pith. sign in

arxiv: 2606.00875 · v1 · pith:ZHMJJRSBnew · submitted 2026-05-30 · 💻 cs.CL

IDEAFix: Evaluation Framework for Creative Defixation Prompting in LLMs

F. Carichon , S. Sharma , M. Girard , R. Rampa , G. Farnadi This is my paper

Pith reviewed 2026-06-28 18:40 UTC · model grok-4.3

classification 💻 cs.CL
keywords LLM evaluationcreative idea generationdefixation promptingdivergent thinkingoutput homogenizationdesign scenariosprompt engineeringoriginality measurement
0
0 comments X

The pith

Task formulation and defixation prompts affect LLM originality while persistent homogenization limits diversity.

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

The paper introduces IDEAFix, a framework that tests large language models on generating original solutions to short design scenarios while varying task attributes and using prompts meant to reduce fixation on common ideas. By keeping the scenarios controlled and short, the approach separates the influence of how the task is stated, which attributes are included, and which prompting tactics are applied. Results indicate that these choices do change how original the outputs score, and that straightforward prompting methods raise originality levels. At the same time, models across the board keep producing similar sets of ideas no matter the changes, pointing to built-in constraints on variety. The framework supplies a repeatable way to study what shapes or restricts creative output in these systems.

Core claim

IDEAFix prompts models to generate multiple original solutions to controlled variations of short design scenarios, task attributes, and defixation prompting strategies. This design enables systematic analysis of how structured guidance influences LLMs' idea generation. Results show that both task formulation and attribute selection significantly affect models' performance, and that simple prompting strategies can boost the originality of solutions. However, persistent output homogenization across models confirms inherent limits in their ability to generate diverse solutions.

What carries the argument

The IDEAFix evaluation framework, which applies controlled variations of short design scenarios, task attributes, and defixation prompting strategies to measure effects on divergent thinking and idea originality.

If this is right

  • Task formulation leads to measurable differences in how original the generated ideas are.
  • Choice of attributes in the scenario influences the performance of the models on originality measures.
  • Simple defixation prompting strategies raise the originality of solutions produced.
  • Output homogenization stays consistent across models despite changes in setup or prompts.
  • Inherent limits restrict LLMs from achieving high diversity in creative solutions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework could be extended to longer or multi-step design tasks to test whether homogenization decreases when more context is available.
  • Combining the prompting strategies tested here with model fine-tuning might address the diversity constraint in future systems.
  • Applying IDEAFix to domains such as story or product concept generation would show if the observed limits are specific to design scenarios.

Load-bearing premise

Short design scenarios with controlled variations of task attributes and defixation prompting strategies can isolate the effects of structured guidance on idea generation without confounding factors from broader creative processes.

What would settle it

An experiment in which altering task formulations or defixation prompts produces no measurable change in originality scores, or in which models generate clearly distinct solution sets without homogenization, would contradict the reported effects and limits.

Figures

Figures reproduced from arXiv: 2606.00875 by F. Carichon, G. Farnadi, M. Girard, R. Rampa, S. Sharma.

Figure 1
Figure 1. Figure 1: IDEAFix protocol depicting the 3 steps to create the dataset (Brief Construction, Attributes [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Examples of alternative Control and AI-specific prompting strategies. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Model performance across task and attribute dimensions. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Effect of the best-performing alternative on model performance. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Homogenization patterns across models. 5 Discussion Our results highlight the presence of fixation effects in LLM idea generation with models repeatedly converging toward familiar solution spaces. These fixation effects seem to be similar to the ones observed for humans on products or known tasks and contribute to constraining their ideation space(Wadinambiarachchi et al., 2024). However, introducing defix… view at source ↗
Figure 6
Figure 6. Figure 6: Examples of briefs present in IDEAFix with their associated attributes. [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Examples of alternative Design Thinking, Brainstorming, and AI specific prompting [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗
read the original abstract

Large language models (LLMs) are increasingly used for tasks involving creative problem solving and idea generation. However, there is a lack of consensus concerning their creative capabilities: some studies report superior performances compared to humans, while others highlight structural limitations such as fixation and the homogenization of outputs. Existing evaluation approaches either rely on narrow, decontextualized tasks that do not capture goal-oriented generation or on broader settings that confound multiple aspects of the creative process, making it difficult to isolate the effects of task formulation, prompting, and evaluation design. Significantly, the role of structured prompting strategies in shaping idea generation remains underexplored. Therefore, we introduce IDEAFix, an evaluation framework for analyzing divergent thinking in open-ended idea generation tasks. We prompt models to generate multiple original solutions to controlled variations of short design scenarios, task attributes, and defixation prompting strategies. This design enables systematic analysis of how structured guidance influences LLMs' idea generation. Our results show that both task formulation and attribute selection significantly affect models' performance, and that simple prompting strategies can boost the originality of solutions. However, we also observe persistent output homogenization across models, confirming inherent limits in their ability to generate diverse solutions. Overall, IDEAFix provides a controlled, extensible framework for studying the mechanisms underlying LLMs' creativity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces IDEAFix, an evaluation framework for divergent thinking in LLMs on open-ended idea generation. It prompts models to produce multiple original solutions to controlled short design scenarios while varying task attributes and defixation prompting strategies, enabling analysis of how structured guidance affects idea generation. Results indicate that task formulation and attribute selection significantly influence performance, simple prompting can increase originality, yet persistent output homogenization across models is observed and interpreted as confirming inherent limits on diverse solution generation.

Significance. If the framework's controlled variations successfully isolate prompting effects without confounds, IDEAFix could provide a reproducible method for studying LLM creativity mechanisms. The observation of homogenization under specific conditions is a useful empirical finding, but the interpretation as evidence of model-intrinsic limits would require additional validation across varied task lengths and framings to strengthen the contribution.

major comments (2)
  1. [Abstract] Abstract: the assertion that persistent homogenization 'confirming inherent limits in their ability to generate diverse solutions' does not follow from the reported design of short design scenarios with controlled attribute variations; the results could be artifacts of scenario length or prompting format, and the manuscript provides no evidence that homogenization persists (or decreases) under alternative framings or longer contexts while holding models fixed.
  2. [Abstract] Results (inferred from abstract description of effects on originality and homogenization): the abstract supplies no metrics, sample sizes, statistical tests, or explicit definitions of 'originality' and 'homogenization' measures, so the load-bearing claims that task formulation/attribute selection 'significantly affect' performance and that prompting 'boosts' originality cannot be evaluated for effect size or robustness.
minor comments (1)
  1. [Abstract] The abstract states the framework is 'extensible' but does not specify how new scenarios or attributes would be added without introducing new confounds.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments on the abstract. We address each point below and will revise the manuscript accordingly where the feedback identifies opportunities to better align claims with the reported evidence.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the assertion that persistent homogenization 'confirming inherent limits in their ability to generate diverse solutions' does not follow from the reported design of short design scenarios with controlled attribute variations; the results could be artifacts of scenario length or prompting format, and the manuscript provides no evidence that homogenization persists (or decreases) under alternative framings or longer contexts while holding models fixed.

    Authors: We agree that the abstract phrasing overstates the generality of the finding. The observed homogenization is tied to the short design scenarios used, and alternative task lengths or framings were not tested. In revision we will reword the abstract to report the homogenization result as observed under the specific controlled conditions of the study, while noting that this suggests but does not confirm model-intrinsic limits without further validation across varied framings. revision: yes

  2. Referee: [Abstract] Results (inferred from abstract description of effects on originality and homogenization): the abstract supplies no metrics, sample sizes, statistical tests, or explicit definitions of 'originality' and 'homogenization' measures, so the load-bearing claims that task formulation/attribute selection 'significantly affect' performance and that prompting 'boosts' originality cannot be evaluated for effect size or robustness.

    Authors: The abstract is intentionally concise; full operational definitions (originality via semantic divergence and expert ratings; homogenization via inter-solution similarity), sample sizes, and statistical tests appear in the Methods and Results sections. To improve evaluability we will add brief parenthetical definitions of the two measures to the abstract and note that reported effects reached statistical significance, while retaining detailed metrics in the body text. revision: partial

standing simulated objections not resolved
  • The manuscript contains no experiments with longer contexts or alternative framings, so we cannot supply direct evidence on whether homogenization persists or decreases under those conditions.

Circularity Check

0 steps flagged

No circularity: empirical framework with independent observations

full rationale

The paper introduces an evaluation framework (IDEAFix) for testing LLM idea generation under controlled prompting variations and reports empirical results on originality and homogenization. No equations, derivations, fitted parameters, or self-referential computations appear. Claims rest on experimental outputs rather than reducing to definitional inputs or self-citation chains. The homogenization observation is presented as a measured outcome, not a constructed tautology. This matches the default case of a non-circular descriptive study.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical models, free parameters, axioms, or invented entities are introduced; the work is a conceptual evaluation framework.

pith-pipeline@v0.9.1-grok · 5777 in / 1196 out tokens · 34580 ms · 2026-06-28T18:40:07.701957+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

72 extracted references · 12 canonical work pages · 1 internal anchor

  1. [1]

    The Wharton School Research Paper Forthcoming , volume=

    Using large language models for idea generation in innovation , author=. The Wharton School Research Paper Forthcoming , volume=. 2024 , publisher=

    2024

  2. [2]

    2025 , eprint=

    Qwen3 Technical Report , author=. 2025 , eprint=

    2025

  3. [3]

    2025 , month =

    xAI , title =. 2025 , month =

    2025

  4. [4]

    Augmenting Human Creativity in Brainstorming Sessions with Artificial Intelligence , author=

  5. [5]

    arXiv preprint arXiv:2504.14191 , year=

    Ai idea bench 2025: Ai research idea generation benchmark , author=. arXiv preprint arXiv:2504.14191 , year=

    work page arXiv 2025

  6. [6]

    Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V

    Ideabench: Benchmarking large language models for research idea generation , author=. Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2 , pages=

  7. [7]

    Nature , year=

    Do AI models produce more original ideas than researchers? , author=. Nature , year=

  8. [8]

    Scientific reports , volume=

    The current state of artificial intelligence generative language models is more creative than humans on divergent thinking tasks , author=. Scientific reports , volume=. 2024 , publisher=

    2024

  9. [9]

    Journal of Product Innovation Management , volume=

    Artificial intelligence and corporate ideation systems , author=. Journal of Product Innovation Management , volume=. 2026 , publisher=

    2026

  10. [10]

    AI & society , volume=

    On the creativity of large language models , author=. AI & society , volume=. 2025 , publisher=

    2025

  11. [11]

    Proceedings of the 15th International Conference on Computational Creativity , pages =

    Is temperature the creativity parameter of large language models? , author=. arXiv preprint arXiv:2405.00492 , year=

    work page arXiv

  12. [12]

    arXiv preprint arXiv:2510.22954 , year=

    Artificial hivemind: The open-ended homogeneity of language models (and beyond) , author=. arXiv preprint arXiv:2510.22954 , year=

    work page arXiv

  13. [13]

    Preprint , year=

    Is prompt engineering the creativity knob for large language models , author=. Preprint , year=

  14. [14]

    Base models beat aligned models at randomness and creativity.arXiv preprint arXiv:2505.00047, 2025

    Base models beat aligned models at randomness and creativity , author=. arXiv preprint arXiv:2505.00047 , year=

    work page arXiv

  15. [15]

    Science advances , volume=

    Generative AI enhances individual creativity but reduces the collective diversity of novel content , author=. Science advances , volume=. 2024 , publisher=

    2024

  16. [16]

    Nature Reviews Psychology , volume=

    Using natural language processing to analyse text data in behavioural science , author=. Nature Reviews Psychology , volume=. 2025 , publisher=

    2025

  17. [17]

    arXiv preprint arXiv:2505.17241 , year=

    Generative AI and creativity: A systematic literature review and meta-analysis , author=. arXiv preprint arXiv:2505.17241 , year=

    work page arXiv

  18. [18]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    CreBench: Human-Aligned Creativity Evaluation from Idea to Process to Product , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  19. [19]

    Brainstorming with a generative language model: the role of creative ability and tool-support for brainstorming performance , author=

  20. [20]

    Proceedings of the 16th conference on creativity & cognition , pages=

    Homogenization effects of large language models on human creative ideation , author=. Proceedings of the 16th conference on creativity & cognition , pages=

  21. [21]

    Journal of Creativity , pages=

    Has the creativity of large-language models peaked?: An analysis of inter-and intra-llm variability , author=. Journal of Creativity , pages=. 2025 , publisher=

    2025

  22. [22]

    Humanities and Social Sciences Communications , volume=

    Inspiration booster or creative fixation? The dual mechanisms of LLMs in shaping individual creativity in tasks of different complexity , author=. Humanities and Social Sciences Communications , volume=. 2025 , publisher=

    2025

  23. [23]

    Frontiers in Psychology , volume=

    The paradox of creativity in generative AI: high performance, human-like bias, and limited differential evaluation , author=. Frontiers in Psychology , volume=. 2025 , publisher=

    2025

  24. [24]

    Proceedings of the 2024 CHI conference on human factors in computing systems , pages=

    The effects of generative AI on design fixation and divergent thinking , author=. Proceedings of the 2024 CHI conference on human factors in computing systems , pages=

    2024

  25. [25]

    arXiv preprint arXiv:2410.17218 , year=

    Creativity in ai: Progresses and challenges , author=. arXiv preprint arXiv:2410.17218 , year=

    work page arXiv

  26. [26]

    2004 , publisher=

    The creative mind: Myths and mechanisms , author=. 2004 , publisher=

    2004

  27. [27]

    arXiv preprint arXiv:2603.19066 , year=

    Parallelograms Strike Back: LLMs Generate Better Analogies than People , author=. arXiv preprint arXiv:2603.19066 , year=

    work page arXiv

  28. [28]

    Divergent creativity in humans and large language models.arXiv preprint arXiv:2405.13012, 2024

    Divergent creativity in humans and large language models , author=. arXiv preprint arXiv:2405.13012 , year=

    work page arXiv

  29. [29]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Towards scientific discovery with generative ai: Progress, opportunities, and challenges , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  30. [30]

    Management Science , year=

    The effects of generative AI on high-skilled work: Evidence from three field experiments with software developers , author=. Management Science , year=

  31. [31]

    Scientific reports , volume=

    A large-scale comparison of human-written versus ChatGPT-generated essays , author=. Scientific reports , volume=. 2023 , publisher=

    2023

  32. [32]

    arXiv preprint arXiv:2411.15560 , year=

    Do LLMs Agree on the Creativity Evaluation of Alternative Uses? , author=. arXiv preprint arXiv:2411.15560 , year=

    work page arXiv

  33. [33]

    arXiv preprint arXiv:2504.15784 , year=

    Automated creativity evaluation for large language models: A reference-based approach , author=. arXiv preprint arXiv:2504.15784 , year=

    work page arXiv

  34. [34]

    Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    Lateval: An interactive llms evaluation benchmark with incomplete information from lateral thinking puzzles , author=. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    2024

  35. [35]

    Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) , pages=

    MacGyver: Are Large Language Models Creative Problem Solvers? , author=. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) , pages=

    2024

  36. [36]

    Advances in neural information processing systems , volume=

    Chain-of-thought prompting elicits reasoning in large language models , author=. Advances in neural information processing systems , volume=

  37. [37]

    ACM computing surveys , volume=

    Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing , author=. ACM computing surveys , volume=. 2023 , publisher=

    2023

  38. [38]

    Harvard business review , volume=

    Why design thinking works , author=. Harvard business review , volume=

  39. [39]

    Creativity and innovation management , volume=

    What is TRIZ? From conceptual basics to a framework for research , author=. Creativity and innovation management , volume=. 2005 , publisher=

    2005

  40. [40]

    DS 32: Proceedings of DESIGN 2004, the 8th International Design Conference, Dubrovnik, Croatia , pages=

    CK theory in practice: lessons from industrial applications , author=. DS 32: Proceedings of DESIGN 2004, the 8th International Design Conference, Dubrovnik, Croatia , pages=

    2004

  41. [41]

    Advances in neural information processing systems , volume=

    Language models are few-shot learners , author=. Advances in neural information processing systems , volume=

  42. [42]

    2023 , publisher=

    Pushing gpt’s creativity to its limits: Alternative uses and torrance tests , author=. 2023 , publisher=

    2023

  43. [43]

    Journal of Creativity , volume=

    The originality of machines: AI takes the Torrance Test , author=. Journal of Creativity , volume=. 2023 , publisher=

    2023

  44. [44]

    The Journal of creative behavior , volume=

    Brainstorming, brainstorming rules and decision making , author=. The Journal of creative behavior , volume=. 2009 , publisher=

    2009

  45. [45]

    Journal for the Education of Gifted young scientists , volume=

    The effectiveness of SCAMPER technique on creative thinking skills , author=. Journal for the Education of Gifted young scientists , volume=. 2016 , publisher=

    2016

  46. [46]

    Technological Forecasting and Social Change , volume=

    A brief methodological guide to scenario building , author=. Technological Forecasting and Social Change , volume=. 2000 , publisher=

    2000

  47. [47]

    Handbook of Human-Centered Artificial Intelligence , pages=

    Design Thinking and AI: Facilitating HCAI Solutions , author=. Handbook of Human-Centered Artificial Intelligence , pages=. 2026 , publisher=

    2026

  48. [48]

    AI EDAM , volume=

    Design creativity in AI: Using the SCAMPER method , author=. AI EDAM , volume=. 2025 , publisher=

    2025

  49. [49]

    Advanced Engineering Informatics , volume=

    AutoTRIZ: Automating engineering innovation with TRIZ and large language models , author=. Advanced Engineering Informatics , volume=. 2025 , publisher=

    2025

  50. [50]

    Do-not- answer: A dataset for evaluating safeguards in llms,

    Do-not-answer: A dataset for evaluating safeguards in llms , author=. arXiv preprint arXiv:2308.13387 , year=

    work page arXiv

  51. [51]

    , author=

    Exploiting open-endedness to solve problems through the search for novelty. , author=. ALIFE , volume=

  52. [52]

    Automatic Chain of Thought Prompting in Large Language Models

    Automatic chain of thought prompting in large language models , author=. arXiv preprint arXiv:2210.03493 , year=

    work page internal anchor Pith review Pith/arXiv arXiv

  53. [53]

    Sentence-bert: Sentence embeddings using siamese bert-networks , author=. Proceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP) , pages=

    2019

  54. [54]

    Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

    Confabulation: The surprising value of large language model hallucinations , author=. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  55. [55]

    The Journal of Creative Behavior , volume=

    The impact of type of examples on originality: Explaining fixation and stimulation effects , author=. The Journal of Creative Behavior , volume=. 2014 , publisher=

    2014

  56. [56]

    Journal of Product Innovation Management , volume=

    The creative process of problem framing for innovation: An integrative review and research agenda , author=. Journal of Product Innovation Management , volume=. 2025 , publisher=

    2025

  57. [57]

    , author=

    Assessing creativity with divergent thinking tasks: exploring the reliability and validity of new subjective scoring methods. , author=. Psychology of Aesthetics, Creativity, and the Arts , volume=. 2008 , publisher=

    2008

  58. [58]

    , author=

    The social psychology of creativity: A componential conceptualization. , author=. Journal of personality and social psychology , volume=. 1983 , publisher=

    1983

  59. [59]

    Creativity Research Journal , volume=

    Malevolent creativity: A functional model of creativity in terrorism and crime , author=. Creativity Research Journal , volume=. 2008 , publisher=

    2008

  60. [60]

    The Routledge companion to creativity , pages=

    Computers and creativity: models and applications , author=. The Routledge companion to creativity , pages=. 2008 , publisher=

    2008

  61. [61]

    International Journal of Design Creativity and Innovation , volume=

    Data-intensive evaluation of design creativity using novelty, value, and surprise , author=. International Journal of Design Creativity and Innovation , volume=. 2015 , publisher=

    2015

  62. [62]

    Design Science , volume=

    Uses of the novelty metrics proposed by Shah et al.: what emerges from the literature? , author=. Design Science , volume=. 2023 , publisher=

    2023

  63. [63]

    The Journal of Creative Behavior , volume=

    Creative problem solving in small groups: The effects of creativity training on idea generation, solution creativity, and leadership effectiveness , author=. The Journal of Creative Behavior , volume=. 2020 , publisher=

    2020

  64. [64]

    Journal of Cross-Cultural Psychology , volume=

    Beyond individual creativity: The superadditive benefits of multicultural experience for collective creativity in culturally diverse teams , author=. Journal of Cross-Cultural Psychology , volume=. 2012 , publisher=

    2012

  65. [65]

    Creativity Research Journal , volume=

    Detecting fixation bias in creative idea generation: Evidence from design novices and experts , author=. Creativity Research Journal , volume=. 2026 , publisher=

    2026

  66. [66]

    Thinking & Reasoning , volume=

    Incubation and creativity: Do something different , author=. Thinking & Reasoning , volume=. 2013 , publisher=

    2013

  67. [67]

    The Journal of Creative Behavior , volume=

    The personality composition of teams and creativity: The moderating role of team creative confidence , author=. The Journal of Creative Behavior , volume=. 2008 , publisher=

    2008

  68. [68]

    Information systems research , volume=

    Modifying paradigms—Individual differences, creativity techniques, and exposure to ideas in group idea generation , author=. Information systems research , volume=. 2001 , publisher=

    2001

  69. [69]

    Creativity Research Journal , volume=

    Creativity and ethics: The relationship of creative and ethical problem-solving , author=. Creativity Research Journal , volume=. 2010 , publisher=

    2010

  70. [70]

    Cognitive psychology , volume=

    Structured imagination: The role of category structure in exemplar generation , author=. Cognitive psychology , volume=. 1994 , publisher=

    1994

  71. [71]

    Journal of Research in Personality , volume=

    An examination of the relationship between conscientiousness and group performance on a creative task , author=. Journal of Research in Personality , volume=. 2010 , publisher=

    2010

  72. [72]

    Proceedings of Fifth IAA International Conference on Low-Cost Planetary Missions, 24-26 Septembre 2003, ESA SP-542, Noordwijk, the Netherlands , pages=

    Mars Hopper vs Mars Rover , author=. Proceedings of Fifth IAA International Conference on Low-Cost Planetary Missions, 24-26 Septembre 2003, ESA SP-542, Noordwijk, the Netherlands , pages=

    2003