[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-81781-en":3,"doc-seo-81781-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},81781,549758252649,"Ivy","https://ap-avatar.wpscdn.com/avatar/8000253669c5317157?_k=1778319167496531819",8,"Research & Report","When to Repair a Graph ANN Index","Graph approximate-nearest-neighbor (ANN) indexes such as HNSW and DiskANN/Vamana can lose recall under insert/delete churn: deletions orphan greedy-search routes that pass through removed nodes. Systems restore navigability by repairing edges on a fixed schedule, consolidating every X operations. This work tests whether triggering local edge repair from a measured navigability-degradation signal beats fixed-cadence repair at matched repair budget, on two real datasets and across settings, and finds no tail-recall advantage.","When to Repair a Graph ANN Index:  \nA Matched-Budget Negative Result, and the Interpolated-Baseline  \nTrap That Hid It  \nMadhulatha Mandarapu∗ Sandeep Kunkunuru†  \nVaidhyaMegha Private Limited, India  \n[https://samyama. ai/](https://samyama. ai/)  \nJuly 2026 (v2)  \narXiv :2607 .00728v2 [ cs .DB] 10 Jul 2026  \nChanges in v2—this version corrects v1 . Version 1 of this paper claimed that navigabilitysignal-triggered repair beats fixed-cadence repair at matched budget on tail recall, and that the probe signal is a leading indicator of recall drop. Both claims are withdrawn. v1 never ran a fixed-cadence policy at the budget where it reported its headline; it interpolated one, across a response curve that is steeply concave. Running the missing baseline erases the effect. v1 also reported a signal-validity correlation computed on recall levels rather than on the recall drops its own pre-registration specified; the pre-registered statistic falsifies the claim. Finally, v1 stated that four pre-registered negative controls “all pass”; none of the four had been implemented at the time of publication. All four have now been written and run: three pass, and the one guarding the headline (budget parity) fails. v1’s scale result is likewise withdrawn and replaced by a four-seed matched-budget measurement at a 100k live set, where every threshold’s delta is negative. What survives—the harness, the exact live-set oracle, the drift-severity regime map, and a corrected budget-parity protocol—is reported here, together with the methodological trap that produced the error.  \nAbstract  \nGraph approximate-nearest-neighbor (ANN) indexes—HNSW, DiskANN/Vamana—lose recall under insert/delete churn: deletions orphan the greedy-search paths that route through removed nodes. Production systems restore navigability by repairing the graph on a fixed schedule (consolidate every X operations) . We asked whether triggering local edge repair on a measured navigability-degradation signal, rather than a blind clock, spends a fixed repair budget better. At matched repair budget, it does not. On two real ANN datasets (SIFT-128 and FashionMNIST-784) under bursty churn, compared against a fixed-cadence baseline actually run atthe triggered policy’s realized consolidation count, the tail-recall advantage is indistinguishable from zero at every operating point, graph degree, and index scale; at several points the clock is better. We trace our earlier positive result to an interpolated baseline: recall is sharply concave in repair budget—one consolidation captures over half of all achievable gain—so reading the baseline off a straight line between zero and four passes understates it by more than the effect claimed. Evaluated by the statistic we pre-registered—correlation with the subsequent recall drop rather than the concurrent recall level—the probe signal is not a leading indicator. What remains: an exact live-set recall oracle, a reproducible churn harness, a drift-severity regime map, and a budget-parity protocol that makes this error detectable. We report the negative  \n∗ [madhulatha@samyama.ai](madhulatha@samyama.ai)  \n†[sandeep@samyama.ai](sandeep@samyama.ai)  \nresult and the trap, because the trap generalizes: any “at matched budget X” comparison whose baseline is read off an interpolated curve, at the scarce end of a concave response, manufacturesan effect favouring the proposal.  \n1 Introduction  \nVector search increasingly runs over churning corpora: embeddings are continuously inserted (new items) and deleted (expired or removed items) . Graph ANN indexes [Malkov and Yashunin, 2020 , Subramanya et al. , 2019] answer a k-nearest-neighbor query by a greedy best-first walk over a navigable proximity graph. A deletion removes a node but leaves the greedy-search paths that routed through it dangling; recall then falls. Restoring monotone reachability requires local edge repair over the affected neighborhood (re-linking the deleted node’s in-neighbors to its out","cbCaif6UNgrcVB4Q","https://ap.wps.com/l/cbCaif6UNgrcVB4Q","pdf",480854,4,1,10,"English","en",105,"# Introduction\n## Repair timing for graph-based ANN\n# Contributions of this version\n## Negative result at matched consolidation budget\n## Interpolated-baseline trap\n## Falsified mechanism using pre-registered statistics\n## Negative controls executed","[{\"question\":\"What problem does the paper address in graph-based ANN indexes?\",\"answer\":\"Graph ANN indexes lose recall under insert/delete churn because deletions orphan greedy-search paths that previously routed through removed nodes.\"},{\"question\":\"What is the main question the authors test?\",\"answer\":\"Whether triggering local edge repair using a measured navigability signal spends a fixed repair budget more effectively than using a fixed repair cadence.\"},{\"question\":\"What conclusion does this v2 version reach compared with v1?\",\"answer\":\"The earlier positive claims are withdrawn; with corrected baselines and protocols, signal-triggered repair does not outperform fixed-cadence repair at matched consolidation budget on tail recall, and the probe signal is not supported by the pre-registered statistic.\"}]",1784176110,25,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"when-to-repair-a-graph-ann-index","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":20},"https://docshare.wps.com/document/when-to-repair-a-graph-ann-index/81781/",{"url":52,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-24","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does the paper address in graph-based ANN indexes?","Question",{"text":75,"@type":76},"Graph ANN indexes lose recall under insert/delete churn because deletions orphan greedy-search paths that previously routed through removed nodes.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What is the main question the authors test?",{"text":80,"@type":76},"Whether triggering local edge repair using a measured navigability signal spends a fixed repair budget more effectively than using a fixed repair cadence.",{"name":82,"@type":73,"acceptedAnswer":83},"What conclusion does this v2 version reach compared with v1?",{"text":84,"@type":76},"The earlier positive claims are withdrawn; with corrected baselines and protocols, signal-triggered repair does not outperform fixed-cadence repair at matched consolidation budget on tail recall, and the probe signal is not supported by the pre-registered statistic.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,134],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":22,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":22,"slug":133},"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]