Canvas

Authorship on Canvas LMS

Canvas is a popular learning management service (LMS), or educational technology platform that institutions purchase or adopt that enables students, administrators, educational staff, and instructors to exchange information. For instance, an instructor might post course readings for students to access and assign grades within an LMS, while students upload assignment drafts or participate in discussion forums where they discuss course readings. In examining Canvas's terms of use and privacy policies, we want to acknowledge that these policies are some of the best we've seen from a platform provider. Still, we want to offer a critical investigation of authorship issues associated with using an LMS for two reasons: (1) these platforms are ubiquitous in educational settings today and (2) LMSs provide gateways where parties beyond a student, an instructor, an institution, and the LMS platform provider can access inputs.

LMSs, as web-based platforms, make use of tracking tools that store user data and metadata, much of which is invisible to student users and even to instructors. Let's specifically consider Canvas's "Instructure Privacy Policy." Analysis of this policy demonstrates that the platform draws a distinction between "Data You Provide to Us" and "Data Collected Via Technology" before concretely noting that it deploys cookies, web beacons, flash cookies, and analytics to gather information about users. While students understand that when they input information requested by Canvas such as first and last name, gender, email and mailing addresses, the data may be stored and used by Canvas, they may not realize that other, less visible data, including "information regarding the date and time of your visit and the solutions and information for which you searched and which you viewed," is being collected and stored. The policy also notes that Canvas retains information such as "files and messages that you store using your account." Further, the policy language fails to offer detailed insight about how the platform might make use of the data and information it collects. The information collected is not displayed to nor used by students or instructors; rather it serves to benefit Canvas by allowing for development of their products, for targeted advertising that brings in revenue, and for agreements with third parties who have an interest in such amassed information.

Figure 13. The Learning Tools Interoperability (LTI) standard enables administrators and educators to plug platforms into an LMS such as Canvas. This screen capture depicts a set of choices that are available to Tim within the Canvas LMS at CSU. Note that he might incorporate a range of platforms made by Unizin, Macmillan, McGraw-Hill, and Amazon [Echo 360].

The point that we want to stress here is that LMSs are not simply neutral channels through which content, data, and metadata flow within institutions and between students, instructors, and administrators. LMSs also function as gateways that connect students' intellectual property, information, and data with platform developers and with third-party tools and applications. Developers make use of the Learning Tools Interoperability (LTI) standard to embed third-party tools directly within LMSs, granting access to the networks of activity and associated IP that are generated within schools and colleges. LTI can be useful for universities because it helps educators, students, and administrators select and integrate tools and services that they might find valuable. For example, at Tim Amidon's home institution, Colorado State University, instructors might choose from a range of plugin tools, including Unizin Engage, McGraw-Hill Connect, and Pearson MyLab (see Figure 13 for a screen capture of tools available for integration within Canvas via LTI).

Many educators at CSU find these types of tools exciting because they offer new and valuable ways of understanding how we learn and teach. Some of Tim's colleagues, for instance, have been using a learning analytics tool called Unizin Engage to better understand how students are engaging with course materials. As Unizin puts it, "Engage provides insights into how students are using content. Instructors can quickly see which students have questions and how often they are interacting with the text." Another example of a learning analytics tool that Kairos audiences might be familiar with is Eli Review, which Jeff Grabill, Bill Hart-Davidson, and Mike McLeod developed "to support evidence-based teaching practices and facilitate rich peer learning environments." Learning analytics tools such as these can be used to better understand how learning and teaching unfolds by offering educators, students, and administrators new windows through which to view educational practices. However, educators, students, and administrators need to not only understand who receives and has access to the content, data, and metadata they generate within these platforms (especially learning analytics platforms that are seamlessly integrated into robust LMSs), but also to understand how university agents, platforms, and third parties beyond universities and platforms might use the IP students and educators have contributed to these platforms.

Table 4. Tensions in Student IP Heuristic Applied to Canvas
Category Category Attributes
Users Students; graduate teaching assistants; instructors; faculty, program, department, and institutional administrators; academic support staff; athletics director; tutors; agents of the institution; LTIs (plugin third party applications); consortia members; platform employees and staff.
Permissions Instructors or programmatic, department, and/or institutional administrators adopt the LMS and compel students to use the platform as condition of participation in course.

Click-wrap agreement where student agrees to platform terms of use.

Policies such as end-user license agreements (EULAs), as well as platform specific copyright and privacy policies govern use.

Instructors and/or program, department, and/or university administrators may adopt or integrate third-party tools which plug into the LMS, requiring students to view additional click-wrap agreements that enter students into terms of service, EULAs, and platforms' policies that govern copyright and privacy on platforms.

Inputs Surface contributions: Student assignments, course assignments, readings/PDFs/course texts, links to content hosted on external platforms (e.g., YouTube videos), discussion forums, direct messages/emails, announcements/replies, schedule.

Nontransparent and hidden contributions: Data and metadata about interactivity, including when students/instructors have uploaded/downloaded content types, assignments, files, when students submit, if submissions are late, how much time students have spent logged in, how many of the files have been downloaded by students, how much time students have spent watching embedded videos within the platform, file types and file metadata, IP address, student name, student grades, interactivity/behavior within website, geolocation, browser, and hardware used.
Operations Platform facilitates student-to-student and student-to-instructor interactions.

Data about interactions is captured and logged within a database.

Algorithms are applied to data to reveal insights about student learning behaviors, individual and group student performance and engagement levels, and/or perform student, instructional, and programmatic assessment.

Learning tools interoperability standards facilitate ability for instructors and administrators to integrate external tools for use in courses and programs within universities and schools, allowing content, data, and metadata to flow from universities and schools to tertiary platforms.
Outputs Insights on learning behaviors at individual and aggregate levels.

Insights about how instructors and students use tools/platforms, including which types of content are interacted with and how.

Databases of content, data, and metadata uploaded to/passing through the site.

Proprietary algorithms trained using database.
Gateways Once LMS is adopted, students and instructors have ability to upload and interact with content within LMS, as well as with additional platforms integrated into LMS through LTI standards. However, students/instructors must agree to secondary/tertiary EULA agreements (e.g., a PDS integrated within Canvas gives appearance that a user has never left LMS, but is now working within a secondary/tertiary platform).

Teachers and administrators have ability to observe/view analytics data about student performance and behaviors within LMS.

De-identified data can be scraped and mined from classes or LTI plugins for institutional research, consortia level research/planning, usability testing of learning applications/tools, as well as to garner insight into trends about how students behave/learn (e.g., engagement research conducted by programs, departments, universities, consortia, platforms, and/or platform-institution or platform-individual collaborations.
Figure 14. In his keynote address to Computers and Writing Conference attendees in 2016, Jeff Grabill warned computers and writing specialists about the perils of not paying attention to edbots. Here we have linked to a full-length transcript of Grabill's keynote.

There are ethical and important reasons for educators to employ learning technologies that make use of student data. Indeed, Grabill observed, in his 2016 keynote address at the Computers and Writing Conference, "The work that others are doing with technologies for learning to write has the potential to impact millions of learners." Yet he also took a moment to reflect on the potential implications of failing to attend to the exchange economies that impact how IP circulates within educational technology platforms. He noted how various platforms can be wielded toward very disparate ends in our schools: "One of the most insidious moves in education technology in K-12...is schools' penchant for free on the surface, which costs them dearly downstream particularly taking its toll on the lives of teachers and the lives of students." While the seamless integration of educational technologies into schools and interoperability of third-party applications within LMSs present opportunities for the improvement of teaching and learning, more frequently those partnerships and agreements result in an appropriation of student IP for the benefit of a platform developer or third party rather than for instructors or students.

In other words, learning management systems and the educational technology and learning analytics tools that can be plugged into them have great potential precisely because "of how easily they can harvest our intellectual property, data, and the minute details of our lives" (Morris & Stommel, 2017). LMSs, as tools designed to facilitate and amplify the flow of IP created by students both within universities and to tertiary markets, users, and audiences, are worth critically interrogating for the ways they participate in and reify power dynamics within IP exchange economies. LMSs directly and indirectly enact asymmetrical power relationships in the exchange of information. Resigning control of student intellectual property by enabling wide and uncontrolled access creates an asymmetrical power relationship between powerless intellectual property creators (students) and powerful data brokers (technology developers and educational technology companies).

Figure 15. Noble pointed toward search results from Google's autosuggest function to illustrate the "structural ways that racism and sexism" are built into computational tools as forms of "algorithmic oppression."

This power dynamic might be seen as another instance of inequity resulting in part from disproportionate levels of access to and control of the benefits of education according to class, racial, ethnic, geographic, gender, and cultural makeup. Resigning control of the IP inputs that students make means giving entities outside of our schools access to valuable student data. These data can be used to support both local and external surveillance regimes, and further disenfranchise those who have been historically most subjected to social, cultural, and economic marginalization. According to Safiya Umoja Noble (2018) "discrimination is...embedded into computer code and, increasingly, in artificial intelligence technologies" (p. 1). In her monograph, Algorithms of Oppression, Noble provided a detailed analysis of how Google's algorithms reflect the types of misogynist and racial bias that exist with society, as algorithms learn to return results based on the data users input into its platform and assumptions that algorithm designers make about what those inputs might suggest.

Educational technologies are not different from the logics and use contexts which influence platform design and use in our everyday lives. Educators, in particular, must be aware of the ways that bias and discriminatory practice may enter into education technologies, as Chris Gilliard and Hugh Culik (2016) argued, because they can be used to engage in digital redlining which seeks to "reinforce existing class structures." As Gilliard and Culik posited: "in one era, redlining created differences in physical access to schools, libraries, and home ownership. Now, the task is to recognize how digital redlining is integrated into edtech to produce the same kinds of discriminatory results."

Figure 16."Digital redlining," as Chris Gilliard and Hugh Culik explained, "is not a renaming of the digital divide. It is a different thing, a set of education policies, investment decisions, and IT practices that actively create and maintain class boundaries through strictures that discriminate against specific groups."

Indeed, in their analysis of acceptable use policies promulgated at over 30 universities, Gillard and Culik discovered that Carnegie classifications directly correlated with how universities defined and conceptualized acceptable use. "These deeply different approaches to digital technologies," Gilliard and Culik observed, "discourage and limit working class students from the open-ended inquiry supported at more elite institutions." We share these concerns, and computers and writing specialists, digital rhetoricans, and digital humanists have much to contribute to critical conversations about these tools. Returning to Grabill's keynote, it's important that, as a matter of stance, we engage with these tools and be critical of them.

Responding to the ways that inequities materialize in digital environments, platforms, and spaces has been a central tenet of our disciplinary history. What we find most concerning about LMSs, as IP scholars, is how complacent students, instructors, administrators, and platform users have become about these asymmetrical exchange and control dynamics. It often seems as if educational stakeholders do not realize—or care to realize?—that the student IP created and composed with these platforms generates vast forms of wealth, knowledge, and value.

We worry that students, instructors, and administrators may not recognize that they are no longer working within an LMS like Canvas when they use a third-party application embedded within the system, an application that appropriates IP generated by both students and faculty. We worry that so many of our students question neither our requests for them to use various tools nor the terms of use governing control over the IP they will contribute to or compose within those platforms. We worry that many users may not realize that they are not only granting LMS platforms like Canvas access to their IP, but also—when they interact with other third-party tools plugged into an LMS—granting access to other downstream parties. We worry, further, that school administrators and educators are not acting as responsible stewards of students' educational records when they freely release volumes of mineable metadata, data, and content downstream without deeply considering how those data streams might be recomposed in the future. Indeed, the Electronic Frontier Foundation (EFF) argued that "[t]he pool of data potentially available to ed tech providers is more revealing than traditional academic records, and can paint a picture of students' activities and habits that was not available before" (2017, p. 24). However, we worry most about the cumulative effect of the pedagogical practices surrounding LMSs: Have students and instructors alike been conditioned to work from the default position that the ideas they think, the texts they compose, and the data that they generate are something that they simply do not or should not have control over?

IP Cast 8: Circulation and LMS


IP Cast 8. Circulation and Learning Management Systems. Jessica, Les, and Tim discuss LMSs, emphasizing how access to and circulatory control over user contributions in these platforms create tensions in the management of student authored IP. In particular, we explore how platforms like LMSs construct a variety of gateways (e.g., password logins; LTI application integration) that front-end users and third parties might access through platform inputs and/or outputs. A transcript of the conversation has also been provided.

"School Officials"

What We Can Learn about Student Authorship and Circulatory Control from FERPA

In their 2017 report, "Spying on Students: Schools Issued Devices and Student Privacy," members of the EFF pointed out that FERPA statutorily "protects students' 'educational records' including personally identifiable information….[Some forms of] information about students' online activity and… behavioral 'metadata' unless it has been 'stripped of all direct and indirect identifiers'" (Alim, et al.). Still, it is common practice for universities and schools to regularly share information that includes "direct and indirect identifiers" with tertiary downstream audiences such as platform designers. How, then, do universities and schools reconcile this tension?

According to the EFF (Alim et al., 2017), section 99.31 of FERPA provides exemptions that empower "educational agencies or institutions" to share protected information with other "school officials" when there is "legitimate educational interest" in doing so. The slippery component of this section of the law outlines circumstances in which universities and schools are able to share student data and records with third parties by designating these "contractors, consultants, [and] volunteers" as "school officials." More specifically, the law notes that "educational agencies" and tertiary "school officials" must designate "specific contract terms that limit what data the contractor may collect from students and how it should be used." The problem, however, is that the language found in contracts, terms of use, and privacy policies governing use and sharing is often legally ambiguous and Canvas's Privacy Policy, for instance, states that "Instructure uses your personal information in the following ways: to create and maintain your account; to identify you as a user in our system; to operate, maintain, and improve our Site, Apps, and Services; to personalize and improve your experience." The description for how data is used is purposefully vague and allows a range of applications, perhaps for benefit to the student but also for the commercial interests of the company.

The point we want to emphasize here is that these types of sharing that occur between students, universities, and platforms are highly determined through an interconnected array of legal agreements, ranging from copyright and privacy laws, local institutional policy, contracts, terms of use, and platform privacy and copyright policies. Still, the circulatory act of sharing inputs by itself enables downstream parties to establish the kinds of access that enable platforms to make derivative products from student and instructor inputs on the basis of their fair and transformative use. Indeed, the legal ambiguity of carving out the right "to personalize and improve your experience," as contractual language that defines "what data the contractor may collect...and how it should be used" is nearly limitless. That universities may identify nearly any reason as a "legitimate educational interest" enables university administrators to freely designate nearly any corporation or individual as a "school official," while simultaneously asserting that they are not sharing student data with outside parties.

Combined, FERPA, LMS, and LTI constitute an assemblage of "circulation gatekeepers" (Edwards, 2018) that enable universities and university administrators to engage in a kind of handshake agreement with platform developers to ensure that they are legally meeting the provisions for protecting student information outlined in FERPA, while also failing to protect students from the types of downstream uses that might be in conflict with their interests. Through agreements between LMSs and third-party applications, schools relegate their stewardship of these data and records, ceding access to and power over any outputs generated. We suspect that the specific contracts governing how platforms make use of data and information are rooted in the same ambiguous language that governs use of data found in Terms of Use and Privacy Policies. Indeed, platforms that use policies containing such vague language could argue that nearly any use of data contributes to improvement of their services. In other words, FERPA might contain provisions that require uses to be disclosed, but those specific uses (e.g., improving user experience) might be so ambiguous as to extend privacy protections that are practically meaningless to students.

Granting LMSs (and third-party tools that plug in to LMSs) access to students' intellectual property and the digital byproducts of learning and teaching can be viewed as deeply problematic for a host of reasons, but the most significant is that it can facilitate the type of uncritical, coercive exchange that enabled PDSs such as Turnitin to amass heaps of profit from the academic labor that students perform in schools and universities. Moreover, as we discussed in the TURNITIN section, the A.V. ex. rel. Vanderhye v. iParadigms (2009) ruling seemed to suggest that once platforms are granted access to these sources of IP, the fair use provisions of copyright law apply, enabling toolmakers to innovate and build derivative products such as database and proprietary algorithms that make use of student inputs. Textbook publishers, reinventing themselves as designers of eLearning products, understand that these derivative works have immense value. eReaders, proprietary learning analytics algorithms, proprietary databases, engagement trackers, and more are the next wave of tools on the horizon, and LMSs provide the toolmakers with gateways necessary for accessing student IP. Students using these tools in classes (and the data they generate) are an important and often unknowing participants within platforms' research and development cycles. Yet the types of scrutiny universities have applied to human subjects research and research data haven't been extended toward digital tools and behavioral data, which is concerning from an ethical perspective.


IP Cast 9. Canvas Skill for Alexa. Tim uses the Canvas Skill for Alexa to access his courses in Canvas, demonstrating how easy it is for third parties to capture content, data, or metadata that passes through a LMS. A transcript of the conversation has also been provided.

Copyright

Disclaimers

The information and ideas contained in this webtext are not intended to be understood as legal advice, but rather as an exploration of the potential tensions that may exist between how authorship functions as a legal concept and how authorship is practiced and theorized in educational contexts.

Webtext design adapted from © Escape Velocity by HTML5 UP under a Creative Commons CC BY 3.0 license.