[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84602-en":3,"doc-seo-84602-105":29,"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":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84602,16904993612988,"Olivia Brown","https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd",8,"Research & Report","Calibrating the Instrument: Controllability of an LLM-Driven Synthetic Population","Generative Synthetic Populations (GSP)—combining statistically grounded population synthesis, agent-based modelling, and LLM-driven agents—are used for urban simulation and institutional communication research, yet they require a prior validation step. The paper defines controllability: whether a synthetic population responds to stimuli with ordered, replicable, and group-structured behavior. Using SIVE with a fictional municipality and 120 synthetic personas, seven pre-registered criteria evaluate fidelity, stability, noise, specificity, sensitivity, and ordering. All criteria pass across temperatures.","arXiv :2607 .009 10v 1 [ cs .MA] 1 Jul 2026  \nCalibrating the Instrument: Controllability of an LLM-Driven  \nSynthetic Population  \nMirko Degli Esposti  \nDepartment of Physics and Astronomy, University of Bologna  \n[mirko.degliesposti@unibo.it](mirko.degliesposti@unibo.it)  \nJuly 2, 2026  \nAbstract  \nGenerative Synthetic Populations (GSP)—the convergence of statistically grounded population synthesis, agent-based modelling, and large language model (LLM) agents—are attracting growing interest as tools for urban simulation and institutional communication research. Before any GSP instrument is used to study a real population, however, a more basic question must be answered: does it respond to stimuli of known valence in an ordered, replicable, and group-structured way? We call this property controllability. We ask not whether a synthetic population tracks humans, but whether it tracks itself —whether the latent structure we impose on it is recovered in its own responses. This internal-validity question is logically prior to any claim about external validity, in the same sense that characterising an instrument’s response function must precede using it to test a theory.  \nWe report SIVE (Synthetic Instrument Validation Experiment): a fictional municipality (Montelago) and a population of 120 synthetic personas with known latent attitudinal structure are exposed to seven experimental conditions spanning strongly positive to strongly negative institutional communications about a municipal water network. Seven pre-registered criteria, evaluated across a temperature sweep (t ∈ {0 .2 , 0.5 , 0. 7}), jointly assess fidelity, stability, noise floor, specificity, sensitivity, and ordering. All seven criteria pass at every temperature.  \nA central finding emerges from a stimulus calibration failure turned diagnostic success: a message designed to be “weakly positive”was identified by the instrument as functionally negative, traced to unresolved problems, uncertainty, and institutional passivity in its text; aredesigned version restored the expected ordering, and its effect interacts with agents’ latent trust in ways not anticipated at design time. A single-profile noise sub-experiment further shows that the instrument’s intrinsic measurement noise is roughly half the cross-agent estimate and stable across temperatures — the instrument is more precise than a naive cross-agent analysis suggests.  \nBeyond the aggregate criteria, individual agent trajectories reveal psychologically coherent micro-dynamics that summary statistics obscure: agents with near-identical trust deltas under the same negative stimulus arrive there by qualitatively different routes. We make the full population and reaction data available via an interactive explorer ([https:](https:)//[mirko-degli-esposti.github.io/montelago-explorer/](mirko-degli-esposti.github.io/montelago-explorer/)).  \n1 Introduction: populations as prepared states  \nCities are among the most instrumented complex systems we have, and among the least controllable. For a mathematical physicist drawn to urban questions, the attraction and the frustration are the same: the dynamics of interest—how information, trust, and behavioural intentions propagate through a heterogeneous population embedded in space — unfold in systems that admit no true replicates, no controlled perturbations, and no resets. Agent-based modelling (ABM) has long served as the computational microscope for such systems, but every microscope needs a preparation: an initial condition. For social simulation, the initial condition is a population.  \nThe question of how to prepare a population that is faithful to what we know, and maximally noncommittal about what we do not, is older than the current wave of artificial intelligence. It is, in Jaynes’s sense, a maximum-entropy question [Jaynes, 1957] . Recent work has shown that population synthesis under moment constraints from official statistics can be carried out at scale with the mach","cbCaiklsaHpIkAYO","https://ap.wps.com/l/cbCaiklsaHpIkAYO","pdf",652450,1,25,"English","en",105,"# Introduction: populations as prepared states\n## Generative Synthetic Populations and controllability\n## SIVE setup and validation criteria\n## Findings from stimulus calibration and noise experiments\n## Micro-dynamics and data availability","[{\"question\":\"What does “controllability” mean for a synthetic population instrument in this work?\",\"answer\":\"Controllability means the synthetic population responds to stimuli with an ordered, replicable, and group-structured pattern, recovering the latent structure the instrument imposes on its own responses.\"},{\"question\":\"How is SIVE (Synthetic Instrument Validation Experiment) designed?\",\"answer\":\"SIVE uses a fictional municipality, Montelago, and 120 synthetic personas with known latent attitudinal structure. The population is exposed to seven conditions ranging from strongly positive to strongly negative institutional communications about a municipal water network, then evaluated with seven pre-registered criteria.\"},{\"question\":\"What diagnostic issue emerged during stimulus calibration, and how was it resolved?\",\"answer\":\"A message intended to be “weakly positive” was identified as functionally negative. Redesigning the message restored the expected ordering, and its effect was found to interact with agents’ latent trust in ways not anticipated during design.\"}]",1784197034,63,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"calibrating-the-instrument-controllability-of-an-llm-driven-synthetic-population","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/calibrating-the-instrument-controllability-of-an-llm-driven-synthetic-population/84602/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","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 does “controllability” mean for a synthetic population instrument in this work?","Question",{"text":75,"@type":76},"Controllability means the synthetic population responds to stimuli with an ordered, replicable, and group-structured pattern, recovering the latent structure the instrument imposes on its own responses.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is SIVE (Synthetic Instrument Validation Experiment) designed?",{"text":80,"@type":76},"SIVE uses a fictional municipality, Montelago, and 120 synthetic personas with known latent attitudinal structure. The population is exposed to seven conditions ranging from strongly positive to strongly negative institutional communications about a municipal water network, then evaluated with seven pre-registered criteria.",{"name":82,"@type":73,"acceptedAnswer":83},"What diagnostic issue emerged during stimulus calibration, and how was it resolved?",{"text":84,"@type":76},"A message intended to be “weakly positive” was identified as functionally negative. Redesigning the message restored the expected ordering, and its effect was found to interact with agents’ latent trust in ways not anticipated during design.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]