Complex Systems with Patrick McKenzie (patio11)

Patrick McKenzie (patio11) reads his classic Bits about Money essay explaining why the debt collection industry earns its “river of effluvia” metaphor. From the accounting standards that force banks to "charge off" delinquent accounts to the large CSV files that constitute the only proof of a debt's existence, he explores how the system prioritizes accounting finality over legal and factual accuracy. The conversation reveals why the single most effective way to resolve a debt is often forcing a collector to read and write paper. They can’t, operationally, and don’t see that as an impediment to making money, especially because the paper will often document their lies and crimes.

–
Full transcript available here: https://www.complexsystemspodcast.com/debt-collections/

–
Presenting Sponsors: Mercury, Granola & Framer
If you have more interesting hobbies than managing your money, Mercury Personal is built for you. It allows you to automate movement between accounts—allocating paychecks and tax prep the moment they hit—with a sensible permissions model for partners or accountants. It works the way tech people expect banking to work. Go to mercury.com/personal to experience banking built by the same folks Patrick trusts for his business.
If meetings consistently leave you with hazy action items and lost context, Granola handles the transcription so you can actually participate and gives you searchable notes afterward. Try it free at granola.ai/complexsystems with code COMPLEXSYSTEMS
Building and maintaining marketing websites shouldn’t slow down your engineers. Framer gives design and marketing teams an all-in-one platform to ship landing pages, microsites, or full site redesigns instantly—without engineering bottlenecks. Get 30% off Framer Pro at framer.com/complexsystems.
–
Links:
https://www.bitsaboutmoney.com/archive/the-waste-stream-of-consumer-finance/ 

–
Timestamps:
(00:00) Intro
(05:20) The lifecycle of a defaulted debt
(14:17) The debt collection industry
(19:58) Sponsors: Mercury | Granola
(23:36) The debt collection industry (cont’d)
(25:55) The operations of a debt collection firm
(34:02) What does the debt collector hope to get from a call?
(36:52) Why does this continue being so broken?
(39:34) Suing people with robots
(40:26) Sponsor: Framer
(41:43) Suing people with robots (cont’d)
(44:14) What can be done about this?

Patrick McKenzie (patio11) and Philip Kiely, early employee at Baseten, discuss the inference stack: the critical layer of software and hardware that sits between a model’s weights and a user’s prompt. They cover inference engineering, how intermediate layers are evolving over a technical stack that is changing every six months, and how sophisticated organizations are actually consuming LLMs beyond just writing their questions into chatbot apps.
–
Full transcript available here: www.complexsystemspodcast.com/inference-engineering-with-philip-kiely/
–
Presenting Sponsors: Mercury, Meter, & Granola

Complex Systems is presented by Mercury—radically better banking for founders. Mercury offers the best wire experience anywhere: fast, reliable, and free for domestic U.S. wires, so you can stay focused on growing your business. Apply online in minutes at mercury.com.
Networking infrastructure has a way of accumulating technical debt faster than almost anything else in IT. Meter handles the full stack (wired, wireless, and cellular) as a single integrated solution: designed, deployed, and managed end-to-end so there's only one vendor to call when something goes wrong. Visit meter.com/complexsystems to book a demo. 

If meetings consistently leave you with hazy action items and lost context, Granola handles the transcription so you can actually participate and gives you searchable notes afterward. Try it free at granola.ai/complexsystems with code COMPLEXSYSTEMS
–
Links:
Download Inference Engineering: https://www.baseten.com/inference-engineering/ 
Philip's website: https://philipkiely.com/ 
Stripe's Emily Sands on Complex Systems: https://www.complexsystemspodcast.com/episodes/the-past-present-and-future-of-ai-with-stripe/ 
Des Traynor on Complex Systems: https://www.complexsystemspodcast.com/episodes/des-traynor/  
–
Timestamps:
(00:00) Intro
(00:30) The AI deployment pipeline
(03:04) Evolution of abstraction layers in engineering
(05:14) Defining inference and model weights
(08:45) Architecture of language and diffusion models
(10:11) AI adoption in the broader economy
(11:30) The shift toward agentic workflows and RL
(14:55) Function calling and real-world actions
(20:10) Sponsors: Mercury | Meter
(22:59) Technologies for agentic tools: MCP and skills
(25:32) The craft of writing a harness
(29:56) Using AI for automated proofreading and tool creation
(34:12) Balancing LLMs with deterministic code
(37:31) Observability and chain of thought reasoning
(39:31) Sponsor: Granola
(41:21) Observability and chain of thought reasoning
(50:45) Speculative decoding and hidden states
(55:37) The value of smaller, task-specific models
(59:55) Internal competencies versus buying solutions
(01:09:27) Self-publishing a technical book in record time
(01:23:20) Wrap

Patrick McKenzie (patio11) deconstructs the "original sin" of payments: building a global financial substrate on shared secrets that were distributed promiscuously to function. He examines the multi-decade game of Whack-a-Mole played by the industry to balance the "optimal amount of fraud" against the catastrophic conversion hit of high-friction security. From the physical failure of terminal buttons to the smartphone finally solving the lifecycle problem of cryptographic tokens, Patrick explores the technical and social reasons why we’ve moved from "something you know" to the "continuity of access" provided by the device in your pocket.
–
Full transcript available here: www.complexsystemspodcast.com/secondary-auth/

–
Presenting Sponsors: Mercury & Granola

If you have more interesting hobbies than managing your money, Mercury Personal is built for you. It allows you to automate movement between accounts—allocating paychecks and tax prep the moment they hit—with a sensible permissions model for partners or accountants. It works the way tech people expect banking to work. Go to mercury.com/personal to experience banking built by the same folks Patrick trusts for his business. 
If meetings consistently leave you with hazy action items and lost context, Granola handles the transcription so you can actually participate and gives you searchable notes afterward. Try it free at granola.ai/complexsystems with code COMPLEXSYSTEMS
–
Links:
Bits about Money: https://www.bitsaboutmoney.com/
Emily Sands on Complex Systems: https://www.complexsystemspodcast.com/episodes/the-past-present-and-future-of-ai-with-stripe/ 
–
Timestamps:
(00:00) Intro
(01:32) Publishing the shared secret… again
(03:39) Manufacturing shared secrets at scale
(07:51) Something you own, take one
(10:10) Sponsors: Mercury | Granola
(13:48) Something you own, take two
(18:26) Something you own, take three
(21:24) One other semi-successful method: positive pay
(24:45) Wrap

Patrick McKenzie (patio11) and Luke Farrell examine the structural "technical imagination" gap that prevents the US government from delivering high-fidelity digital services. They discuss why states routinely pay full price 29 times for the same buggy codebase, why failure is the default outcome, and why rooms full of government administrators cannot muster the expertise to say a two line code change should be trivial. They also discuss Luke’s work on the "means testing industrial complex,” why the government redundantly pays a private vendor to do a SQL query for information the IRS already knows, and what vendors would say about their own discontents.
–
Full transcript available here: http://www.complexsystemspodcast.com/understanding-government-procurement-with-luke-farrell/
–
Presenting Sponsors: Mercury &  Framer

If you have more interesting hobbies than managing your money, Mercury Personal is built for you. It allows you to automate movement between accounts—allocating paychecks and tax prep the moment they hit—with a sensible permissions model for partners or accountants. It works the way tech people expect banking to work. Go to mercury.com/personal to experience banking built by the same folks Patrick trusts for his business.
Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column N.A., Members FDIC.
Building and maintaining marketing websites shouldn’t slow down your engineers. Framer gives design and marketing teams an all-in-one platform to ship landing pages, microsites, or full site redesigns instantly—without engineering bottlenecks. Get 30% off Framer Pro at framer.com/complexsystems.
–
Links:
Luke Farrell's Substack: https://donmoynihan.substack.com/
Luke Farrell, The Means-Testing Industrial Complex: https://donmoynihan.substack.com/p/the-means-testing-industrial-complex
–
Timestamps:
(00:00) Intro
(01:52) Transitioning from Google to the US Digital Service (USDS) 
(05:18) How rule buildup and administrative burdens create "Kafkaesque" mazes 
(08:21) Using diagrams and funnels to visualize benefit denials 
(11:49) Software logic errors that improperly kicked children off Medicaid 
(18:25) Why government payroll IT costs hundreds of millions of dollars 
(20:02) Sponsors: Mercury and Framer
(22:02) How recursive legal requirements and DOD standards inflate IT scope 
(26:57) Market consolidation and the lack of competition in procurement 
(33:47) Aligning program administrator incentives with successful service delivery 
(36:03) Using in-house technologists to push back on vendor change orders 
(39:27) Shifting from "Big Bang" contracts to iterative, agile development 
(53:10) The moral incoherence of asset limits 
(01:11:36) Insourcing electronic income verification databases 
(01:16:56) Building public sector competence to manage modern technical risk 
(01:20:08) Wrap

Patrick McKenzie (patio11) reads an essay about "industrial-scale" fraud and why it should be treated as a professional business process rather than a series of isolated accidents. He explains how fraudsters leverage specialized supply chains—shared CPAs, incorporation agents, and "least attentive" banks—to loot public funds. Patrick argues that the government’s "pay-and-chase" model is fundamentally broken and suggests that simple "proof of work" functions, like a 30-second cell phone video of a workspace, could provide the visceral signal that paperwork lacks, and examines the state’s lack of "object permanence" regarding serial fraudsters and how scaled data provides the defense-side advantage needed to catch modern frauds.
–
Full transcript available here: www.complexsystemspodcast.com/fraud-as-infrastructure/
–
Presenting Sponsor: Mercury
Complex Systems is presented by Mercury—radically better banking for founders. Mercury offers the best wire experience anywhere: fast, reliable, and free for domestic U.S. wires, so you can stay focused on growing your business. Apply online in minutes at mercury.com.
Mercury is a fintech company, not an FDIC-insured bank. Banking services provided through Choice Financial Group and Column N.A., Members FDIC.
–
Links:
Bits about Money: https://www.bitsaboutmoney.com/archive/fraud-investigation/ 
Dan Davies on Complex Systems: https://open.spotify.com/episode/5QKxzgumJXSQuaWCmYAoM9 
Jetson Leder-Luis on Complex Systems podcast: https://open.spotify.com/episode/3NiC7x9edoxJXkNW9vRfAT 
Stripe’s Emily Sands on Complex Systems: https://open.spotify.com/episode/64Dyh6Gbg1lg4qUFwId0hc 
–
Timestamps:
(00:00) Intro
(05:23) In which we briefly return to Minnesota
(09:26) Common signals, methods, and epiphenomena of fraud
(09:30) Fraudsters are playing an iterated game
(11:29) The fraud supply chain is detectable
(14:27) Investigators should expect to find ethnically clustered fraud
(20:11) Sponsor: Mercury
(21:47) High growth rate opportunities attract frauds
(26:04) Fraudsters find the weakest links in the financial system
(32:35) Frauds openly suborn identities
(35:57) Asymmetry in attacker and defender burdens of proof
(40:13) Fraudsters under-paperwork their epiphenomena
(44:22) Machine learning can adaptively identify fraud
(48:14) Frauds have a lifecycle
(50:34) Should we care about fraud investigation, anyway