Executive summary

Thinklab is a platform for open review of research grant proposals. We're creating Thinklab because we believe the interests of science and society are best served if the entire scientific process is open — not just the software, data, and papers produced at the end. And while we're convinced this kind of openness is in the best interests of science as a whole, we understand many scientists will not feel it's in their personal interests. That's why our goal is not just to enable open research, it's to actually create incentives for it. We aim to create value for both scientists and science funders. Thinklab is intended for broad use across all of science, including biomedical research. An early version can be seen at thinklab.com.

How Thinklab works

Thinklab targets two groups: scientists and science funders. We'll start this proposal by looking at Thinklab from the perspective of scientists. Later on we'll explore how Thinklab can help science funders open up research and accelerate their impact.

Posting an open proposal

The first challenge Thinklab faces is persuading scientists to publicly share their research proposals. This is tricky because there's a fear that ideas will be stolen. We believe this fear is valid but overblown. We also believe there are a number of benefits to grant writers that are not fully appreciated:

A validated research plan — Before embarking on a research project it's critically important to make sure you're working on the right thing and have the right approach. Early feedback can validate the research plan and give researchers confidence that when results come in, they'll be trusted and valued by the community.

Improved funding odds — It's easy to forget that not everyone has the same set of background knowledge that we have. Getting an outside perspective can make sure the value of a proposed project is being communicated effectively — presumably improving the proposals funding odds.

More visibility and citations — Research proposals are valuable scientific outputs in and of themselves. Thinklab gives them a DOI and makes them a citable part of the scientific record.

More collaboration opportunities — Instead of encouraging others to compete with you, an open proposal could do just the opposite. If people see you're already working on something it's likely they'll consider collaborating with you, or simply pursue another avenue of research.

The open review process

Once a proposal has been published it's up to the research team to invite peers to review it. Alternatively, scientists (or any approved user) can review a proposal they find interesting. Thinklab will strongly encourage a two-part review process where ideas are first submitted as "one-pagers", and only later developed into full proposals. Here's what the review process looks like:

Review Part 1: Annotate — In this step reviewers are asked to read the proposal and make annotations wherever they have feedback (see Figure 1). To ensure reviewers form opinions independently we hide annotations and comments from others at this stage.

Review Part 2: Summarize — Next, reviewers are asked to summarize their most important thoughts and advice. They may also be asked to rate the proposal on various metrics.

Review Part 3: Discuss — In this step we reveal comments from other users. Reviewers are asked to use the notes they made in part one to join a real-time discussion with the authors and other reviewers. Discussion can take place in inline comments or in longer form discussions that exist on their own page.

Review Part 4: Finalize — After discussing the proposal, reviewers have the option to update their review scores and/or provide updates on how their overall assessment of the proposal has changed.

This review process has been used to gather feedback on this proposal. Four reviewers made a total of 150 comments and 139 annotations. Casey Greene, a first time reviewer, said: "The UI is very slick!"

Figure 1. Proposal annotation options

This menu appears when proposal reviewers select text to make an annotation.

Thinklab Impact Points

To reward reviewers for sharing valuable feedback and ideas, Thinklab has created an impact points system. This system allows us to highlight the "most impactful reviewers" for each proposal, while also creating a global impact leaderboard.

Reviewers earn impact points based on peer assessment of the value of their contributions. Contributions are rated by the research team, by users selected by a sponsoring funder, or by the community at large. To avoid any temptation for "reciprocal voting" ratings are submitted privately.

To create additional incentive, Thinklab aims to partner with science funders and offer monetary rewards that are tied to impact points. (More on this later.)

Thinklab Inbox

Users are notified of new content and discussion that is relevant to them via the Thinklab Inbox. Right now it simply shows any discussion that directly matches anything the user is following. Users can follow proposals, topics, or even funding opportunities. Over time we intend to make the system smarter such that we more efficiently direct scientific attention to the exact problems and discussions that need it.

Real-time open science

When a proposal gets funded, the research team has the option to continue their work as an open research project. Researchers can engage the community by sharing ideas, project plans, updates, and questions in real-time as the project progresses. Reviewers that made valuable suggestions during the proposal stage can continue to share ideas and give feedback throughout the project. This all occurs using the exact same discussion system that is used for proposal review.

Daniel Himmelstein is currently using Thinklab to lead an open research project: Repurposing drugs on a hetnet [1]. His project has accumulated over 300 comments from 28 contributors. In his review of this proposal Daniel says:

The best thing about Thinklab is that it not that it solves a major problem in science but that it solves most major problems in science. Having used Thinklab for over a year now, I have experienced the plethora of benefits.

Alternatives to Thinklab

If a researcher wants to get feedback on a grant proposal, a common solution is to put it in Google Docs, and send it to colleagues. Here are the important ways Thinklab is different:

• Thinklab is designed to have peer review and reviewer comments become a persistent, citable part of the scientific record. Each Thinklab discussion has a DOI and is citable.
• The Thinklab Inbox serves to direct researcher attention throughout the network by automatically showing users discussion that matches their interests and expertise.
• Conversation is not constrained by the need to fit in the sidebar of a Google Doc. Users can post more substantial discussion to it's own page where it's easier to read, and can attract more attention.
• Thinklab recognizes reviewer contribution through an impact points system that is based on peer assessment.
• Thinklab intends to build features that will integrate with workflows at funding organizations and enable them to manage an open peer review process.

Openness & sustainability

Thinklab sees itself as a service that helps open up science. That's our mission and our most important work. For example, all content and discussion posted to Thinklab is licensed CC-BY.

While we'd like to be in a position to open source the code for Thinklab itself, at this point we feel that might jeopardize our ability to create a sustainable business. First, it creates the possibility that a group of academics would launch a competing service and thereby hamper our ability to gain traction. Second, it would likely make it very hard for us to get funding from traditional investors. While we prefer to be funded by impact investors, the reality is we need to keep our options open at this point.

Thinklab as a service for science funders

Thinklab's primary goal is to accelerate a transition to a much more open, collaborative, Internet-native model of research. We believe creating a service for grant writers — as outlined above — will produce some success towards that goal. However, given the inertia of our scientific system, and given the incentives that drive scientists towards secrecy, it is likely that progress will be slow.

We see this as a problem of incentives. To drive rapid adoption of open grant proposal review we need the incentives of scientists to be aligned with what is good for science as a whole. We believe science funders hold the power to align these incentives. In fact, they're already starting to do this in other areas. Many funders have started to require that the results of the research they fund be made publicly available.

Thinklab wants to help funders take the next step — we want to help funders create grant programs that require openly posted proposals. In addition, we want to help them manage an open peer review process where anyone from the scientific community can share ideas and feedback to help improve research plans.

Monetary rewards for crowdsourced review

The big payoff from open research comes not just from being open, it comes from the real-time collaboration that openness enables. It comes from bringing a cognitively diverse set of people together to tackle challenging problems. [2]

To fully realize the potential of Thinklab, there needs to be strong incentives — not just for grant writers — but for reviewers too! If we want scientists to take time away from their research to share feedback and ideas on that of their peers, we need strong incentives.

Thinklab proposes that science funders set aside a portion of grant money to financially reward an open crowdsourced review process. These rewards would go directly to reviewers and be paid out in proportion to how much value they add. Distribution is tied to the impact points system we described previously. The only way to game the system should be to actually write comments that peers find valuable! At some point there will likely be attempts to game the system via "voting rings", but we're confident this issue can be managed.

There has been some concern about the overjustification effect. This is where external incentives (money) will decrease a person's intrinsic motivation to do something. However, it should be noted that this doesn't really apply if very few people are doing the desired activity in the first place. And this is by and large the case when it comes to getting scientists to comment on their peers work on the Internet.

Benefits for funders

The following describes the benefits a Thinklab partnership can offer science funders. The theme here is accelerating scientific progress by making research more open and more collaborative.

Reviewers are less likely to reject good proposals for faulty reasons. An open real-time discussion gives authors a chance to address reviewer criticism and possibly alleviate concerns.

Fundamentally flawed research is more likely to be filtered out. If proposals are first submitted as one-page ideas, it's reasonable that 10-20 people could read each one. Anyone that notices a potential flaw can start a discussion and bring in experts to address that particular issue.

Ideas that have potential can be developed into great ideas. The first ideas people have are not always the best. Thinklab gives reviewers the opportunity to work with authors to develop an average idea into a great idea.

Best practices can spread across science faster. Best practices are changing fast. Thinklab gives people an opportunity to share those best practices, even across disciplines.

The exchange of ideas is accelerated. Publishing proposals (and the review discussion related to them) creates a valuable resource of cutting edge ideas that scientists can learn from and build upon. Without Thinklab these ideas might not be shared until years later. Fundamentally, science progresses in proportion to the speed at which we share scientific information.

More collaboration and less redundancy. If scientists know what each other are working on, it creates more possibilities for working together. At the same time it will likely reduce the redundancy of having multiple research groups working on the same thing in isolation.

Product roadmap

• Continued development of our open review system. We will be working in close consultation with both grant writers and reviewers to create the most useful and user friendly service possible.
• Development of features to enable a pilot program with a funder.
• Creation of APIs so Thinklab content can be more easily accessed.
• Integration with more external services such as CLOCKSS, Altmetrics, and PubMed.

Team and resources

Thinklab is founded by Jesse Spaulding. Jesse has created and sold several startups including CourseTalk, a student review site for MOOCs (Massive Open Online Courses). Previously, Jesse independently developed a high frequency trading algorithm that made 500k in profits. As someone who has spent his career in the startup world, Jesse is able to bring a fresh set of eyes to the challenges facing our scientific system.

Jesse is working with Gleb Pitsevich who has a broad background in mathematics, programming, and web development. Gleb has a master's degree in computer science and is the founder of Razor Theory, a web and mobile development company based out of Belarus.

Thinklab is looking for a talented individual with a strong academic science or philanthropic background to join the team as a co-founder and lead business and community development.

Conclusion

We believe open review of grant proposals will improve research plans, reduce wasted resources, spread best practices, accelerate the exchange of ideas, and lead to more collaboration with less redundancy. In short: it will significantly accelerate scientific progress. Perhaps the most compelling argument for Thinklab is this: what we're doing has the potential to catalyze changes that will affect all of science. Anytime you can make changes that positively affect a system there's potential for massive impact. Science needs people thinking outside the box!