Pith. sign in

REVIEW 2 major objections 4 minor 27 references

The Stress of Improvisation: Instructors' Perspectives on Live Coding in Programming Classes

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Instructors report that live coding imposes more mental stress than presenting static slides, driven by the improvisational and unpredictable character of writing code in front of a class.

desk verdict Solid small qualitative study, but the abstract oversells the stress comparison; should be revised before acceptance. read the letter →

arxiv 2506.03402 v1 pith:GZHVONNL submitted 2025-06-03 cs.HC

Xiaotian Su , April Wang This is my paper
classification cs.HC
keywords livecodingprogrammingeducationinstructorexperiencementalworkloadcognitiveloadcontextualinquiryteachingassistantsstagefright
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 live coding — writing and executing code in front of a class — puts a heavier mental burden on instructors than presenting static slides, and that the burden comes from the format's improvisational, unpredictable character. Based on interviews with five teaching assistants and contextual observation of four lecturers, it finds that unpredictability makes time management and student engagement hard to control, and that this drives stress from thinking aloud, multitasking, fear of mistakes, and stage fright. The paper's point is that instructor-side cognitive load is a real, understudied cost of a technique usually evaluated only by student outcomes. If the claim holds, the case for designing better classroom tools — simplified IDEs, real-time code sharing, adaptive presentation, AI assistance — becomes concrete rather than speculative.

What carries the argument

The load-bearing mechanism is the improvisational, unpredictable character of live coding, which turns teaching into a dual task: the instructor must write and debug code in real time while verbalizing reasoning and monitoring student engagement. Deviations from prepared scripts, student questions, and unexpected bugs amplify the unpredictability, and the paper traces the resulting stress through four channels — think-aloud load, multitasking, fear of mistakes, and stage fright. The supporting method is a two-part qualitative design: formative interviews with five teaching assistants capture reflective accounts, while contextual inquiry with four lecturers — classroom observation plus follow-up interview — grounds those accounts in observed behavior. This combined structure is what lets the paper compare live coding against static slides on the instructor's own reported experience.

What would settle it

A direct test would be a between-subjects or crossover study in which instructors teach matched topics once by live coding and once by static slides, with physiological measures (heart rate variability or skin conductance) and a validated workload questionnaire, plus a larger and more diverse instructor sample. If such measures show no systematic difference in stress between the two formats, or show that experienced instructors are unaffected, the paper's comparative stress claim would not generalize.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the improvisational and unpredictable nature of live coding is what makes it stressful for instructors: it makes time management difficult and threatens student engagement, and the net effect is more mental stress than teaching from static slides. The evidence is qualitative and consistent across the two studies: all teaching assistants and all but the most experienced lecturer rated live coding as more mentally demanding than conventional lectures. The paper identifies distinct sources of that load — thinking aloud while typing, multitasking between code, explanation, and audience, fear of making mistakes, and stage fright in front of a large room — and argues that these are not incidental but inherent to live coding's improvisatory character. Accordingly, the paper frames the design implication as supporting instructors in real time rather than assuming they can simply 'code better' under pressure.

Load-bearing premise

The claim rests on the assumption that nine self-selected instructors from one university, five of them teaching assistants and four of them lecturers, are representative enough of programming instructors generally, and that their self-reports of mental workload and stress track their actual in-class experience.

Editorial extensions

If this is right

  • If the claim holds, instructor cognitive load becomes a first-class design constraint for programming education tools, not just an afterthought.
  • Live coding's time cost — up to twice as long as static slides — should be treated as a real constraint in lecture planning, and tools that shorten debugging or preparation time would directly reduce stress.
  • Tailored educational IDEs with minimalist interfaces, automatic syntax correction, and guided templates could lower the unpredictability that drives instructor stress.
  • Real-time code sharing and classroom-aware collaboration features (access control, student annotation, in-IDE questions) could relieve the multitasking burden without losing live coding's engagement benefits.
  • AI-assisted debugging and adaptive content presentation could help instructors respond to unexpected student questions, the main source of deviation from prepared plans.

Reading between the lines

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

  • Beyond the paper, this suggests a possible feedback loop the authors do not name: as stress rises, instructors become more absorbed in debugging and lose the audience, which worsens engagement and raises stress further; tool support that externalizes debugging could interrupt that loop.
  • A testable extension would be to measure whether the stress gap between live coding and slides shrinks with teaching experience; the paper's data hint that the most experienced lecturer reported no added demand, implying a possible practice effect worth quantifying.
  • The preparation overhead finding implies that much 'improvisation' is actually rehearsed; if so, evaluating tools should consider both in-class and pre-class time, not just lecture minutes.
  • The paper's single-university sample invites a cross-institutional replication, and also a comparison with other performance-based teaching formats (e.g., chalkboard derivations in mathematics), to see whether the stress is specific to code or general to live exposition.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 4 minor

Summary. This extended abstract reports a qualitative study of instructors' experiences with live coding in programming classes. The authors carried out two complementary studies: formative interviews with five teaching assistants and contextual inquiries (classroom observations plus follow-up interviews) with four university lecturers. They report motivations for live coding (comprehension, adaptivity, engagement, modeling good practices), implementation strategies (preparation, procedure, student activities), and obstacles (decline of engagement, unpredictability, mental stress, time pressure), and they translate these into design implications for IDEs and presentation tools. The abstract's headline claim is that the improvisational nature of live coding makes time management and student engagement difficult, 'resulting in more mental stress than presenting static slides.' The paper is transparent about its small sample and self-reported data in Section 5.2, but the central comparative claim needs to be aligned with the evidence actually presented.

Significance. The manuscript addresses a genuinely underexplored perspective: prior live-coding research has focused largely on student outcomes, whereas the instructor-side cognitive and psychological costs have been mostly hypothesized. The paper's strengths include its use of both reflective interviews and in-situ classroom observations, its honest account of abandoning an initial lab-based simulation, the inclusion of concrete participant quotes, and its clearly stated limitations. If the findings are read as instructors' perceived stress rather than as a measured comparative effect, they offer useful direction for tool designers and for future quantitative work. The main weakness is that the abstract overstates what the data can support. With the central claim appropriately reworded and the deviant case (P8) treated as a substantive boundary condition, this would be a solid and useful contribution for an extended-abstract venue.

major comments (2)
  1. [Abstract; §4.3.3] The abstract's claim that live coding leads to 'more mental stress than presenting static slides' is not supported by the evidence reported in §4.3.3. That section states that lecturers rated live coding as more mentally demanding than 'normal classes' (with P8 explicitly excepted) and that TAs described it as more mentally demanding than 'conventional lectures'; neither 'normal classes' nor 'conventional lectures' is defined as or shown to be equivalent to 'static slides,' and no participant is quoted directly comparing live coding with static slides on the dimension of stress. No standardized workload instrument (e.g., NASA-TLX) was used, and the causal phrase 'resulting in' goes beyond what retrospective self-reports can establish. Because this comparative sentence is the paper's central finding, it is load-bearing. Please either provide direct static-slides comparison data or reword the claim to something like 'instructors perceive live coding as more mentally demanding and stressful than conventional or slide-based teaching,' with P8's exception discussed explicitly as a deviation rather than glossed over.
  2. [§3.2; Table 1; §5.2] The central comparative claim rests on nine self-selected participants from a single university: four lecturers, all male, and five TAs, one female, as shown in Table 1. Section 5.2 acknowledges the small sample, but the abstract and conclusion nonetheless state the stress finding as a general result about live coding. Moreover, P8, the most experienced lecturer, does not rate live coding as more mentally demanding, which suggests the effect is experience-dependent rather than uniform. For a qualitative study, this does not invalidate the findings, but it does mean the claims should be explicitly scoped to the studied population or framed as hypotheses for a larger, more diverse sample. Treating P8 as a boundary condition rather than a footnote would make the analysis more persuasive.
minor comments (4)
  1. [§5.2] In the Limitations paragraph, 'across different institution' should be 'across different institutions.'
  2. [Table 1] The column group 'H U C' is not explained in the table itself; the definitions appear only in the following line of text. Please add an explicit caption or footnote defining these column abbreviations.
  3. [§4.2.2] There is a stray space in the citation 'Hwang et al . [11]' that should be removed.
  4. [§2.2] The sentence 'it was demonstrated that verbalization increases cognitive load' would be clearer with an explicit subject, e.g., 'Hertzum et al. [16] demonstrated that verbalization increases cognitive load.'

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is an inductive qualitative study with no derivational chain, no fitted parameters, and no load-bearing self-citations.

full rationale

The paper is an inductive qualitative study based on semi-structured interviews and classroom observations, not on a formal derivation. Its central claims about instructor stress and time pressure are grounded in participant quotes and field notes, not in a model fitted to its own conclusion. No equation is defined in terms of the target result, and no parameter is fit and then relabeled as a prediction. The authors' references do not include prior work by Su and Wang; the only design-relevant prior system discussed, Improv, is external work by Chen and Guo. The abstract's phrase 'more mental stress than presenting static slides' generalizes beyond the data, because participants mostly compared live coding to 'normal classes' or 'conventional lectures' and P8 dissented, but that is an evidence-strength or overclaim issue, not circularity. Since there is no derivation chain in which an output equals an input by construction, and no self-citation chain that forces the conclusion, the correct circularity finding is none.

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

This is a qualitative study, so there are no fitted numerical parameters and no invented physical or conceptual entities. The central claim rests instead on assumptions about sample representativeness, self-report validity, observation adequacy, and coding stability.

assumptions (4)
  • domain assumption The nine volunteer instructors from one university are sufficiently representative of programming instructors for the reported stress and time-pressure findings to generalize.
    Recruitment via campus posters for TAs and one university's computer science department for lecturers; Table 1 shows all four lecturers are male. Section 5.2 acknowledges the small sample and limited generalizability.
  • domain assumption Self-reported mental workload and psychological stress are a valid proxy for actual instructor workload during live coding.
    Section 5.2 states reliance on self-reported data and suggests future objective measures such as physiological data. No physiological or performance measures were collected.
  • domain assumption Two observed lectures per lecturer provide a representative sample of each instructor's live coding practice.
    Section 3.2 states each lecturer was observed for two 45-minute lectures. Topic, time of day, and student composition could make any particular pair of lectures unrepresentative.
  • domain assumption The inductive thematic coding of nine participants' accounts is stable and complete without inter-rater reliability checks.
    Section 3.1 describes an inductive approach after Boyatzis [5], but no codebook, inter-rater agreement statistics, or audit trail are reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The Stress of Improvisation: Instructors' Perspectives on Live Coding in Programming Classes." pith.science (2026). https://pith.science/paper/GZHVONNL

@misc{pith2026250603402,
  author       = {Pith},
  title        = {Pith review of: The Stress of Improvisation: Instructors' Perspectives on Live Coding in Programming Classes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GZHVONNL}},
  note         = {Machine review of arXiv:2506.03402}
}
read the original abstract

Live coding is a pedagogical technique in which an instructor writes and executes code in front of students to impart skills like incremental development and debugging. Although live coding offers many benefits, instructors face many challenges in the classroom, like cognitive challenges and psychological stress, most of which have yet to be formally studied. To understand the obstacles faced by instructors in CS classes, we conducted (1) a formative interview with five teaching assistants in exercise sessions and (2) a contextual inquiry study with four lecturers for large-scale classes. We found that the improvisational and unpredictable nature of live coding makes it difficult for instructors to manage their time and keep students engaged, resulting in more mental stress than presenting static slides. We discussed opportunities for augmenting existing IDEs and presentation setups to help enhance live coding experience.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 16 canonical work pages

  1. [26]

    Miller, and Amber Settle

    Andrea Watkins, Craig S. Miller, and Amber Settle. 2024. Comparing the Experi- ences of Live Coding versus Static Code Examples for Students and Instructors. InProceedings of the 2024 on Innovation and Technology in Computer Science Education V. 1(Milan, Italy)(ITiCSE 2024). Association for Computing Machinery, New York, NY, USA, 506–512. https://doi.org/...

    arXiv 2024

  2. [1]

    Abdulaziz Alaboudi and Thomas D. LaToza. 2019. An Exploratory Study of Live- Streamed Programming. In2019 IEEE Symposium on Visual Languages and Human- Centric Computing (VL/HCC). 5–13. https://doi.org/10.1109/VLHCC.2019.8818832

    arXiv 2019

  3. [2]

    Mahnaz Behroozi, Shivani Shirolkar, Titus Barik, and Chris Parnin. 2020. Does stress impact technical interview performance?. InProceedings of the 28th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering(Virtual Event, USA)(ESEC/FSE 2020). Association for Computing Machinery, New York, NY,...

    arXiv 2020

  4. [3]

    Jens Bennedsen and Michael E Caspersen. 2005. Revealing the programming process. InProceedings of the 36th SIGCSE technical symposium on Computer science education. 186–190

    work page 2005

  5. [4]

    2022.Live coding: a user’s manual

    Alan F Blackwell, Emma Cocker, Geoff Cox, Alex McLean, and Thor Magnusson. 2022.Live coding: a user’s manual. MIT Press

    work page 2022

  6. [5]

    1998.Transforming qualitative information: Thematic analysis and code development

    R Boyatzis. 1998.Transforming qualitative information: Thematic analysis and code development. Sage

    work page 1998

  7. [6]

    Neil CC Brown and Greg Wilson. 2018. Ten quick tips for teaching programming. PLoS computational biology14, 4 (2018), e1006023

    work page 2018

  8. [7]

    Chen and Philip J

    Charles H. Chen and Philip J. Guo. 2019. Improv: Teaching Programming at Scale via Live Coding. InProceedings of the Sixth (2019) ACM Conference on Learning @ Scale(Chicago, IL, USA)(L@S ’19). Association for Computing Machinery, New York, NY, USA, Article 9, 10 pages. https://doi.org/10.1145/3330430.3333627

    work page arXiv 2019

Show all 27 references
  1. [8]

    Lasecki, and Tao Dong

    Yan Chen, Walter S. Lasecki, and Tao Dong. 2021. Towards Supporting Program- ming Education at Scale via Live Streaming.Proc. ACM Hum.-Comput. Interact.4, CSCW3, Article 259 (Jan. 2021), 19 pages. https://doi.org/10.1145/3434168

    2021 doi

  2. [9]

    Ellen Corcoran. 1981. Transition Shock: The Beginning Teacher’s Paradox Ellen Corcoran.Journal of teacher education32, 3 (1981), 19–23

    1981

  3. [10]

    K Anders Ericsson and Herbert A Simon. 1980. Verbal reports as data.Psycholog- ical review87, 3 (1980), 215

    1980

  4. [11]

    Derek Hwang, Vardhan Agarwal, Yuzi Lyu, Divyam Rana, Satya Ganesh Susarla, and Adalbert Gerald Soosai Raj. 2021. A Qualitative Analysis of Lecture Videos and Student Feedback on Static Code Examples and Live Coding: A Case Study. InProceedings of the 23rd Australasian Computin...

    2021

  5. [12]

    1994.Social phobia: From shyness to stage fright.Basic Books/Hachette Book Group

    John R Marshall and Suzanne Lipsett. 1994.Social phobia: From shyness to stage fright.Basic Books/Hachette Book Group

    1994

  6. [13]

    George Herbert Mead. 1934. Mind, self, and society from the standpoint of a social behaviorist. (1934)

    1934

  7. [14]

    Shinji Miyake. 2001. Multivariate workload evaluation combining physiological and subjective measures.International Journal of Psychophysiology40, 3 (2001), 233–238. https://doi.org/10.1016/S0167-8760(00)00191-4 Psychophysiology in

    2001 doi

  8. [15]

    Stephen Monsell. 2003. Task switching.Trends in cognitive sciences7, 3 (2003), 134–140

    2003

  9. [16]

    Hansen Morten Hertzum and Hans H.K

    Kristin D. Hansen Morten Hertzum and Hans H.K. Andersen. 2009. Scrutinising usability evaluation: does thinking aloud affect behaviour and mental workload? Behaviour & Information Technology28, 2 (2009), 165–181. https://doi.org/10. 1080/01449290701773842 arXiv:https://doi.org...

    2009 doi

  10. [17]

    Mai Otsuki, Tzu-Yang Wang, and Hideaki Kuzuoka. 2022. Assessment of Instruc- tor’s Capacity in One-to-Many AR Remote Instruction Giving. InProceedings of the 28th ACM Symposium on Virtual Reality Software and Technology(Tsukuba, Japan)(VRST ’22). Association for Computing Mach...

    2022

  11. [18]

    John Paxton. 2002. Live programming as a lecture technique. 18, 2 (dec 2002), 51–56

    2002

  12. [19]

    Adalbert Gerald Soosai Raj, Pan Gu, Eda Zhang, Arokia Xavier Annie R, Jim Williams, Richard Halverson, and Jignesh M. Patel. 2020. Live-coding vs Static Code Examples: Which is better with respect to Student Learning and Cognitive Load?. InProceedings of the Twenty-Second Aust...

    2020

  13. [20]

    Patel, Richard Halverson, and Erica Rosen- feld Halverson

    Adalbert Gerald Soosai Raj, Jignesh M. Patel, Richard Halverson, and Erica Rosen- feld Halverson. 2018. Role of Live-coding in Learning Introductory Programming. InProceedings of the 18th Koli Calling International Conference on Computing Education Research(Koli, Finland)(Koli...

    2018

  14. [21]

    Marc J. Rubin. 2013. The effectiveness of live-coding to teach introductory pro- gramming. InProceeding of the 44th ACM Technical Symposium on Computer Science Education(Denver, Colorado, USA)(SIGCSE ’13). Association for Comput- ing Machinery, New York, NY, USA, 651–656. http...

    2013 doi

  15. [22]

    Susie Scott. 2007. College hats or lecture trousers? Stage fright and performance anxiety in university teachers.Ethnography and Educa- tion2, 2 (2007), 191–207. https://doi.org/10.1080/17457820701350582 arXiv:https://doi.org/10.1080/17457820701350582

    2007 doi

  16. [23]

    Ana Selvaraj, Eda Zhang, Leo Porter, and Adalbert Gerald Soosai Raj. 2021. Live Coding: A Review of the Literature. InProceedings of the 26th ACM Conference on Innovation and Technology in Computer Science Education V. 1(Virtual Event, Germany)(ITiCSE ’21). Association for Com...

    2021

  17. [24]

    Griswold, and Adalbert Gerald Soosai Raj

    Anshul Shah, Emma Hogan, Vardhan Agarwal, John Driscoll, Leo Porter, William G. Griswold, and Adalbert Gerald Soosai Raj. 2023. An Empirical Evaluation of Live Coding in CS1. InProceedings of the 2023 ACM Conference on International Computing Education Research - Volume 1(Chic...

    2023

  18. [25]

    Ben Stephenson. 2019. Coding Demonstration Videos for CS1. InProceedings of the 50th ACM Technical Symposium on Computer Science Education(Minneapolis, MN, USA)(SIGCSE ’19). Association for Computing Machinery, New York, NY, USA, 105–111. https://doi.org/10.1145/3287324.3287445

    2019

  19. [27]

    2021.Engineering psychology and human performance

    Christopher D Wickens, William S Helton, Justin G Hollands, and Simon Banbury. 2021.Engineering psychology and human performance. Routledge

    2021

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.