for report to funders and
evaluators. The can also embed this information on their CV. Publishers and repositories can
embed ImpactStory data right next to their products.
Finally, ImpactStory goes beyond these three key tenets of evaluation data to make altmetrics
data meaningful through context: metrics are presented in a framework accounting for both
audience and engagement type. Users unfamiliar with Mendeley, for instance, can still
understand that their product has been highly saved by scholars. Significant Wikipedia activity
might earn a “highly cited by the public” badge. The metrics are presented in comparison to
other products (with a 95% confidence interval), rather than in isolation to facilitate
interpretation. See the figures below for examples:
24
25
Appendix D: Letters of support
[All the quotations and names have been redacted from this section, since we didn’t
have permission from all the authors to reproduce them.]
From librarians
[redacted]
[redacted]
From toolmakers
[redacted]
26
Appendix G: Conflict of Interest statement
Heather Piwowar has been funded by DataONE and the Dryad Data Repository. Heather
Piwowar serves on the advisory board for F1000 Research.
Jason Priem and Heather Piwowar do not have other potential conflicts of interest or sources of
bias.
27