Beyond Perspectives: A Trio-Ethnography of Interpretation Evolution in LLM-Supported Programming Education

Jennie Ren; Jordan H. McDowell; Kyrie Zhixuan Zhou

REVIEW 3 major objections 4 minor 30 references

Beyond Perspectives: A Trio-Ethnography of Interpretation Evolution in LLM-Supported Programming Education

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A three-person reflective dialogue can reveal how students actually learn with AI, beyond what their submitted code shows.

desk verdict A candid experience report whose reflective value is real, but whose claim to yield 'more accurate' interpretations is not supported by its self-referential design. read the letter →

arxiv 2607.22463 v1 pith:VUP2K7EM submitted 2026-07-24 cs.HC cs.AI

Jennie Ren , Jordan H. McDowell , Kyrie Zhixuan Zhou This is my paper · ORCID

classification cs.HCcs.AI

keywords computingeducationgenerativeAIlargelanguagemodelstrio-ethnographyprogrammingpedagogyAI-supportedlearningeducatorreflectionstudentperspectives

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that structured three-person dialogue—two computing educators with different teaching styles and one undergraduate student—can give educators a more accurate picture of how students learn with AI than classroom observation alone. Across three conversations, the educators' initial assumptions about AI use did not simply get confirmed or rejected; they evolved. The student's accounts showed learning steps that leave no trace in submitted code, such as reading, modifying, and practicing with AI-generated answers. By treating student dialogue as evidence, the educators revised their view of AI as a tutor rather than an answer provider, and began to redesign assessment and instruction around reasoning rather than final products. If this works, trio-ethnography offers a practical reflective method for computing educators adapting to generative AI.

What carries the argument

The carrying mechanism is trio-ethnography: a structured reflective conversation among three co-inquirers, here two educators with contrasting teaching philosophies and one student who regularly uses generative AI. The study implements it in three stages—an initial educator duo-dialogue, a semi-structured student interview, and a reflective reconstruction where educators re-read their earlier positions in light of student evidence. The method works by turning the student's lived experience into evidence that can confirm, extend, or complicate educators' prior interpretations, so the dialogue itself is the instrument that produces the more accurate understanding.

What would settle it

Give a diverse group of students the same AI-assisted programming task while recording their interaction logs, code revisions, and think-aloud comments; then run a trio-ethnographic dialogue with their educators. If educators' post-dialogue interpretations match the logged behavior substantially better than their pre-dialogue interpretations did, the central claim survives. If the dialogue leads educators away from the logged evidence, or if no improvement appears, the claim that dialogue produces more accurate interpretations is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that educators can reach more accurate interpretations of students' AI-supported learning by engaging a student in sustained dialogue, and that these interpretations evolve through that dialogue rather than appearing at once. The student's narratives confirmed some educator interpretations, such as the importance of AI transparency and AI as a learning partner, and complicated others: the same submitted code may result either from copying an AI answer or from a long invisible process of reading, note-taking, practice, and verification. This distinction led the educators to move from treating AI as a classroom-policy issue to treating it as a teachable part of pro

Load-bearing premise

The load-bearing premise is that the student's self-reported account of learning is accurate enough to serve as ground truth for what 'more accurate' educator interpretations mean; the study does not independently verify those learning processes against code, logs, or other behavioral evidence.

Editorial extensions

If this is right

If educators adopt this reflective practice, AI policy in programming courses should shift from regulating use to explicitly teaching when and how AI supports learning.
Assessment designs should evaluate reasoning and process—explanations, justifications, reflections—not only final code.
Instruction should include explicit support for debugging and for comparing AI-generated code with one's own attempts.
The 'one more step' idea expands into multiple post-AI learning activities that instructors can make visible through reflection prompts and process documentation.
Future work with more diverse students is needed to test whether the patterns hold beyond highly motivated learners.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Beyond the paper: the same trio-ethnographic structure could be applied to other AI-supported skill domains—writing, data analysis, design—wherever final artifacts hide process.
Beyond the paper: the findings generate a testable hypothesis that learning gains from AI-generated code depend on what students do after receiving the answer; a comparison of process-scaffolded versus unsupported AI use could quantify the value of the invisible steps.
Beyond the paper: since the study relies on one self-reported, highly motivated student, a natural next step is to pair trio-ethnography with interaction logs or code-revision histories; that would give educators an independent check on whether dialogue-based interpretations track actual behavior.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear colleague,

Quick take: this is a thoughtfully written experience report on using trio-ethnography to help computing educators reflect on how students learn with LLMs. The reflective value is real. But the paper's stronger claim—that the dialogue produces 'more accurate' interpretations of student learning—outruns its evidence. The authors are the participants, the student's self-report is the only window on learning, and there is no external benchmark against which 'accuracy' is measured.

What is genuinely useful: the paper moves trio-ethnography from comparing perspectives to tracing how interpretations evolve. The three themes (AI stigma → transparency → explicit guidance; AI as answer provider → tutor; 'one more step' → multiple invisible steps) are coherent and well illustrated with quotes. The authors are honest about the single, highly motivated student in §5.2, and the limitations section is not pro forma. As a self-study, it demonstrates a structured way for instructors to surface learning processes that assignments hide.

Where it is soft: the circularity is load-bearing. The 'student dialogue' is produced by a co-author, the educators are the researchers, and the coding/analysis is done by the same people. So claims in §3.3 and §5.1.3 about 'narrowing the gap' and 'more accurate understandings' are, at bottom, participants' retrospective assessments of their own change. That's a legitimate form of reflective evidence, but it is not evidence about actual accuracy unless it is triangulated with interaction logs, artifacts, or independent coding. The paper would be stronger if it reframed the outcome as 'perceived refinement' or 'reflective insight,' and left 'accuracy' to studies with external validation.

The analytic framework is weakest when it treats the student's account of 'notes to take down, I practice... muscle memory' as evidence of learning processes without questioning whether those activities actually improved learning. That's a minor concern given the paper's scope.

Bottom line: the central argument holds as an experience report, not as an empirical demonstration. I'd send it to peer review and ask for a substantial revision—reframe the accuracy language, add a reflexivity statement, and make the protocol and transcript excerpts more complete. It's useful for computing education researchers interested in qualitative methods and AI pedagogy, and it deserves a serious referee rather than a desk reject.

Referee Report

3 major / 4 minor

Summary. This experience report describes a trio-ethnography involving two computing educators (R1, R2) and one undergraduate computer science student (R3), all of whom are also the paper's authors. Over three stages—an initial educator dialogue, a semi-structured student interview, and a reflective reconstruction—the paper traces how the educators' interpretations of students' AI-supported programming learning evolved. It reports four themes: moving from AI stigma to transparency and explicit guidance; repositioning AI from answer provider to learning partner/tutor; recognizing learning processes invisible in final code; and reconstructing teaching beliefs about assessment, debugging, and active learning. The central methodological claim is that trio-ethnography can narrow the gap between observable student behavior and students' actual learning processes, leading to 'more accurate' educator interpretations.

Significance. The paper's reflective method is timely and potentially useful: structured dialogue between educators and a student can surface perspectives that classroom artifacts hide, and the study transparently reports its procedure, data excerpts, limitations, and ethics. The 'invisible learning' theme and the pedagogical implications for assessment and debugging instruction are plausible and actionable. However, the paper's central claim—that the triad achieved 'more accurate' interpretations—is not supported by the design, because accuracy is assessed without any external benchmark. The study is best read as an existence proof that trio-ethnography can prompt reflective belief change; with recalibrated claims, it would be a useful contribution to computing-education practice.

major comments (3)

[Abstract, §3.4, §5.1.3] The paper repeatedly claims that the dialogue produced 'more accurate' interpretations and 'narrowed the gap' between educators' and students' actual learning. The design provides no external measure of accuracy: the data consist only of self-produced transcripts, the participants are the researchers, and the 'evolution toward accuracy' is judged by the same individuals whose beliefs are the object of study. The reflective value of the method survives this concern, but the accuracy language exceeds the evidence. Please either add independent corroboration (e.g., code artifacts, interaction logs, think-aloud data, or external raters) or reframe the contribution as producing 'more nuanced,' 'better informed,' or 'revised' interpretations.
[§4.3, §5.2] The claim that student dialogue revealed 'invisible learning' rests on a single student's self-report. The quotation in §4.3—'Usually I have notes to take down, I practice...'—is accepted as evidence of actual learning processes without triangulation, and §5.2 itself notes that the student was a highly motivated learner. This supports an existence proof for the reflective potential of the method, but not a robust empirical description of student learning. Please scope the conclusions to this single participant and make clear that the learning activities were reported, not independently observed.
[§3.2, §7] The authors are simultaneously the participants, the interviewers, and the analysts. The collaborative coding and reflective reconstruction are performed by R1 and R2, the same individuals whose interpretive evolution is the outcome. This creates a circularity risk that is acknowledged only indirectly through the ethics statement. The manuscript should explicitly discuss how the analysis guarded against confirmation bias (e.g., an external analyst, an audit trail, or a preregistered coding scheme), or restrict the claims to self-reported belief change rather than objective interpretive accuracy.

minor comments (4)

[§5.1.2] The subsection is labeled 'A third implication' but there is no second implication heading; §5.1.1 presents a 'key implication' with two directions. Renumber or label the implications consistently.
[§3.3] The term 'student dialogue' is used both for the Stage 2 interview and, in places, for the entire triadic process. Clarify the terminology to distinguish the interview from the reflective reconstruction.
[§2.1] Several citation clusters bundle four or more references (e.g., [4, 24] and [6, 17, 19]) without distinguishing which claim each supports. Consider separating them for readability and verifiability.
[§7] The ethics statement says pseudonyms R1, R2, and R3 are used, but these are not pseudonyms; they are arbitrary labels. Consider adding actual pseudonyms or clarifying that these are anonymized identifiers.

Circularity Check

1 steps flagged · score 6.0 of 10

Accuracy claim is self-referential: the student dialogue is both the evidence and the criterion for 'more accurate' educator interpretations.

self definitional [§3.3 Stage 3 and §5.1.3, with §7 authorship disclosure]
"Following the student dialogue, the educators revisited their original discussions, treating the student’s dialogue as evidence to refine and reconstruct their interpretations of students’ AI-supported learning. This reflective process reconstructed the educators’ interpretations ... enabling more accurate understandings of students’ AI-supported learning. ... This experience report presents a trio-ethnographic inquiry among the authors."

The claimed outcome—'more accurate understandings of students’ AI-supported learning'—is evaluated against the very student dialogue that is also the method's only input. The student (R3) is a co-author; the educators (R1, R2) are the authors who designed the prompts and performed the analysis. No independent trace (interaction logs, code artifacts, think-aloud, external benchmark) of the student's actual learning processes is used. Therefore 'accuracy' is operationally defined as agreement with the student's self-narrative, making 'dialogue produces more accurate understanding' true by construction rather than empirically demonstrated.

full rationale

The paper's central methodological contribution is that trio-ethnography helps educators develop 'more accurate understandings' of students' AI-supported learning. The only evidence for this accuracy gain is the student's self-reported narrative, which is produced and interpreted by the same people who are the paper's authors. There is no external benchmark or independent measure of the student's actual learning processes; the paper itself discloses in §7 that the inquiry was conducted 'among the authors.' Thus the outcome measure is the same as the intervention's input: the student dialogue is used both to generate the new interpretations and to certify their accuracy. This is not an equation-level tautology, and the paper's weaker claim—that the exercise is reflective and prompts pedagogical reconsideration—is independently plausible. But the load-bearing 'more accurate' language reduces to an internal, self-defined criterion. The limitations section honestly notes the single highly motivated student, but that does not break the evaluative loop. No self-citation chains or imported uniqueness theorems are involved, so the circularity is localized to the accuracy claim; hence a score of 6 rather than higher.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new formal models, parameters, or invented entities. Its central claim depends on domain assumptions about self-report validity, the observation gap, and the evidentiary value of self-authored dialogue rather than on measured quantities.

assumptions (4)

domain assumption The student's self-reported learning narratives accurately represent how the student learned with AI.
The analysis treats R3's descriptions (e.g., §4.3 'usually I have notes to take down, I practice...') as evidence of actual learning processes, with no triangulation against code artifacts, logs, or observation.
domain assumption Educators' self-reported changes in interpretation constitute 'more accurate' understanding.
The claimed outcome is measured by the participants' own retrospective accounts (§3.4, §4.4), not by any independent measure of interpretive accuracy.
domain assumption Classroom-observable products (submitted code, assignments) underrepresent student learning processes.
This premise motivates the entire study and is stated in §1, but the paper provides no direct measurement of the observation gap.
domain assumption Trio-ethnographic dialogue is a valid method for generating evidence about interpretation evolution.
The paper relies on the qualitative research tradition (duoethnography, trioethnography) cited in §2.2; this is a methodological premise, not universally accepted as a replacement for external validation.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Beyond Perspectives: A Trio-Ethnography of Interpretation Evolution in LLM-Supported Programming Education." pith.science (2026). https://pith.science/paper/VUP2K7EM

@misc{pith2026260722463,
  author       = {Pith},
  title        = {Pith review of: Beyond Perspectives: A Trio-Ethnography of Interpretation Evolution in LLM-Supported Programming Education},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VUP2K7EM}},
  note         = {Machine review of arXiv:2607.22463}
}

read the original abstract

Generative AI is reshaping programming education, yet educators often infer students' AI-supported learning from classroom observations alone. This experience report presents a trio-ethnography involving two computing educators with different teaching philosophies and one undergraduate computer science student to examine how these interpretations evolve through dialogue. Across three conversations, the educators reflected on students' AI use, discussed changes to programming pedagogy, and revisited their assumptions after engaging with the student's lived experiences. Rather than simply confirming or contradicting the educators' perspectives, the student's narratives revealed learning processes that were largely invisible in the classroom, prompting both educators to reconsider assumptions about AI use, assessment, transparency, and programming instruction. We argue that trio-ethnography offers a valuable reflective approach for helping computing educators move beyond observable student behaviors toward a richer understanding of AI-supported learning and for informing instructional adaptation in the era of generative AI.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

30 extracted references

[1]

2016.Handbook of autoethnography

Tony E Adams, Stacy Holman Jones, and Carolyn Ellis. 2016.Handbook of autoethnography. Routledge

2016
[2]

Areej Ali, Aayushi Hingle Collier, Umama Dewan, Nora McDonald, and Aditya Johri. 2025. Analysis of generative AI policies in computing course syllabi. In Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1. 18–24

2025
[3]

Dario Luis Banegas and David Gerlach. 2021. Critical language teacher education: A duoethnography of teacher educators’ identities and agency.System98 (2021), 102474

2021
[4]

Brett A Becker, Paul Denny, James Finnie-Ansley, Andrew Luxton-Reilly, James Prather, and Eddie Antonio Santos. 2023. Programming is hard-or at least it used to be: Educational opportunities and challenges of ai code generation. In Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1. 500–506

2023
[5]

Yasemin Cakcak Tezgiden and Ufuk Ataş. 2024. Becoming and being a critical language teacher educator: A duoethnography. (2024)

2024
[6]

Doga Cambaz and Xiaoling Zhang. 2024. Use of ai-driven code generation models in teaching and learning programming: a systematic literature review. In Proceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. 172–178

2024
[7]

2016.Collaborative autoethnography

Heewon Chang, Faith Ngunjiri, and Kathy-Ann C Hernandez. 2016.Collaborative autoethnography. Routledge

2016
[8]

Paul Denny, Juho Leinonen, James Prather, Andrew Luxton-Reilly, Thezyrie Amarouche, Brett A Becker, and Brent N Reeves. 2024. Prompt Problems: A new programming exercise for the generative AI era. InProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. 296–302

2024

Show all 30 references

[9]

Paul Denny, James Prather, Brett A Becker, James Finnie-Ansley, Arto Hellas, Juho Leinonen, Andrew Luxton-Reilly, Brent N Reeves, Eddie Antonio Santos, and Sami Sarsa. 2024. Computing education in the era of generative AI.Commun. ACM67, 2 (2024), 56–67

2024
[10]

Smit Desai, Tanusree Sharma, and Pratyasha Saha. 2023. Using ChatGPT in HCI research—A trioethnography. InProceedings of the 5th International Conference on Conversational User Interfaces. 1–6

2023
[11]

Arto Hellas, Juho Leinonen, Sami Sarsa, Charles Koutcheme, Lilja Kujanpää, and Juha Sorva. 2023. Exploring the responses of large language models to beginner programmers’ help requests. InProceedings of the 2023 ACM Conference on International Computing Education Research-Volu...

2023
[12]

Muntasir Hoq, Yang Shi, Juho Leinonen, Damilola Babalola, Collin Lynch, Thomas Price, and Bita Akram. 2024. Detecting ChatGPT-generated code sub- missions in a CS1 course using machine learning models. InProceedings of the 55th ACM Technical Symposium on Computer Science Educa...

2024
[13]

Dhruv Jain, Venkatesh Potluri, and Ather Sharif. 2020. Navigating graduate school with a disability. InProceedings of the 22nd International ACM SIGACCESS Conference on Computers and Accessibility. 1–11

2020
[14]

Majeed Kazemitabaar, Justin Chow, Carl Ka To Ma, Barbara J Ericson, David Weintrop, and Tovi Grossman. 2023. Studying the effect of AI code generators on supporting novice learners in introductory programming. InProceedings of the 2023 CHI conference on human factors in comput...

2023
[15]

Ban it till we understand it

Sam Lau and Philip Guo. 2023. From" Ban it till we understand it" to" Resistance is futile": How university programming instructors plan to adapt as more students use AI code generation and explanation tools such as ChatGPT and GitHub Copilot. InProceedings of the 2023 ACM Con...

2023
[16]

Giang Nguyen H Le, Vuong Tran, and Trang Thuy Le. 2021. Combining photog- raphy and duoethnography for creating a trioethnography approach to reflect upon educational issues amidst the COVID-19 global pandemic.International Journal of Qualitative Methods20 (2021), 16094069211034391

2021
[17]

Rongxin Liu, Carter Zenke, Charlie Liu, Andrew Holmes, Patrick Thornton, and David J Malan. 2024. Teaching CS50 with AI: leveraging generative artificial intelligence in computer science education. InProceedings of the 55th ACM technical symposium on computer science education...

2024
[18]

Amy E Long, Rachel Wolkenhauer, and Mary Higgins. 2021. Learning through Research in a Professional Development School: A Duoethnographic Approach to Teacher Educator Professional Learning.School-university partnerships14, 1 (2021), 36–44

2021
[19]

Joseph Maguire, Rosanne English, Qi Cao, and Chee Kiat Seow. 2025. Themes in the declared use of generative artificial intelligence in assessment. InProceedings of the 9th Conference on Computing Education Practice. 17–20

2025
[20]

Lauren E Margulieux, James Prather, Brent N Reeves, Brett A Becker, Gozde Cetin Uzun, Dastyni Loksa, Juho Leinonen, and Paul Denny. 2024. Self-regulation, self-efficacy, and fear of failure interactions with how novices use LLMs to solve programming problems. InProceedings of ...

2024
[21]

James Prather, Juho Leinonen, Natalie Kiesler, Jamie Gorson Benario, Sam Lau, Stephen MacNeil, Narges Norouzi, Simone Opel, Virginia Pettit, Leo Porter, et al. 2024. How instructors incorporate generative AI into teaching computing. InProceedings of the 2024 on Innovation and ...

2024
[22]

It’s weird that it knows what i want

James Prather, Brent N Reeves, Paul Denny, Brett A Becker, Juho Leinonen, Andrew Luxton-Reilly, Garrett Powell, James Finnie-Ansley, and Eddie Antonio Santos. 2023. “It’s weird that it knows what i want”: Usability and interactions with copilot for novice programmers.ACM trans...

2023
[23]

James Prather, Brent N Reeves, Juho Leinonen, Stephen MacNeil, Arisoa S Randri- anasolo, Brett A Becker, Bailey Kimmel, Jared Wright, and Ben Briggs. 2024. The widening gap: The benefits and harms of generative ai for novice programmers. InProceedings of the 2024 ACM Conferenc...

2024
[24]

Sami Sarsa, Paul Denny, Arto Hellas, and Juho Leinonen. 2022. Automatic generation of programming exercises and code explanations using large language models. InProceedings of the 2022 ACM Conference on International Computing Education Research-Volume 1. 27–43

2022
[25]

2012.Duoethnography: Dialogic methods for social, health, and educational research

Richard D Sawyer and Darren Lund. 2012.Duoethnography: Dialogic methods for social, health, and educational research. Vol. 7. Left Coast Press

2012
[26]

Richard D Sawyer and Joe Norris. 2013. Duoethnography: Understanding quali- tative research

2013
[27]

Judy Sheard, Paul Denny, Arto Hellas, Juho Leinonen, Lauri Malmi, and Simon
[28]

JD Zamfirescu-Pereira, Laryn Qi, Björn Hartmann, John DeNero, and Narges Norouzi. 2025. 61a bot report: Ai assistants in cs1 save students homework time and reduce demands on staff.(now what?). InProceedings of the 56th ACM Technical Symposium on Computer Science Education V. ...

2025
[29]

Cynthia Zastudil, Magdalena Rogalska, Christine Kapp, Jennifer Vaughn, and Stephen MacNeil. 2023. Generative ai in computing education: Perspectives of students and instructors. In2023 IEEE Frontiers in Education Conference (FIE). IEEE, 1–9

2023
[2024]

InProceedings of the 55th ACM Technical Symposium on Computer Science Education V

Instructor perceptions of ai code generation tools-a multi-institutional interview study. InProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. 1223–1229

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

2016.Handbook of autoethnography

Tony E Adams, Stacy Holman Jones, and Carolyn Ellis. 2016.Handbook of autoethnography. Routledge

2016

[2] [2]

Areej Ali, Aayushi Hingle Collier, Umama Dewan, Nora McDonald, and Aditya Johri. 2025. Analysis of generative AI policies in computing course syllabi. In Proceedings of the 56th ACM Technical Symposium on Computer Science Education V. 1. 18–24

2025

[3] [3]

Dario Luis Banegas and David Gerlach. 2021. Critical language teacher education: A duoethnography of teacher educators’ identities and agency.System98 (2021), 102474

2021

[4] [4]

Brett A Becker, Paul Denny, James Finnie-Ansley, Andrew Luxton-Reilly, James Prather, and Eddie Antonio Santos. 2023. Programming is hard-or at least it used to be: Educational opportunities and challenges of ai code generation. In Proceedings of the 54th ACM Technical Symposium on Computer Science Education V. 1. 500–506

2023

[5] [5]

Yasemin Cakcak Tezgiden and Ufuk Ataş. 2024. Becoming and being a critical language teacher educator: A duoethnography. (2024)

2024

[6] [6]

Doga Cambaz and Xiaoling Zhang. 2024. Use of ai-driven code generation models in teaching and learning programming: a systematic literature review. In Proceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. 172–178

2024

[7] [7]

2016.Collaborative autoethnography

Heewon Chang, Faith Ngunjiri, and Kathy-Ann C Hernandez. 2016.Collaborative autoethnography. Routledge

2016

[8] [8]

Paul Denny, Juho Leinonen, James Prather, Andrew Luxton-Reilly, Thezyrie Amarouche, Brett A Becker, and Brent N Reeves. 2024. Prompt Problems: A new programming exercise for the generative AI era. InProceedings of the 55th ACM Technical Symposium on Computer Science Education V. 1. 296–302

2024