Recap post - 2025
I plan, for as long as this newsletter lives, to make the last post each year a 100-word (or less) summary of each post published earlier that year, sorted in chronological order. Thanks for reading, and Happy New Year!
1: GenAI’s Impact on Critical Thinking
A CHI 2025 paper examines the relationship between GenAI usage and critical thinking, finding that “higher confidence in GenAI is associated with less critical thinking, while higher self-confidence is associated with more critical thinking.” I think schools should foster intellectual growth without completely shunning AI tools (which is impossible anyway), but that’s much easier said than done. If AI tools become so powerful that nobody can outthink them (see, e.g., chess), then at that point, we’ll probably face a vastly different set of problems.
2: Productive Failure
A SIGCSE 2025 paper used real-time heart-rate data to infer cognitive load in a comparison between Direct Instruction (DI) and Productive Failure (PF); its findings “suggest that PF-based approaches may lead to more robust learning.” In DI, students receive a lesson before attempting a practice task, while this order is reversed in PF. Cognitive load is typically something that instructors should reduce, but in this case, the authors say that “appropriately subjecting students to an initially heavy cognitive load may be fruitful for learning.”
3: Perceived Obligations
Active learning is a huge deal in STEM education, so why don’t more instructors incorporate it? A paper published in Educational Studies in Mathematics in January 2025 investigates: the authors worked with three mathematicians and found that they felt obligated to cover a (large) set of topics, which left them with insufficient time for active learning. After some informal experiments with entrance/exit tickets, the researchers and mathematicians converged on a solution that had students answering multiple-choice questions that the instructors had planned on answering in class anyway.
4: Rawls vs. Einstein
According to a 1975 survey, 94% of faculty consider themselves “above-average teachers.” That sounds implausible, but to me, it’s not obvious what it even means for one teacher to be “better” than another. Sometimes, perhaps the teacher should focus on maximizing the minimum score, but other times, perhaps they should focus on maximizing the maximum score! I think these two approaches, while extreme, illustrate the balancing act of teaching large classes. In the figure below, both sections clearly improved over the baseline, but it’s hard to say which section’s outcomes are “better.”
5: (Un)productive Failure (and Success)
Another post about productive failure: the topic of this one — a 2016 article by Manu Kapur published in Educational Psychologist — was cited by the SIGCSE 2025 paper from earlier. In it, Kapur criticizes unguided problem solving as unproductive failure, highlights the “problem-solving phase followed by a consolidation (or instruction) phase” approach as productive failure, discusses problem-based learning as an example of productive success, and speculates that rote memorization is an example of unproductive success. Pedagogical experiments are good, but I think variety, or at least a bit of it, is also good.
6: Grade Deflation at Princeton
In 2004, to combat grade inflation, the faculty of Princeton adopted a policy that set an “expectation” of 35% A-grades (i.e., A-, A, or A+) across undergraduate courses in each department. The percentage of A-grades at Princeton decreased from 46 in 2003–04 to about 40 in 2005–11. After a three-year uptick, the 35% expectation was removed in 2014. In 2024–25, the percentage of A-grades at Princeton was 66.7, and a committee released a report that focuses on reducing the percentage of A+ grades. As I wrote in the post, maybe there should be a limit for each department…