Flextensions

Flextensions is a tool for automating assigment extensions—from student request to instructor approval. Flextensions currently works with Google Forms and Gradescope. Canvas/bCourses support is in active development.

AY24-25 Student leads: Edwin Vargas Navarro, Conan Smallwood

Faculty contact: Lisa Yan and Michael Ball, EECS/Data Seamless Learning Team

Recent Updates

Project Overview

[minor edits to original Teach-Net post, January 2024]

The education literature tells us that flexible (soft) deadlines reduce student stress, keep students accountable for their own learning (and meta-learning), and ensure that the course grade represents the work that students have submitted rather than being distorted by late penalties (Feldman, 2018). At Berkeley, students are requesting more accommodations than prior to the pandemic for many reasons (DSP accommodations, ASE strike, pandemic, personally affected by global crises, etc.)

With the Flextensions system we’ve been developing, students submit a Google Form detailing their extension request—which assignment, how many days requested, and why. If the requested extension is less than an instructor-specified threshold (e.g., 2 days), Flextensions notifies the student by email that it’s been approved. For assignments submitted via Gradescope, the tool automatically updates Gradescope to accept that student’s assignment late. (We’re working on similar automation for bCourses.) If the request is for longer than that threshold, the course staff are notified and can ask the student to either exchange email with or meet with a member of course staff (instructor-settable) before the extension is requested, just to make sure there are no systemic or student-distress issues that may need attention in order to keep the student on track for success. (Anecdotally, a number of students have reported being pleasantly surprised and relieved when their professor meeting wasn’t to chew them out for being late, but to actually check on their well-being.)

In other words, Flextensions helps centralize the majority of administrative work required to track extensions for individual students.

Since the prototype was launched by Computer Science TAs half a decade ago—lecturer Peyrin Kao was a huge student contributor before he joined the faculty—Flextensions has been adopted by many large CDSS and EECS courses including Data 8, C88C, 100, the CS 61 series, CS 161, and CS 169. These courses have redesigned course policies to minimize abuse of the soft deadlines system—of which there’s minimal, particularly at the upper-division level. Ultimately, it’s been a huge benefit to us in our giant STEM courses: ~90% of requests are quick and one-time, and (u)GSIs can immediately identify students who need instructor meetings or more support. We’ve been studying various components of the project, including how it could close learning equity gaps, promote student self-efficacy, and reduce staff logistical burden. We will be working on supporting bCourses assignments in the future.

So far, we’ve only tried the Flextensions paradigm in CS/Data courses, where assignments are majority autograded. Extensions with non-autograded assignments are currently constrained by staff grading capacity. We’d love to hear if you have ideas for how this could work in natural sciences, social sciences, or humanities courses for a subset of assignments, say, those submitted via online bCourses portals.

Overview of the Student Experience

Students submit a short Google form to request an extension on a particular assignment, indicating how long they want and a brief reason for the request. Most forms are automatically processed with no staff intervention: for example, DSP-related requests, or extensions shorter than an instructor-specified threshold. In these cases, Flextensions notifies the student by email that the extension was approved, and if Gradescope is being used to accept submissions, Flextensions updates Gradescope to allow a later submission time for the student on that assignment. (We are working on similar automation for bCourses and other online submission systems.) In other cases, Flextensions notifies the student that their extension will be manually approved after they either email more details or meet directly with a TA or instructor – this ensures a “contact point” so the staff can make sure the student isn’t experiencing more systemic problems. If your course staff uses Slack for communication, Flextensions can also send a Slack message to a designated channel to notify course staff when new requests arrive.

For more information, check out our slides from the 2024 Teaching and Learning Conference at UC Berkeley, presented by Jordan Schwartz. Please also see our publications below.

Access GitHub

Research and Development

There are two main academic thrusts for Flextensions:

Since 2023, this project has spawned numerous opportunities for undergraduate and Master’s student research.

Proceedings and Presentations

Additional Information

Tool Support

At present, Flextensions is supported by generous time from instructors and course staff using this tool in their classes. We are looking for a way to support this tool more widely for any course instructors, regardless of software fluency.

History and Faculty involvement

Flextension is a tool built on the shoulders of many faculty, TAs, and students in EECS and Data Science. The public Flextensions tool was developed in Spring 2022 CS161: Computer Security, taught by Nick Weaver. Several TAs including Shomil Jain, Peyrin Kao, Vron Vance, and Zephyr Barkan wrote and contributed to the first iteration of this tool, along with an accommodations policy that encouraged student usage. Fuzail Shakir wrote the first Flextensions experience report in his May 2023 MS EECS Technical Report, advised by Armando Fox.

Starting Summer 2023, Flextensions software and maintenance has been integrated into software development and maintenance projects sponsored by the Seamless Learning Team, run by Michael Ball and Lisa Yan. In Fall 2023, Narges Norouzi was integral in kicking off research projects studying the impact of Flextensions on student learning. Also in Fall 2023, this tool became widely used by almost all Data Science Undergraduate Studies (DSUS) courses; Silas Santini, DSUS Software Engineer, has supported software tool onboarding for different DSUS courses.