Traditionally, researchers start their investigations with studies in simplified contexts; only later do they work on scaling up their successful innovation to realistic teaching and learning settings. The traditional process — sequential and phased —is slow and often ends with failures to scale up and replicate results across settings. Researchers also tend to choose problems based on the next step in their scientific program, and consequently their “solutions” can drift away from the pressing problems of practice that occur in realistic school settings. In contrast to both these tendencies, researchers now have increasing opportunities to start their work where teachers and students are spending more and more educational time: on digital learning platforms. They can better focus their research around problems as they actually occur in the lives of students and teachers.

SEERNet is a program of support for research on DLPs that was initiated and funded by the US Department of Education Institute of Education Sciences in 2021. This program will be led by Digital Promise and Empirical Education over the next five years. Five platforms — E-Trials, UpGrade, Terracotta, and Learning at Scale — are currently participating in SEERNet, and later others will join. 

Here we provide an illustrative example of how SEERNet might address the challenges of improving mathematics teaching and learning. An illustrative scenario can clarify our argument, but please keep in mind that this is just one possible scenario where SEERNet could make a difference. SEERNet is not limited to mathematics applications nor to this particular set of challenges. Nor will every research need to work with all five platforms.

In 2018, a report called The Opportunity Myth described four barriers that prevent students from progressing equitably in mathematics. Students do not equally have (1) grade-appropriate assignments, (2) strong instruction, (3) deep engagement, and (4) teachers with high expectations. The Opportunity Myth was released pre-pandemic and the problems still exist today. However, a big change has occurred. When the 2018 study was conducted, the research team painstakingly collected paper materials because that is what teachers and students were using. But today, many more students are learning mathematics on digital learning platforms. How could research teams leverage SEERNet and its DLPs to address the four barriers in The Opportunity Myth, thereby tackling a big mathematics education challenge with SEERNet over five years?

1. Improving math assignments. One SEERNet platform, Terracotta, is a plug-in to Canvas. Terracotta enables systematically varying the assignments made by teachers to students via Canvas — and in many school districts, this is centralized infrastructure for assigning and monitoring student work. Today, an army of researchers works on creating high quality instructional materials, but very few work on why existing high quality instructional materials are underutilized in schools. Over five years, we can imagine research teams shifting their attention to patterns of inequity in opportunities to do challenging mathematics and how they could provide supports to teachers and students within existing large scale platforms to address this. For example, could recommenders help teachers choose more challenging assignments that are appropriate for their students or help them modify their own assignments to increase cognitive demand?
2. Strengthening instruction. Another SEERNet platform, ASSISTments eTrials, tackles a focused but important part of routine math instruction: how students receive feedback and explanations of mathematics when the students show signs of struggle in doing a particular assignment. Today, many research teams have an interest in adaptive instruction, but few research teams can build a platform, get it into wide use, and then study how to improve feedback and explanations in their platform. Over five years, research teams who have a strong drive to strengthen instruction for particular populations, for example, students with an Individualized Educational Plan (IEP), could leverage this platform to build and evaluate scalable solutions to better support the focal populations they most care about. In turn, the ASSISTments team would willingly incorporate and scale up any improvements that reliably work better than their existing “best available” methods of providing feedback and explanation.
3. Deepening Engagement. A third SEERNet platform, Mathia / Upgrade, allows researchers to systematically modify curricular units in an established, well-used product that already provides adaptive support to students. Mathia has a strong grounding in cognitive research, but now educators and researchers understand the need for a whole child approach in supporting students who previously faced barriers to now deeply engage in a challenging topic like mathematics. Via SEERNet, research teams could develop analytics to reveal when students are experiencing weaker engagement with Mathia and could explore how to provide whole-child resources to deepen and sustain engagement in challenging mathematics, especially for those students who have been weakly supported in the past.
4. Teachers with High Expectations. A fourth platform, OpenStax, is an open educational resource that enables instructors to modify high school and college textbooks and to insert interactive supports that students can activate. Unfortunately, by the time that many students get to high school and college, their expectations of performing well in mathematics have been shattered by past negative experiences with courses like Algebra. Via OpenStax, we imagine how instructors and researchers at a community college that serves a Latino/Latina population could decide to produce video and textual “explainers” that feature Latino and Latina students who have succeeded in their courses, and thus can both serve as positive role models and talk about how they learned powerful mathematical ideas. (We drew upon a TNTP report in generating this example). Via OpenStax Kinetic, these explainers could be embedded throughout a college-level mathematics textbook, for students to choose and use at their discretion. Researchers could study which Explainers  keep students engaged.
5. A Testbed for Pulling it All Together. The fifth DLP in SEERNet involves the online course programs of a major equity-oriented university, Arizona State University. Supporting students growth in mathematical proficiency is a major challenge for all equity-oriented universities. Although ASU’s SEERNet capabilities will be available last, we see ASU as the ideal testbed where the results of all four research programs above could be integrated and used to disrupt the barriers described in The Opportunity Myth and instead promote strong math learning among the diverse populations that ASU serves.

Hence, although we described how each SEERNet platform might work on aspects of a problem, we also see the possibility for a community-driven program of research that tackles a bigger problem than any one team can tackle on its own. Further, underlying our examples, there are common reasons why shifting research from its conventional forms to be based on DLPs can enable dramatic change in how impactful research is. SEERNet could: 

• shift the attention of researchers towards problems of practice, as they occur in realistic settings. 
• lead to improvements being designed and studying within widely used systems, so that successful improvement can be more rapidly scaled up. 
• lead to partial solutions grounded in one perspective, for example, a cognitive perspective, becoming open to research teams who offer complementary perspectives, such as a socio-emotional or whole-child perspective. 
• translate broad calls to action for “high expectations” from inactionable towards a toolkit for disrupting the cycle of low expectations. 
• provide a testbed for integrating multiple successful factors from prior studies and applying them to make a big difference for a large population.

Changes like this become possible when research shifts to digital learning platforms that are already in wide use and are made more available to research teams outside the original platform developers. Over five years, re-grounding research in newly open DLPs could bring a stronger focus on solving problems of practice within realistic contexts, and this could dramatically increase the ability of learning scientists to make big, important impacts in education. 

In a famous quote, J.C.R. Licklidder wrote “People tend to overestimate what can be done in one year and to underestimate what can be done in five or ten years.” We expect SEERNet will start slowly because there are many operational and infrastructural issues to work out. But our sights are on bigger changes to the impacts of educational research.