# Research Source: https://dinand.com/research/ Author: Dinand Tinholt > Simulations I run when I want to know if an idea about AI survives contact with real numbers. Ideas across time · Research # Research Simulations I run when I want to know if an idea about AI survives contact with real numbers. Pre-registered campaigns ## The questions I put to the test Each campaign writes down its predictions before a single unit runs, then reports what held, what failed and what the result does not claim. 10 campaigns so far. Paper · Oct 16, 2026 ### [When watching every hour starts to pay](https://dinand.com/research/when-watching-every-hour-starts-to-pay/) A pre-registered simulation held hourly monitoring of an agent fleet and six-hourly inspection to the same nominal reviewer budget, and found the budget at which hourly monitoring starts to catch faults. [Read the paper ](https://dinand.com/research/when-watching-every-hour-starts-to-pay/) Annual report · 2026 ### [The Measured Enterprise 2026](https://dinand.com/research/measured-enterprise-2026/) The lab’s annual report: the Delegation Index, the protocol, seven pre-registered campaigns, the corrections logged and every measured claim, in one document. [Read the report ](https://dinand.com/research/measured-enterprise-2026/) Three-part publication · September 2026 ### [The Decision Head](https://dinand.com/research/decision-head/) A self-hosted service that answers a workflow's typed questions with a calibrated probability. The executive briefing, the full paper and the blog post that started it. [Open the publication ](https://dinand.com/research/decision-head/) Film · October 2026 ### [Who Checks the Agents](https://dinand.com/research/who-checks-the-agents/) The lab’s work on AI delegation and oversight in one film of just under nine minutes, with chapters, a full transcript and every number on screen listed with its source. [Watch the film ](https://dinand.com/research/who-checks-the-agents/) Oct 7, 2026Mixed ### [The open decision head](https://dinand.com/research/campaigns/decision-head/) A retrained open decision head scored 0.728 against 0.566 for a rule table on 800 fresh cases, a gap that excludes zero; the first round missed, with 0.545 against 0.575 (gap -0.029, interval +0.001 to -0.060). [Read the campaign ](https://dinand.com/research/campaigns/decision-head/) Oct 1, 2026Supported ### [A Trial Balance for Agent Omissions](https://dinand.com/research/campaigns/trial-balance/) Reading where a commitment ledger fails to balance found 0.80 of skipped steps at 745 tokens of reading; log review never reached 0.80 at any budget. [Read the campaign ](https://dinand.com/research/campaigns/trial-balance/) Sep 17, 2026Mixed ### [Sprawl Cost Under a Fixed Monitoring Budget](https://dinand.com/research/campaigns/agenesis-2/) With the monitoring budget held flat, sprawl cost grows as fleet size to the power 1.4665 (95% interval 1.4101 to 1.5230); with per-agent monitoring it is 0.9001. [Read the campaign ](https://dinand.com/research/campaigns/agenesis-2/) Sep 14, 2026Measured ### [Herding: do coupled copies of one model start failing together?](https://dinand.com/research/campaigns/herding/) Coupling a fleet adds at most 0.013 to error correlation, against +0.447 from shared weights; the rise in errors came from longer prompts. [Read the campaign ](https://dinand.com/research/campaigns/herding/) Sep 11, 2026Measured ### [Instructie: which clauses of a delegation charter change what an agent does?](https://dinand.com/research/campaigns/instructie/) Of six charter clauses, only boundaries, decision rules and escalation conditions changed agent behaviour; mandate, reporting and revocation had no measurable effect. [Read the campaign ](https://dinand.com/research/campaigns/instructie/) Sep 10, 2026Refuted ### [SIGIL: does a liability ledger change what an overseer authorizes?](https://dinand.com/research/campaigns/sigil/) The registered prediction that review pays on irreversible decisions failed in all 12 series; a liability ledger raised wrongful blocking at the top tier to 42.7%. [Read the campaign ](https://dinand.com/research/campaigns/sigil/) Sep 9, 2026Stopped ### [Studies the gates stopped](https://dinand.com/research/studies-the-gates-stopped/) Two pre-registered studies stopped on rules written before they ran. One of them also retracts a published claim. [Read the campaign ](https://dinand.com/research/studies-the-gates-stopped/) Sep 3, 2026Mixed ### [Delegation Cliff](https://dinand.com/research/campaigns/delegation-cliff/) Each added level of delegation depth costs 1.50 on the quality margin; reviewer capacity buys back the most at +1.46. [Read the campaign ](https://dinand.com/research/campaigns/delegation-cliff/) Sep 2, 2026Measured ### [Error Independence: does different hardware buy a second opinion?](https://dinand.com/research/campaigns/error-independence/) The same model on different silicon shares its errors (phi +0.447 on contested items); models from different families do not (mean phi −0.102). [Read the campaign ](https://dinand.com/research/campaigns/error-independence/) Apr 4, 2026Measured ### [AGENESIS: first-mover AI adoption](https://dinand.com/research/campaigns/agenesis/) 100 runs across 4 industry verticals, 5 enterprises of 150 agents each per run: a first-mover premium of 210%, with the market tipping in 77% of markets. [Read the campaign ](https://dinand.com/research/campaigns/agenesis/) Lab notebook ## Studies & simulations 18 published studies, most of them pre-registered simulations: agent oversight, delegation, supply-chain stress and multi-agent markets. Latest study · Oct 18, 2026 ### [When watching every hour starts to pay](https://medium.com/@tinholt/when-watching-every-hour-starts-to-pay-caea3d7200ab) A pre-registered simulation held hourly monitoring of an agent fleet and six-hourly inspection to the same nominal reviewer budget, and found the budget at which hourly monitoring starts to catch faults. [Read on Medium ](https://medium.com/@tinholt/when-watching-every-hour-starts-to-pay-caea3d7200ab)[Read on LinkedIn ](https://lnkd.in/p/gtFwntpV) Oct 11, 2026Medium + LinkedIn ↗ ### [The Judge’s Mirror](https://medium.com/@tinholt/the-judges-mirror-0abe871ade6b) A preregistered study of how often a model judge agrees with itself on 197 blind case files, and how far down one model family a judge stays usable. Oversight & accountability Oct 7, 2026LinkedIn ↗ ### [The Open Decision Head: An Executive Briefing](https://www.linkedin.com/feed/update/urn:li:activity:7513629422663434240/) A retrained open decision head beat both a rule table and the hosted service on 800 fresh cases, after a first version that did not. The agentic enterprise Oct 1, 2026Medium ↗ ### [A Trial Balance for Agent Omissions](https://medium.com/@tinholt/a-trial-balance-for-agent-omissions-568a41c59aa1) Agent logs record what happened and have no row for what didn’t. A pre-registered test of commitment accounting, an obligation ledger reconciled against executed calls, for catching omissions in procurement-workflow traces. Oversight & accountability Sep 17, 2026Medium ↗ ### [Sprawl Cost Is Superlinear Under a Fixed Monitoring Budget: A Pre-Registered Calibration of Agent Portfolio Retirement Dynamics](https://medium.com/@tinholt/sprawl-cost-is-superlinear-under-a-fixed-monitoring-budget-a-pre-registered-calibration-of-agent-e24cbaa7e62a) AGENESIS-2 adds a portfolio layer to the AGENESIS enterprise simulation to measure whether agent sprawl costs more than proportionally under a fixed monitoring budget. Oversight & accountability Sep 17, 2026LinkedIn ↗ ### [Your AI Review Team May Be Waiting for Failures It Cannot See](https://www.linkedin.com/feed/update/urn:li:activity:7506213115978489856/) The executive reading of the sprawl-cost result. Oversight & accountability Sep 10, 2026Medium ↗ ### [Accountability Makes Oversight Worse: A Pre-Registered Test of Liability-Exposed AI Supervision Across Irreversibility Tiers](https://medium.com/@tinholt/accountability-makes-oversight-worse-a-pre-registered-test-of-liability-exposed-ai-supervision-ef4fdf46bad3) A pre-registered test of whether making an AI supervisor answerable for past harm improves its oversight, and why it backfired across irreversibility tiers. Oversight & accountability Sep 9, 2026LinkedIn ### [Skin in the Game Made the AI Supervisor Worse](https://dinand.com/publications/skin-in-the-game/) A controlled experiment on the cost of wrongful blocks and the incentives that shape an AI reviewer. Oversight & accountability Sep 3, 2026Medium ↗ ### [Pricing Agent Autonomy](https://medium.com/@tinholt/pricing-agent-autonomy-0a2bb5b97caf) A simulation sweeping twelve dimensions of organisational design to price the bet that moving work further from human review is cheaper. Delegation depth carries the strongest weight, and it is negative. Oversight & accountability Sep 2, 2026Medium ↗ ### [When the Second Opinion Shares the Blind Spot](https://medium.com/@tinholt/when-the-second-opinion-shares-the-blind-spot-7e4443dcb081) One model on two machines, with different quantisation and serving stacks, made the same mistakes. What separates one reviewer from another is the model family: a case for dissimilar redundancy. Oversight & accountability Aug 28, 2026Medium ↗ ### [A ledger for agentic decisions](https://medium.com/@tinholt/a-ledger-for-the-decisions-nobody-priced-625c452e43fd) Difficulty is a property of the work; risk is a property of the consequences. Testing, in simulation and as a product, the rule that decides which agentic decisions a person must see. Oversight & accountability Aug 27, 2026Medium ↗ ### [The Breaking Point](https://medium.com/@tinholt/the-breaking-point-2b68dbd7d7c2) 45,312 simulated supply-network designs, each run through a year of compounding disruption, and the exact stress level at which service breaks. The agentic enterprise Aug 16, 2026Medium ↗ ### [The Delegation Cliff](https://medium.com/@tinholt/the-delegation-cliff-ed787b1ebe22) 90,880 simulated designs of an AI-delegating organization. At a 95% quality floor they degrade gradually; at 99% the same design space develops a cliff. Oversight & accountability Jun 15, 2026Medium ### [Inventory Optimization Under Uncertainty: Broad random shocks trained a tougher inventory planner than a clever AI adversary did](https://dinand.com/publications/inventory-under-uncertainty/) An inventory-planning experiment explores which kinds of stress prepare a policy for unfamiliar disruptions. The agentic enterprise Apr 4, 2026Medium ↗ ### [What Happens When One Company Adopts AI Before Its Competitors?](https://medium.com/@tinholt/what-happens-when-one-company-adopts-ai-before-its-competitors-cd976e5a8dba) Results from a 100-run multi-agent enterprise simulation across four… The agentic enterprise Mar 31, 2026Medium ↗ ### [AI-Driven Inventory Optimization Under Deep Uncertainty](https://medium.com/@tinholt/ai-driven-inventory-optimization-under-deep-uncertainty-607fd80386d1) An Autoresearch… The agentic enterprise Mar 13, 2026Medium ↗ ### [Your AI Agent’s Brain: Specialized or Well-Briefed?](https://medium.com/@tinholt/your-ai-agents-brain-specialized-or-well-briefed-88ec339c6db4) What a Controlled Experiment Reveals About Enterprise AI… The agentic enterprise Feb 28, 2026Medium + LinkedIn ### [Building a Fake Economy Run by AI Agents to Negotiate Trade Promotions: Lessons from a CPG](https://dinand.com/publications/trade-promotion-agents/) Manufacturer and retailer agents negotiate promotions inside a simulated market with competing incentives. The agentic enterprise The machinery behind the studies Several experiments run on AGENESIS and the other systems profiled in the workshop. [AGENESIS profile ](https://dinand.com/experiments/agenesis/)[All experiments ](https://dinand.com/experiments/) Unpublished ## From the notebook Working pieces and explorations that live only on this site. 4 min read Study summary · draft ## [When monitoring stops scaling](https://dinand.com/research/when-monitoring-stops-scaling/) What a simulation suggests about the cost of overseeing a growing agent portfolio. [Read more ](https://dinand.com/research/when-monitoring-stops-scaling/) 4 min read Published article overview ## [When liability changes the reviewer](https://dinand.com/research/oversight-and-liability/) A controlled simulation of how a record of past harm affected an AI supervisor. [Read more ](https://dinand.com/research/oversight-and-liability/) 35 min read Paper ## [When watching every hour starts to pay](https://dinand.com/research/when-watching-every-hour-starts-to-pay/) A pre-registered simulation held hourly monitoring of an agent fleet and six-hourly inspection to the same nominal reviewer budget, and found the budget at which hourly monitoring starts to catch faults. [Read more ](https://dinand.com/research/when-watching-every-hour-starts-to-pay/) 5 min read Study summary ## [The Judge’s Mirror](https://dinand.com/research/the-judges-mirror/) How often a model judge agrees with itself on 197 blind case files, and how far down one model family a judge stays usable. [Read more ](https://dinand.com/research/the-judges-mirror/) --- Site guide for AI agents: https://dinand.com/llms.txt