Day: August 30, 2026

I have written on several occasions about the educational potential of debate. Researchers who focus on this topic often use the alternative term “argumentation“. Debate requires a deep understanding of the contested topic and offers a concrete way to exercise critical thinking. It also involves a motivational element because it involves competition. I read some educational research on debate and made a personal connection, noting how quality argumentation differed from the arguments I read on social media. Most of my previous posts about argumentation are available on my blog. 

A couple of years ago, I encountered a proposal that AI chatbots could serve as debate partners and saw a connection to Deanna Kuhn’s research, which investigated text messages as a way to develop argumentation skills because the process created a written record that could be analyzed and discussed. I saw a similar opportunity in the written record produced in an AI chat and the AI system could provide an opponent. 

I had done little more with this idea until recently, but I began thinking about a more formal debate format and decided to see how AI would implement the format I had in mind. 

Here is the prompt I have been exploring.

I want you to debate me on the topic of whether a computer and keyboard or handwriting on paper is the best note-taking approach. First, write a 200-word position statement supporting handwriting, and I will do the same for using a keyboard. After evaluating my position, write a challenge to my position. I will reply and challenge your position.

I chose this topic because I have read much of the related research and popular press and I think I understand the issues and competing positions. I also argue the less popular digital approach. This combination seemed a good combination to evaluate the experience. 

The change I made from my previous approach to an AI debate was to add the expectation that each participant create an introductory position the two parties a way to identify what appeared to be issues that were in disagreement. One more thing, while I first explored this experience using an AI subscription I use (Claude), I then tried the Mac mini on which I self-host an AI system (running Ollama) and repeated the process. The self-hosted experience was great and had the strange benefit of being to take an hour or so off for lunch and come back to the computer with the debate still active. I was particularly impressed by the content the AI was able to through at me. This Mac was not connected to the Internet and the only explanation I can offer is that I use the same setup to run a Karpathy Wiki and I have a great deal of relevant content within that system.

I have a substantial amount of content stored from this session and I often include such material as an Appenedix, but this was simply too much material. As an alternative, I decided I would upload the transcript of the entire debate to NotebookLM (Gemini Notebook) and include a link to this notebook for anyone interested in reading the transcript. 

I had one final idea. What if I use the AI embedded with the notebook to ask for an evaluation of the arguments and a winner?

Even if you don’t examine the transcript of the debate, NotebookLM’s evaluation gives you a sense of the level of the content and the interaction produced. I was a little disappointed because I thought my position was beyond logical refutation, but Notebook seemed unwilling to declare me the clear winner. The experience was not unlike actual debates. Can you get your opponent to engage with you on what you want the judge to accept as the key or deciding issues? 

Prompt: I want you to evaluate this debate considering whether taking notes by hand or using a computer is best in preparing a learner for tests based on lecture content. Are the best arguments made by the pro-keyboard or pro-handwriting supporter?

Evaluating this debate through the lens of preparing for tests based on lecture content reveals a fundamental tension between efficient information storage and active cognitive processing.

Both sides offer strong, distinct arguments depending on the type of test being taken and the student’s studying habits.

The Pro-Handwriting Case: Best for Conceptual Tests and Internal Recall

The pro-handwriting supporter makes the most compelling arguments regarding internalization, conceptual understanding, and performance under actual exam conditions.

• The “Recall” vs. “Retrieval” Advantage: In an exam setting, students do not have access to their devices or a search engine. The handwriting supporter argues that digital note-taking promotes “cognitive offloading” and “retrieval” (knowing where to find a fact in a database) rather than “recall” (pulling a concept from internal memory). Handwriting forces the brain to do the heavy lifting immediately, creating the robust neural pathways needed to recall information under pressure without a digital “crutch”.
• Performance on Conceptual Questions: Backed by the Mueller and Oppenheimer (2014) study, the handwriting advocate highlights that because typing is fast, it leads to mindless “transcription”. Because physical writing is slower, hand-writers must summarize and prioritize key concepts in real-time. This “selective encoding” means they perform significantly better on conceptual and synthesis test questions.
• The “Sieve” vs. “Bucket” Study Cycle: Handwriting acts as a “sieve” that filters information during the lecture. When test preparation begins weeks or months later, the hand-writer reviews a distilled, pre-processed “mental map”. In contrast, a typed “bucket” of notes requires the student to spend limited study time filtering a mountain of raw, flat data before they can even begin deep synthesis.

The Pro-Keyboard Case: Best for Detailed Tests and Long-Term Study Flexibility

The pro-keyboard supporter makes highly sophisticated arguments regarding information completeness, managing cognitive load, and the reality of long-term study cycles.

• Performance on Detail-Oriented Questions: The keyboard allows students to type fast enough to capture almost everything accurately. For exams that test factual details and specific data points, the pro-keyboard side is heavily favored, as research indicates laptop users perform better on detail-oriented questions due to their comprehensive records.
• Managing Lecture Cognitive Overload: The keyboard supporter rightly points out that a fast-paced lecture is not the ideal setting for deep conceptual thinking. Attempting to synthesize while keeping up with a speaker can cause cognitive overload. Keyboarding minimizes the friction of capture. This “offloads” the processing during class so the student can focus on capturing data, leaving the deep reflection to a low-pressure study environment.
• Iterative Study and Optimization: Digital notes are highly malleable. Features like searchability, audio-linking, and inserting placeholders (like “XXX”) for missed content allow students to efficiently patch gaps, reorganize ideas, and elaborate on notes over a semester-long period. This active post-processing provides a substantial generative advantage over rigid, handwritten pages that are cumbersome to restructure.

Conclusion: Who Makes the Best Arguments?

The pro-handwriting supporter makes the superior psychological and cognitive arguments for final test performance. Their points are deeply anchored in established learning science—such as “desirable difficulty” and the “generation effect”—proving that the physical friction of writing is a feature that directly constructs a student’s internal cognitive map. Under test conditions, a “remembered concept” will always outperform a “searchable note”.

However, the pro-keyboard supporter makes the more realistic argument regarding modern study workflows. If a student is highly disciplined and actively uses the study phase to review, edit, and link their digital notes, the digital “workspace” provides a highly scalable and superior infrastructure for long-term learning.

Ultimately, if the goal of note-taking is to build a high-functioning cognitive map to ace conceptual exams, the pro-handwriting arguments are the strongest. If the goal is to build a durable, searchable archive for detailed reference, the pro-keyboard arguments win out.

Would you like me to create an interactive study guide or a set of flashcards based on these opposing cognitive theories to help you further analyze these learning principles?

References:

Kuhn, D., Goh, W., Iordanou, K., & Shaenfield, D. (2008). Arguing on the Computer: A Microgenetic Study of Developing Argument Skills in a Computer-Supported Environment. Child Development, 79(5), 1310-1328

Mayweg-Paus, E., Macagno, F., & Kuhn, D. (2016). Developing Argumentation Strategies in Electronic Dialogs: Is Modeling Effective. Discourse Processes, 53(4), 280–297. https://doi.org/10.1080/0163853X.2015.1040323