ai-research-survey

Systematic scan of agentic development research. What's signal, what's noise.
git clone https://git.shiptheloop.com/ai-research-survey.git
Log | Files | Refs

scan-v5.json (19633B)


      1 {
      2   "scan_version": 5,
      3   "paper_type": "position",
      4   "paper": {
      5     "title": "Towards Extending the Range of Bugs That Automated Program Repair Can Handle",
      6     "authors": [
      7       "Omar I. Al-Bataineh",
      8       "L. Moonen"
      9     ],
     10     "year": 2022,
     11     "venue": "International Conference on Software Quality, Reliability and Security",
     12     "arxiv_id": "2211.03911",
     13     "doi": "10.1109/QRS57517.2022.00031"
     14   },
     15   "checklist": {
     16     "claims_and_evidence": {
     17       "abstract_claims_supported": {
     18         "applies": true,
     19         "answer": false,
     20         "justification": "The abstract states 'the study shows that integrating dynamic APR with formal analysis techniques reduces complexity and improves reliability,' but the paper only sketches algorithms without implementing or evaluating the hybrid system. The 85% figure refers to existing termination provers on standard benchmarks, not the proposed approach.",
     21         "source": "haiku"
     22       },
     23       "causal_claims_justified": {
     24         "applies": true,
     25         "answer": false,
     26         "justification": "The paper makes causal claims that combining termination provers with APR 'reduces complexity and improves reliability,' but these are theoretical arguments — the proposed algorithms are explicitly unimplemented ('we are in the process of empirically validating the ideas').",
     27         "source": "haiku"
     28       },
     29       "generalization_bounded": {
     30         "applies": true,
     31         "answer": false,
     32         "justification": "The conclusion states the hybrid approach 'reduces complexity and improves overall reliability' without bounding these claims to specific bug types, program sizes, or tool configurations; the 85% success rate from existing tools is generalized to support the proposed unimplemented system.",
     33         "source": "haiku"
     34       },
     35       "alternative_explanations_discussed": {
     36         "applies": true,
     37         "answer": false,
     38         "justification": "The paper does not consider alternative explanations for why APR fails on non-observable bugs, nor alternatives to formal methods (e.g., LLM-based repair, learned specifications) as competing approaches to the identified gap.",
     39         "source": "haiku"
     40       },
     41       "proxy_outcome_distinction": {
     42         "applies": true,
     43         "answer": false,
     44         "justification": "The paper uses termination prover success on existing benchmarks (85%) as evidence for the proposed hybrid APR system's feasibility, without distinguishing that this measures existing tool capability on standard programs, not the proposed integrated repair pipeline.",
     45         "source": "haiku"
     46       }
     47     },
     48     "limitations_and_scope": {
     49       "limitations_section_present": {
     50         "applies": true,
     51         "answer": false,
     52         "justification": "There is no dedicated limitations or threats-to-validity section. Limitations are briefly acknowledged in the future work section ('we are in the process of empirically validating'), which does not constitute a formal limitations discussion.",
     53         "source": "haiku"
     54       },
     55       "threats_to_validity_specific": {
     56         "applies": true,
     57         "answer": false,
     58         "justification": "No threats to validity are identified. While undecidability and state explosion are mentioned as technical challenges, they are not framed as threats to the validity of the paper's claims.",
     59         "source": "haiku"
     60       },
     61       "scope_boundaries_stated": {
     62         "applies": true,
     63         "answer": false,
     64         "justification": "The paper does not explicitly state what classes of programs or bugs are out of scope for the proposed hybrid approach; the acknowledgment that termination is undecidable does not bound the claims made in the abstract and conclusion.",
     65         "source": "haiku"
     66       }
     67     },
     68     "conflicts_of_interest": {
     69       "funding_disclosed": {
     70         "applies": true,
     71         "answer": true,
     72         "justification": "Funding is clearly disclosed: 'This work has been financially supported by the Research Council of Norway through the secureIT project (RCN contract #288787).'",
     73         "source": "haiku"
     74       },
     75       "affiliations_disclosed": {
     76         "applies": true,
     77         "answer": true,
     78         "justification": "Both authors clearly identify their affiliation as Simula Research Laboratory, Oslo, Norway.",
     79         "source": "haiku"
     80       },
     81       "funder_independent_of_outcome": {
     82         "applies": true,
     83         "answer": true,
     84         "justification": "The Research Council of Norway is a government funding body with no financial interest in the specific tools (AProVE, 2LS, T2) evaluated in this paper.",
     85         "source": "haiku"
     86       },
     87       "financial_interests_declared": {
     88         "applies": true,
     89         "answer": false,
     90         "justification": "No competing interests statement or declaration of financial interests (patents, equity, consulting) is provided beyond the funding disclosure.",
     91         "source": "haiku"
     92       }
     93     },
     94     "scope_and_framing": {
     95       "key_terms_defined": {
     96         "applies": true,
     97         "answer": true,
     98         "justification": "The paper provides formal definitions (Definitions 1–15) for all key terms including 'program bug,' 'observable bug,' 'hang bugs,' 'bug tractability,' and 'valid repair,' grounding the framework with mathematical precision.",
     99         "source": "haiku"
    100       },
    101       "intended_contribution_clear": {
    102         "applies": true,
    103         "answer": true,
    104         "justification": "The paper explicitly enumerates three contributions: (1) a novel bug classification system, (2) four APR approaches mapped to bug classes, and (3) hybrid APR algorithms for termination bugs.",
    105         "source": "haiku"
    106       },
    107       "engagement_with_prior_work": {
    108         "applies": true,
    109         "answer": true,
    110         "justification": "The related work section substantively engages with prior bug classification schemes (Tan et al., Cotroneo et al.), existing APR tools (GenProg, SemFix, Angelix), and termination analysis techniques, explaining how the proposed classification differs from prior schemes.",
    111         "source": "haiku"
    112       }
    113     }
    114   },
    115   "type_checklist": {
    116     "position": {
    117       "argument_quality": {
    118         "argument_internally_consistent": {
    119           "applies": true,
    120           "answer": true,
    121           "justification": "The argument flows coherently: the classification properties (observability, reproducibility, tractability) directly inform which detection techniques apply to each bug class, and the hybrid algorithms follow from this analysis.",
    122           "source": "haiku"
    123         },
    124         "counterarguments_addressed": {
    125           "applies": true,
    126           "answer": false,
    127           "justification": "The paper does not engage with the best opposing view — e.g., that the practical complexity of integrating formal methods may outweigh benefits, or that emerging LLM-based repair could address non-observable bugs without formal specifications. Only individual technique limitations are noted.",
    128           "source": "haiku"
    129         },
    130         "analogies_appropriate": {
    131           "applies": false,
    132           "answer": false,
    133           "justification": "The paper does not use analogies; it relies on formal definitions and technical comparisons.",
    134           "source": "haiku"
    135         },
    136         "prescriptions_proportional": {
    137           "applies": true,
    138           "answer": true,
    139           "justification": "Prescriptions are narrowly scoped to future research directions within APR (integrating termination provers, developing fault localization for liveness bugs), proportional to the theoretical framework presented.",
    140           "source": "haiku"
    141         },
    142         "evidence_for_claims_cited": {
    143           "applies": true,
    144           "answer": true,
    145           "justification": "Factual claims are backed by citations — the 85% termination success rate references SNU and PowerStone benchmarks, and all tool capabilities are tied to specific published papers.",
    146           "source": "haiku"
    147         },
    148         "alternatives_discussed": {
    149           "applies": true,
    150           "answer": true,
    151           "justification": "The paper systematically compares four APR approaches (dynamic, static, dynamic-static, formal) and three bug detection techniques with explicit comparison tables discussing relative strengths and weaknesses.",
    152           "source": "haiku"
    153         },
    154         "historical_context_accurate": {
    155           "applies": true,
    156           "answer": true,
    157           "justification": "Historical references appear accurate — the paper correctly traces APR from GenProg through semantic approaches and references foundational work on liveness (Alpern & Schneider 1987) and temporal logic (Pnueli 1977).",
    158           "source": "haiku"
    159         }
    160       },
    161       "clarity_and_scope": {
    162         "key_terms_defined_precisely": {
    163           "applies": true,
    164           "answer": true,
    165           "justification": "Key terms are defined with formal mathematical precision — Definitions 1–15 cover all major concepts (bugs, observability, tractability, hang bugs, valid repair), going well beyond casual usage.",
    166           "source": "haiku"
    167         },
    168         "engages_with_existing_literature": {
    169           "applies": true,
    170           "answer": true,
    171           "justification": "The related work section substantively engages with prior bug classification systems, APR tools, and termination analysis techniques, explaining the limitations of each relative to the proposed framework.",
    172           "source": "haiku"
    173         },
    174         "intended_audience_clear": {
    175           "applies": true,
    176           "answer": true,
    177           "justification": "The paper is clearly written for the APR research community, evident from the technical formalism, venue (QRS 2022), and explicit framing as stimulating 'systematic study' within 'the community.'",
    178           "source": "haiku"
    179         },
    180         "assumptions_stated": {
    181           "applies": true,
    182           "answer": false,
    183           "justification": "Key assumptions are not explicitly stated — e.g., that formal specifications exist for bugs of interest, that model abstractions can be constructed within feasible bounds, or that termination provers can be practically integrated into existing APR pipelines.",
    184           "source": "haiku"
    185         },
    186         "scope_of_applicability_discussed": {
    187           "applies": true,
    188           "answer": false,
    189           "justification": "While undecidability and state explosion are acknowledged, the paper does not systematically discuss where the proposed approach would fail or what types of programs are out of scope (e.g., programs without available formal specifications).",
    190           "source": "haiku"
    191         }
    192       }
    193     }
    194   },
    195   "claims": [
    196     {
    197       "claim": "A significant class of bugs (liveness, non-functional, information flow) cannot be handled by current APR approaches that rely on dynamic analysis.",
    198       "evidence": "Argued from first principles: dynamic analysis requires observable erroneous behavior in finite execution steps, which these bug classes do not produce. Supported by prior literature on liveness bugs.",
    199       "supported": "moderate"
    200     },
    201     {
    202       "claim": "The proposed three-property classification (observability, reproducibility, tractability) enables methodical analysis of which APR approaches can handle which bug types.",
    203       "evidence": "Demonstrated analytically for arithmetic, non-functional, and liveness bugs using the classification properties, but no empirical validation of the framework's utility is provided.",
    204       "supported": "weak"
    205     },
    206     {
    207       "claim": "Integrating dynamic APR with termination provers reduces complexity and improves reliability for sequential termination bugs.",
    208       "evidence": "Argued theoretically through hybrid algorithm sketches and avoidance of patch overfitting; no implementation or empirical evaluation is provided.",
    209       "supported": "weak"
    210     },
    211     {
    212       "claim": "Existing termination provers (AProVE, 2LS) successfully prove termination of approximately 85% of programs in the SNU and PowerStone benchmarks.",
    213       "evidence": "Attributed to application of these tools on benchmarks 'using very little computational time,' but no citation to a published evaluation report for this specific figure is given.",
    214       "supported": "weak"
    215     },
    216     {
    217       "claim": "Formal APR combining termination provers and software model checkers eliminates the patch overfitting problem for termination bugs.",
    218       "evidence": "Derived from formal correctness specification (Formula 2), which is theoretically sound but not empirically validated in the paper.",
    219       "supported": "weak"
    220     }
    221   ],
    222   "methodology_tags": [
    223     "theoretical",
    224     "position"
    225   ],
    226   "key_findings": "The paper proposes a three-property bug classification system (observability, reproducibility, tractability) to enable systematic comparison of APR techniques across bug types that current dynamic APR cannot handle. It maps four APR approaches (dynamic, static, dynamic-static, formal) to bug classes and sketches hybrid algorithms combining termination provers with software model checkers for sequential and concurrent termination bugs. Existing termination provers reportedly succeed on ~85% of standard benchmark programs, offered as evidence of feasibility for the validation component. The hybrid approach theoretically avoids patch overfitting by replacing test-based validation with formal verification, though implementation and empirical evaluation are deferred to future work.",
    227   "red_flags": [
    228     {
    229       "flag": "Unimplemented proposal presented as demonstrated",
    230       "detail": "The abstract claims 'the study shows' the hybrid approach 'reduces complexity and improves overall reliability,' but Section VII explicitly states 'we are in the process of empirically validating the ideas described in this work.' The algorithms are sketches, not implementations."
    231     },
    232     {
    233       "flag": "85% success rate misattributed to proposed system",
    234       "detail": "The 85% termination prover success rate is for existing standalone tools (AProVE, 2LS) on standard benchmarks, not for the proposed hybrid APR system. It is used to argue feasibility of the proposed unbuilt pipeline."
    235     },
    236     {
    237       "flag": "No limitations section",
    238       "detail": "The paper lacks a dedicated limitations or threats-to-validity section. Practical barriers — scalability of model checking, specification availability, tool integration complexity — are not discussed as limitations of the proposed approach."
    239     }
    240   ],
    241   "cited_papers": [
    242     {
    243       "title": "Automated Program Repair (Le Goues, Pradel, Roychoudhury, 2019)",
    244       "relevance": "Survey of APR field providing foundational context for the paper's positioning of hybrid approaches."
    245     },
    246     {
    247       "title": "GenProg: A Generic Method for Automatic Software Repair",
    248       "relevance": "Primary example of dynamic APR against which the proposed hybrid approach is contrasted; illustrates patch overfitting problem."
    249     },
    250     {
    251       "title": "The ManyBugs and IntroClass Benchmarks for Automated Repair of C Programs",
    252       "relevance": "Key APR benchmark dataset referenced for evaluating existing dynamic APR; illustrates timeout-based handling of termination bugs."
    253     },
    254     {
    255       "title": "Proving Termination of Programs Automatically with AProVE",
    256       "relevance": "Core tool proposed for termination validation in the sequential hybrid APR approach."
    257     },
    258     {
    259       "title": "T2: Temporal Property Verification",
    260       "relevance": "Termination prover proposed for patch validation in concurrent programs, with concurrent extension by Cook et al."
    261     },
    262     {
    263       "title": "SemFix: Program Repair via Semantic Analysis",
    264       "relevance": "Representative semantic-based APR approach using symbolic execution, contrasted with proposed formal verification approach."
    265     },
    266     {
    267       "title": "Recognizing Safety and Liveness (Alpern & Schneider, 1987)",
    268       "relevance": "Foundational paper defining liveness properties, underpinning the paper's analysis of liveness bugs as an APR challenge."
    269     },
    270     {
    271       "title": "Towards More Reliable Automated Program Repair by Integrating Static Analysis Techniques (Al-Bataineh et al., 2021)",
    272       "relevance": "Prior work by the same authors on integrating static analysis into APR, directly extended by this paper to formal methods and non-observable bugs."
    273     }
    274   ],
    275   "engagement_factors": {
    276     "practical_relevance": {
    277       "score": 2,
    278       "justification": "Identifies a real and significant gap in APR (non-observable bugs affect safety-critical real systems) but offers only sketched algorithms not yet usable by practitioners."
    279     },
    280     "surprise_contrarian": {
    281       "score": 1,
    282       "justification": "The observation that most APR focuses only on observable bugs is a known gap in the field; the classification framework is a novel organizing contribution but not a counterintuitive finding."
    283     },
    284     "fear_safety": {
    285       "score": 1,
    286       "justification": "Briefly identifies security-related information flow vulnerabilities as an important non-observable bug class and mentions safety-critical systems, but this is not a central focus."
    287     },
    288     "drama_conflict": {
    289       "score": 0,
    290       "justification": "No controversy or conflict angle; purely a constructive research vision paper proposing a new direction."
    291     },
    292     "demo_ability": {
    293       "score": 0,
    294       "justification": "The hybrid algorithms are sketched but not implemented; there is nothing for a reader to run or demonstrate."
    295     },
    296     "brand_recognition": {
    297       "score": 0,
    298       "justification": "Simula Research Laboratory is a respected institution but not a brand-name AI or software lab; no famous products associated."
    299     }
    300   },
    301   "hn_data": {
    302     "threads": [
    303       {
    304         "hn_id": "33154040",
    305         "title": "Evaluating K-NN in the Classification of Data Streams with Concept Drift",
    306         "points": 3,
    307         "comments": 0,
    308         "url": "https://news.ycombinator.com/item?id=33154040"
    309       },
    310       {
    311         "hn_id": "29157895",
    312         "title": "Robust Deep Reinforcement Learning for Quadcopter Control",
    313         "points": 3,
    314         "comments": 0,
    315         "url": "https://news.ycombinator.com/item?id=29157895"
    316       },
    317       {
    318         "hn_id": "35417390",
    319         "title": "Real-time quantum error correction beyond break-even",
    320         "points": 2,
    321         "comments": 1,
    322         "url": "https://news.ycombinator.com/item?id=35417390"
    323       },
    324       {
    325         "hn_id": "29906315",
    326         "title": "Automated Reinforcement Learning (AutoRL): A Survey and Open Problems",
    327         "points": 2,
    328         "comments": 0,
    329         "url": "https://news.ycombinator.com/item?id=29906315"
    330       },
    331       {
    332         "hn_id": "29123008",
    333         "title": "Solving the sampling problem of the Sycamore quantum supremacy circuits",
    334         "points": 2,
    335         "comments": 0,
    336         "url": "https://news.ycombinator.com/item?id=29123008"
    337       },
    338       {
    339         "hn_id": "39894027",
    340         "title": "Instruction-Following Evaluation for Large Language Models",
    341         "points": 1,
    342         "comments": 0,
    343         "url": "https://news.ycombinator.com/item?id=39894027"
    344       },
    345       {
    346         "hn_id": "35055918",
    347         "title": "A multi-segment soft growing robot with selective steering",
    348         "points": 1,
    349         "comments": 0,
    350         "url": "https://news.ycombinator.com/item?id=35055918"
    351       }
    352     ],
    353     "top_points": 3,
    354     "total_points": 14,
    355     "total_comments": 1
    356   }
    357 }

Impressum · Datenschutz