Podcasts about bayesian

I. Matthew Adelstein (aka Bentham's Bulldog) and Richard Chappell debate the existence of God. So do JP Andrew, Benjamin Tettu, Sangwon Lee, Ethan Muse, other Ethan, and "Atheistic Journals", all within this month! I thought the omnipresence of the early '00s online atheism debate was a time-bound historical phenomenon. But now that the blogosphere has reincarnated as Substack, it's back, nearly as strong as ever. I enjoyed participated in the early 00's version, but the current incarnation leaves me cold. I can't get as interested as I once did in the details of the Fine-Tuning Argument, the Argument From Miracles, et cetera. It's not just that I've already served my time in the theology mines. My lack of belief in God intuitively feels like it precedes all of these things. Consider by analogy a society where people regularly debated whether aliens assisted the rise of Napoleon. There are many reasons to believe in such assistance: Napoleon's rise from Corsican nobody to Emperor and would-be world conqueror was a bizarre deviation from the usual pattern of history. He won battles outnumbered 4-to-1, when everyone knows victory usually goes to the larger army. He became monarch of France immediately after it had declared eternal enmity against the concept of monarchy and resolved to murder anyone who didn't hate monarchy enough. And the arguments against such intervention are comparatively flimsy. True, nobody saw aliens, but a sufficiently advanced civilization could conduct its interventions in secret. True, our prior for alien intervention is generally low, but maybe this is unjustified: we have no idea how many historical events the aliens have intervened in; Alexander the Great was pretty strange too. True, Napoleon finally lost, but this is not decisive; maybe the aliens only wanted him to blaze brightly for a few years before his eventual defeat, or maybe stronger aliens were backing the Duke of Wellington. In a world where this hypothesis somehow entered the space of serious debate, it would be hard to dislodge. Our own historical community is protected from it not because they can precisely add up the Bayesian evidence for and against each of these considerations and conclude its falsehood, but because common sense precludes entertaining it in the first place. This isn't to say that Reason doesn't work - if someone could precisely add up the Bayesian evidence for and against these considerations, they would get the right answer - but as a historical matter, this type of delicate reasoning wasn't how our society resolved this question. My atheism is more deeply rooted in whatever prevented us from considering this question in the first place than in the type of delicate reasoning that could get us out of it. https://www.astralcodexten.com/p/why-im-staying-out-of-the-substack

Today's clip is from Episode 165, featuring Alex Fengler. In this conversation, Alex introduces Bayesify , a tool that uses AI to analyze research papers and assess how well they follow a Bayesian workflow.He explains how the tool breaks an analysis down step by step, identifies strengths and weaknesses, and provides suggestions for improving the paper. They also discuss how BasiFi can be used as an educational resource, a review engine for researchers, and potentially as a way to study how the quality of Bayesian analyses has changed over time.Alex also explains why the team is building a human-rated "gold set" of papers to evaluate how well the tool's scoring aligns with expert judgment. It's an interesting example of how AI can be used not just to generate research, but to help verify and evaluate statistical workflows.Full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free):Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

A physics professor ran the energy budget on the Nimitz tic-tac. The maneuver needs about a thousand gigawatts, more than ten times every nuclear plant in the United States combined. Nothing we build survives its own waste heat at that scale. So what was flying? Kevin Knuth is a professor of physics at the University at Albany and a former NASA scientist who now applies Bayesian statistics to UAP sensor data. We cover: why a 5,400 g maneuver would turn a pilot into soup, where the energy went when the object stopped and no explosion followed, the 1954 radar measurements nobody wants to talk about, the New Zealand frigate that went dark and adrift in three seconds, and why the records from that deployment stay classified until 2062. Knuth says the power numbers alone rule out human technology. I push back on whether the sensors agree. Newsletter, full transcript, and my Monday M.A.G.I.C. Message: briankeating.com/aliens CHAPTERS 00:00 Why this field needs gatekeepers 03:50 How anomalous is anomalous? 06:40 The UFO photo that was a seagull 10:40 Radar clocked them at 19 km/s in 1954 15:00 5,400 g and a body weighing 227 tonnes 19:00 The radar return problem he had not considered 21:45 A gigawatt of waste heat and nowhere to put it 24:30 The 747 chased by something carrier sized 29:00 Why a number without an error bar is worthless 34:30 The NASA job he was not qualified for 37:00 Proxima Centauri in a day and a half 41:00 The warship that lost all power in 3 seconds 46:30 Pushing a column of water that cannot compress 48:45 The light that burned bark off the trees 53:00 What the power numbers rule out 56:00 The 95 percent problem Learn more about your ad choices. Visit megaphone.fm/adchoices

Frank Hutter, co-founder of Prior Labs, talks about TabPFN, a tabular foundation model that makes predictions in a single forward pass, and the research behind it.TabPFN is pre-trained on synthetic datasets drawn from a prior over structural causal models, rather than on real data. At prediction time it takes the whole training table as context and outputs an approximation of the Bayesian posterior predictive distribution, without per-dataset training or hyperparameter search. Frank explains how this grew out of his earlier work on AutoML and neural architecture search, how the priors are built and revised, and why tabular data was hard for deep learning for so long.The conversation also covers the TabArena benchmark, how the architecture changed from TabPFN v1 to v3, scaling to larger tables, using the model with coding agents, test-time compute, Google's TabFM, causal inference and interventions, and relational data. At the end, a short update Frank recorded after the interview covers the TabPFN-3.5 release.Prior Labs:TabPFN-3.5: https://priorlabs.ai/tabpfn-3-5https://priorlabs.ai/careersTOC:00:00 Introduction00:44 Welcome and Frank's background02:05 Why tabular data was hard for deep learning10:17 Pre-training on synthetic data12:52 The TabArena benchmark19:28 From AutoML to neural architecture search26:34 TabPFN as a learned algorithm30:50 Bayesian prediction in one forward pass39:37 Scaling to larger tables47:48 Using TabPFN with coding agents57:47 Output heads and architecture from v1 to v31:05:29 Test-time compute and adaptation1:13:32 Google's TabFM1:16:53 How the priors are designed1:18:40 Correlation, causation and interventions1:35:22 Relational and multimodal data1:38:31 Use in organisations1:46:38 The open research arm1:50:21 Update: TabPFN-3.5REFS:TabPFN v2, Nature (Hollmann et al., 2025)https://www.nature.com/articles/s41586-024-08328-6Transformers Can Do Bayesian Inference (Müller et al.)https://arxiv.org/abs/2112.10510TabArena (Erickson et al.)https://arxiv.org/abs/2506.16791AutoGluon-Tabular (Erickson et al.)https://arxiv.org/abs/2003.06505Beyond IID: How General Are Tabular Foundation Models, Really?https://arxiv.org/abs/2606.30410Neural Architecture Search: A Survey (Elsken, Metzen & Hutter)https://arxiv.org/abs/1808.05377Auto-WEKA (Thornton et al.)https://www.cs.ubc.ca/~hutter/papers/AutoWEKA-KDD2013.pdfTabPFN v1 (Hollmann et al., 2022)https://arxiv.org/abs/2207.01848TabPFN-3 technical reporthttps://arxiv.org/abs/2605.13986TabPFN-2.5 reporthttps://arxiv.org/abs/2511.08667CAAFE (Hollmann et al.)https://arxiv.org/abs/2305.03403TabICL (Qu et al.)https://arxiv.org/abs/2502.05564TabICLv2 (Qu et al.)https://arxiv.org/abs/2602.11139Google TabFMhttps://research.google/blog/introducing-tabfm-a-zero-shot-foundation-model-for-tabular-data/TALENT benchmark (Ye et al.)https://arxiv.org/abs/2407.00956Do-PFN (Robertson et al.)https://arxiv.org/abs/2506.06039CausalPFN (Balazadeh et al.)https://arxiv.org/abs/2506.07918Causal Foundation Models with Partial Graphs (Reuter et al.)https://arxiv.org/abs/2602.14972RelBench (Robinson et al.)https://arxiv.org/abs/2407.20060RelArena-α, TabPFN-Rel and RPIhttps://arxiv.org/abs/2608.16319TabPFN on GitHubhttps://github.com/PriorLabs/TabPFNTabPFN-3.5 technical reporthttps://arxiv.org/abs/2609.17895Otto Group Product Classification Challenge (Kaggle, 2015)https://www.kaggle.com/competitions/otto-group-product-classification-challenge---RESCRIPT:https://app.rescript.info/share/e99676c25ee6189fbf54c9be07eb623e

A platelet count of 40 and a central line to put in. Transfuse first, or go ahead? Right now the answer depends on which ICU you work in, and nobody actually knows which is right. In this episode of the eCritCare Podcast, Dr Swapnil Pawar is joined by Professor Peter Watkinson (Professor of Intensive Care Medicine, University of Oxford, and Chief Investigator) and Professor Simon Stanworth (Professor of Haematology and Transfusion Medicine, University of Oxford) to discuss the T4P trial, Threshold for Platelets. T4P is a Bayesian adaptive randomised trial across five platelet transfusion thresholds, from below 10 to below 50, in critically ill patients undergoing low bleeding risk invasive procedures. It is recruiting 2,550 patients across the UK, Australia and Canada and has just passed the halfway mark. In this conversation: Why platelet transfusion practice in critical care was borrowed from blood cancer patients, and why the evidence never caught up The UK survey that showed thresholds anywhere from under 10 to under 50 for the same procedure The risks side of the ledger: a biological product, transfusion reactions, unknown effects, and around £290 per unit before lab and administration costs Why five thresholds instead of two: drawing the threshold-response curve so the optimum does not have to be one of the arms Which procedures are specified (central lines, pleural aspiration, paracentesis) and which are at clinician discretion The adherence question, and why even the below-10 arm is being followed The haematology confusion: why a patient's daily platelet threshold does not stop them being in T4P Rotating trainees, cross-specialty equipoise, and asking clinicians to step outside a comfort zone they have practised in for years Operator experience, ultrasound-guided lines, and what PACER did and did not show in critical care What success looks like: the biggest platelet transfusion study in critical care, and a practice-changing curve Read this episode on Substack: https://critcareedu.substack.com/p/t4p-trial-platelet-transfusion-thresholds Trial registration: ISRCTN79371664. Funded by the NIHR Health Technology Assessment Programme (NIHR131822).

Earlier this year, the FDA released draft guidance endorsing Bayesian methods and simulation-based evidence in clinical trials in a significant shift in how regulators think about predictive medicine.  In a new pharmaphorum podcast, Dr Irina Babina, CEO of oncology R&D platform Concr, discusses how current regulatory frameworks were built for static, one-test-one-answer diagnostics, as well as the potential of digital twins, and how, if regulators keep approving AI medical tools the way they approve blood tests, the most promising predictive technologies will never reach patients. You can listen to episode 276 of the pharmaphorum podcast in the player below, download the episode to your computer, or find it – and subscribe to the rest of the series – on Apple Podcasts, Spotify, Overcast, Pocket Casts, Podbean, and pretty much wherever else you download your other podcasts from.

Support & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!Takeaways:Q: What is HSSM and how does it relate to HDDM?A: HSSM stands for hierarchical sequential sampling models, a generalization of HDDM (hierarchical drift diffusion models), the older toolbox for the same class of decision-making models, but HSSM is built from the ground up on simulation-based inference. That's what lets it handle any variation of the underlying process model, not just the ones with a tractable closed-form likelihood.Q: What is the drift diffusion model and why has cognitive science relied on it so heavily?A: The drift diffusion model treats a decision as a random walk that accumulates evidence until it crosses one of two boundaries, with parameters controlling boundary separation, starting bias, and drift rate. It's been used in thousands of published papers largely because it has a closed-form likelihood, which makes standard Bayesian and maximum-likelihood inference fast. Small variations on the model are often just as scientifically motivated, but if their likelihoods aren't analytically convenient, the literature using them stays sparse.Q: What is a likelihood approximation network (LAN) and what does it actually learn?A: A LAN is a neural network trained to take in a process's parameters and a trial's outcome and output how likely that outcome was, learned purely from repeated simulation rather than derived analytically. Once trained, it functions as a fast, reusable likelihood you plug directly into Bayes' rule, in place of a closed-form solution that may not exist for the model you actually want to fit.Q: What's the difference between amortizing the likelihood and amortizing the posterior?A: Amortizing the likelihood, HSSM's approach, means training a network once to approximate the likelihood, then reusing that same network across arbitrarily many downstream models: different priors, hierarchical structures, or regression backends, with no retraining. Amortizing the posterior directly, the approach tools like BayesFlow take, gives near-instant inference once trained, but locks the network into the specific scenario it was trained for.Chapters:00:00:00 What is HSSM and how does it fit into the Bayesian inference landscape?00:12:09 How did HSSM evolve from HDDM, and what does it apply to?00:30:25 How do neural networks learn likelihoods for Bayesian inference?00:37:01 What makes amortized Bayesian inference so flexible?00:41:04 What are the real computational costs of amortized inference?00:55:16 How does HSSM integrate with libraries like BayesFlow?00:58:57 What does a live demo of HSSM and BayesFlow look like?01:18:33 What is Bayesify and how does it score a paper's Bayesian workflow?01:23:12 What new model classes are coming to the HSSM ecosystem?01:30:12 How is AI reshaping development in the HSSM ecosystem?01:38:42 How should society incentivize keeping hard cognitive skills alive?Thank you to my Patrons for making this episode possible!Links from the show.

What if you could forecast demand but couldn't build, order, or control the supply needed to meet it? A matter of fact you don't even own your supply and have to forecast inventory as well.I sit down with Airbnb data scientist Harrison Katz, PhD, a designer of Bayesian workflows for structured forecasting, to explore the challenge of forecasting when supply depends entirely on others. We discuss uncertainty on both sides of the equation, where demand and supply planning become blurred, and how constraints, demand shaping, and probability come together. Harrison shares how he approaches these problems, how results should be measured, and even turns the tables to ask me a few tough questions.IBF On Demand sponsored by Arkieva, your one-plan S&OP software. Learn more about Arkieva's innovative approach to forecasting here https://arkieva.com/Support the showTo sign up for regular updates and the latest research, events, articles, podcasts and more from the Institute of Business Forecasting & Training, visit www.ibf.org

Medical Disclaimer: The information shared on This Week in HRV is for educational purposes only and is not intended as medical advice. Always consult a qualified healthcare professional before making changes to your health routine. This week on This Week in Heart Rate Variability, we explore the theme of Medical — studying HRV across populations dealing with identifiable disease states and clinical questions. From noise exposure in everyday life to smoking, cardiac surgery rehabilitation, rare autoimmune disease, AI-powered sleep staging, and physical activity in chronic lung disease, today's research spans the full clinical spectrum of what HRV can tell us. 1. Every 10 dB increase in environmental noise reduces SDNN by up to 16% over 160 minutes: evidence from the Apple Hearing Study PUBLICATION: Journal of Exposure Science and Environmental Epidemiology (Nature Portfolio)AUTHORS: Xin Zhang, Sung Kyun Park, Lauren M. Smith, and Richard L. Neitzel KEY FINDING: Using wearable-derived SDNN and sound-exposure data from roughly 980–1,180 participants per analysis in the Apple Hearing Study, hierarchical Bayesian distributed-lag models found that each 10 dB increase in environmental noise was associated with an overall SDNN reduction of 6.8% over a short-term (20-minute) window and 16.0% over a long-term (160-minute) window. Headphone sound produced smaller, more consistent reductions of 7.1–7.4% regardless of window length. Reductions were greater in older adults and those with tinnitus. SIGNIFICANCE: This elevates noise — including overnight traffic noise and headphone volume — to the same level of concern as sleep quality, training load, and nutrition in shaping HRV. The data also provides context for unexpectedly low morning HRV readings after noisy nights. Read the full study 2. Ambulatory monitoring reveals what office readings miss: smokers have heart rates nearly 7 bpm higher in daily life — and elevated diastolic blood pressure PUBLICATION: Hypertension Research (Nature Portfolio)AUTHORS: Yuya Akagi, Kimika Arakawa, Atsushi Sakima, Mai Kabayama, Ken Sugimoto, Yuichi Akasaki, Ako Fukami, Satoko Sakata, Hirochika Ryuno, Hiroyuki Kadoya, Toshiya Yamamoto, Hiroko Yoshida, Yoichi Nozato, Taisuke Ueno, Makiko Abe, Hisatomi Arima, Mitsuru Ohishi, Nobuhito Hirawa, Chisa Matsumoto, Shin-Ichiro Miura, Masaki Mogi, Akira Nishiyama, Akihiro Nomura, Takayoshi Ohkubo, Yusuke Ohya, Shigeru Shibata, Yasuharu Tabara, Koichi Yamamoto, and Kazuomi Kario KEY FINDING: This systematic review and meta-analysis of four non-randomized crossover comparative studies (104 participants total) compared ambulatory blood pressure and heart rate within the same current smokers during active smoking versus abstinence periods (1 day to 1 week). Twenty-four-hour diastolic blood pressure was 2.67 mmHg higher during smoking (95% CI 0.72–4.62), and daytime diastolic was 3.17 mmHg higher. Twenty-four-hour heart rate was 6.47 bpm higher during smoking (95% CI 3.98–8.96); daytime 7.19 bpm higher; nighttime 3.58 bpm higher. Systolic blood pressure showed no significant difference between periods. SIGNIFICANCE: This helps explain the long-standing office blood pressure paradox in smokers: they appear similar to non-smokers in clinical readings because they abstain before appointments and are at rest. Ambulatory monitoring reveals the true picture — persistent sympathetic arousal manifesting as chronically elevated heart rate around the clock. For HRV practitioners, this means a smoking patient's measured baseline is not their true autonomic baseline, and it positions HRV tracking as a motivational tool in cessation progra...

Recorded at ANZCA and the Faculty of Pain Medicine's Annual Scientific Meeting in Auckland, TopMedTalk hosts Andy Cumstey and Kate Leslie interview visiting pain research leaders, Professor Irene Tracey, the Vice-Chancellor of the University of Oxford and Professor Lesley Colvin, the Chair of Pain Medicine at the University of Dundee and an Honorary Consultant in Anaesthesia and Pain Medicine at NHS Tayside. They discuss how thinking has shifted from chronic pain as a simple transition from acute pain, to a distinct disease with separate mechanisms and early "pre-vulnerability" factors including stress-axis disruption and reduced endogenous inhibition. Colvin outlines epidemiological and neuroimaging evidence linking adverse childhood experiences, including repeated medical procedures, to higher risks of chronic pain and other long-term conditions, with reward-system blunting also seen in depression and more severe pain. Tracey explains the "Bayesian brain," where dominant priors can sustain pain, influencing treatment and trial design, while both highlight brain plasticity, emerging neurotech such as VR-based training, and behavior-change approaches to increase physical activity. -- The 2026 International Practicum on Cardiopulmonary Exercise Testing will be held at the Balmer Lawn Hotel in Brockenhurst, UK, from September 16th to 18th this year. It is organised by iPOETTS , the international perioperative testing and training society. Come and join us at this premier educational event designed for clinicians, scientists, and healthcare professionals interested in sport, exercise, and perioperative medicine. This is an International Perioperative Testing and Training Society accredited event so when you attend you can get your iPOETTS accreditation, showing that you are a practitioner who has reached a high, standardized level of competence in performing and interpreting Cardiopulmonary Exercise Testing (CPET) for patients preparing for major surgery. Go now to http://www.ebpom.org

Today's clip is from Episode 164, featuring Andrew Gelman, Aki Vehtari & Richard McElreath. In this conversation, Andrew explains how Bayesian principal stratification can be used to reason about treatment effects when there is an intermediate treatment or outcome that is only partially observed.He discusses how latent variables can represent whether someone would take a stage-two treatment, and how pre-treatment characteristics such as age, location, and past spending can help build a model for this process.Richard connects the problem to the broader distinction between per-protocol and intent-to-treat analyses, and they discuss how standard approaches such as instrumental variables can be understood as special cases of more general Bayesian models. It's a useful example of how Bayesian modeling can represent the full process behind a causal question rather than relying on simplifying assumptions.Full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free): Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

I've tried various times to summarize the core question my research is trying to tackle (and, indeed, I often think of research progress as a process of asking increasingly good core questions). This post gives the deepest version of that question I've found thus far: how should you relate to the parts of the world you can't directly model or control? Let me explain further in terms of a distinction between two perspectives. From the third person perspective you think of yourself as “outside” the world, looking in. You're a good Bayesian, in that you have a set of mutually exclusive collectively exhaustive hypotheses. You choose actions by multiplying your credences by your utilities over those hypotheses, and you treat those actions as the only way you influence the world. Some problems with the third person perspective (aka Cartesian or dualistic agency) were described in Scott and Abram's sequence on embedded agency. One crucial issue is that most realistic environments contain other agents which are modeling you back, which means that your thoughts might affect the world via channels that aren't just your actions. Game theory somewhat mitigates this problem, but only in the very specific case where all [...] ---Outline:(05:08) Rationality of reward(09:09) Letters from spirits(12:21) Languages as Schelling points(15:43) Actions and entanglements The original text contained 1 footnote which was omitted from this narration. --- First published: September 1st, 2026 Source: https://www.lesswrong.com/posts/pYFBD2SnqiWkuNns5/explaining-knightianism-on-one-foot --- Narrated by TYPE III AUDIO.

On a humid August night off the coast of Sicily, British billionaire, Mike Lynch, along with 21 other passengers, were sleeping soundly on Lynch's luxury yacht, the Bayesian. Then, all of a sudden, a surging storm hammered the vessel. Shockingly, within minutes the Bayesian had capsized, tragically sending Lynch, and 7 others down with her. Now, you'd presume this was just an unlucky case of nature's brutality. However, some theorists out there believe that somehow, Lynch was deliberately targeted that night! Intrigued? Well, get your trench coat and magnifying glass, as we re-open the case files on this, and plenty more, supposedly ‘solved' mysteries!Our Sponsors:* Check out BetterHelp: https://www.betterhelp.comAdvertising Inquiries: https://redcircle.com/brandsPrivacy & Opt-Out: https://redcircle.com/privacy

In this episode, Dr Swapnil Pawar sits down with Professor Ian Seppelt, intensivist at Nepean Hospital, Sydney, and one of the lead investigators of the SuDDICU trial, to unpack the decades-long story of Selective Digestive Decontamination (SDD) in intensive care. Ian traces SDD from its haematology origins in the 1970s, through its adoption in the Netherlands, to the extraordinary behind-the-scenes journey of SuDDICU: failed international grants, the pivot from a parallel-group to a crossover cluster design, manufacturing a pharmaceutical-grade SDD preparation from scratch, and navigating ethics approval for a trial without individual patient consent. They discuss: What SDD actually is, and why "selective" matters Why infectious disease physicians resisted a therapy with one of the largest evidence bases in critical care The SuDDICU results: a non-significant mortality benefit, fewer positive blood cultures, fewer resistant organisms, and no signal of emerging resistance The Bayesian view, with a high posterior probability of survival benefit Where SDD goes next: enriched subgroups (neurological injury, pancreatitis), the PREVENT trial of prophylactic ceftriaxone in acute brain injury, and the unanswered question of high-resistance settings Why we adopt some interventions and ignore others: the psychology of evidence in ICU

Artificial intelligence is advancing rapidly in healthcare, but what could it mean for pre-hospital care? In this compilation episode, we bring together two conversations exploring how AI could support clinicians, improve decision-making, and transform the way emergency care is delivered.We begin with Zane Perkins, exploring AI-TRiPS, an AI Risk Prediction and Decision Support System for trauma care. We discuss how AI and Bayesian networks could support clinical reasoning, identify risk and reduce cognitive burden in high-pressure environments such as major trauma and air ambulance care.But with that potential comes important questions around trust, cognitive bias, human factors and clinical responsibility. What happens when clinicians disagree with an algorithm? And could we become too reliant on technology?We then broaden the discussion with Nico Preston, exploring the wider applications of AI across pre-hospital care, from dispatch and demand prediction to diagnostics, risk stratification and clinical decision support.Across both conversations, one theme becomes clear: the future of AI in emergency care is unlikely to be about replacing clinicians. Instead, the opportunity may be to create human–AI systems that combine computational power with clinical experience, judgement and contextual awareness. But getting that balance right will be critical. Can AI make us better clinicians, and how do we ensure we use it safely?This episode is sponsored by PAX: The gold standard in emergency response bags.When you're working under pressure, your kit needs to be dependable, tough, and intuitive. That's exactly what you get with PAX. Every bag is handcrafted by expert tailors who understand the demands of pre-hospital care. From the high-tech, skin-friendly, and environmentally responsible materials to the cutting-edge welding process that reduces seams and makes cleaning easier, PAX puts performance first. They've partnered with 3M to perfect reflective surfaces for better visibility, and the bright grey interior makes finding gear fast and effortless, even in low light. With over 200 designs, PAX bags are made to suit your role, needs, and environment. And thanks to their modular system, many bags work seamlessly together, no matter the setup.PAX doesn't chase trends. Their designs stay consistent, so once you know one, you know them all. And if your bag ever takes a beating? Their in-house repair team will bring it back to life.PAX – built to perform, made to last.Learn more at ⁠⁠https://www.pax-bags.com/en/⁠

Today's clip is from Episode 164, featuring Andrew Gelman, Aki Vehtari & Richard McElreath. In this conversation, Andrew explains why a Bayesian workflow goes far beyond simply fitting a model. He discusses the importance of building, fitting, and checking models, and why moving between simpler and more complicated models can reveal insights that a single model might miss. He also explores how simulation and generative modeling can help researchers evaluate new models and gain confidence in their results, even when there isn't an established method or published study to rely on. It's a look at why good statistical practice isn't just about getting an answer, but knowing how much you can trust it. Full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free): Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

Support & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome workTakeaways:Q: What is the "Bayesian Workflow" book about, and who is it for?A: It covers what the three authors know that isn't already in Bayesian Data Analysis (BDA3) or Statistical Rethinking, organized around case studies that walk through full analyses end to end rather than just giving a recommendation. It's not an introduction to Bayesian inference -- it assumes you already know the basics -- but a guide to making theoretically informed, professional decisions at the many branching points a real analysis involves that source books rarely acknowledge.Q: What's a concrete way to report Bayesian results without just handing over a posterior distribution?A: Report a few named scenarios from the distribution, such as pessimistic, median, and optimistic. This is easier to discuss than a full posterior and helps shift the conversation toward what would move outcomes from the median toward the optimistic case.Chapters:00:00:00 Who are this episode's three guests, and what is Bayesian Workflow?00:02:35 What's new: the LBS Instagram account and the Carnegie Mellon workshop?00:05:22 What is Richard McElreath's origin story, from anthropology to statistics?00:18:22 What is the elevator pitch for the Bayesian Workflow book?00:20:12 Where does workflow sit between statistical theory and case studies?00:27:21 Why express your scientific background in a generative model?00:35:04 What came out of the LBS listener contest?00:36:43 How is a Bayesian workflow different from a pipeline?00:39:03 What is reverse Bayes, and how does it help with prior sensitivity?00:43:53 How do Bayesians reinterpret non-Bayesian methods?00:45:02 How is the Bayesian Workflow book structured?00:46:51 Who is Dorota, the LBS contest grand prize winner?00:52:24 When does a hierarchical model stop being an innocuous assumption?00:58:17 Can multilevel regression and poststratification pool detection across sites?00:59:32 Why start with a big generative simulation before the statistical model?01:02:05 What is the "secret weapon" of comparing shrinkage to fixed-effects estimates?01:11:02 How do you detect which assumptions are actually driving your inference?01:15:24 How do you get regulated industries to accept a posterior instead of a score?01:22:04 Should statisticians soften uncertainty for decision makers?01:23:11 Why report three scenarios instead of a single number?01:25:16 How do you communicate survival probabilities to cancer doctors?01:27:51 How do you handle a leaky instrument in causal inference?01:29:16 What is a principal stratification model?01:34:47 What are the three authors working on next?01:39:53 If you had unlimited time and resources, which problem would you solve?01:41:26 Could statistical workflow be made more axiomatic?01:42:03 Which great scientific mind would you have dinner with?Thank you to my Patrons for making this episode possible!Links from the show

In this episode of Communicable, Amy Legg (Brisbane, Australia), Rekha Pai Mangalore (Melbourne, Australia) & Angela Huttner (Geneva, Switzerland) return for the third part of the ‘We will make you love PK/PD' series. The hosts are joined by two experts, Amanda Gwee (Melbourne, Australia) & Sebastian Wicha (Hamburg, Germany), to dive into an important application of PK/PD: model-informed precision dosing (MIPD), and how it allows for more informed, individualised dosing decisions. The episode also covers the open-access educational tools, KidsCalc and TDMx, developed by Gwee and Wicha respectively, and the steps clinicians can take to put MIPD into practice, keeping the individual patient at the centre of dosing decisions.Resources TDMx, https://www.tdmx.eu/KidsCalc, https://www.kidscalc.org/Further readingBroeker A, et al. Towards precision dosing of vancomycin: a systematic evaluation of pharmacometric models for Bayesian forecasting. CMI 2019. DOI: 10.1016/j.cmi.2019.02.029Brusamarello C, et al. How important are MIC determination methods when targeting vancomycin levels in patients with Staphylococcus aureus infections? J Antimicrob Chemother 2021. DOI: 10.1093/jac/dkab065De Cock PA, et al. Bedside model-informed precision dosing of vancomycin in severely ill neonates and children in Belgium (the BENEFICIAL trial): a multicentre, randomised controlled trial. Lancet Child Adolesc Health 2026. DOI: 10.1016/S2352-4642(25)00385-2  Gwee A, et al. Identifying a therapeutic target for vancomycin against staphylococci in young infants. J Antimicrob Chemother 2022. DOI: 10.1093/jac/dkab469Haiping X, et al. Model-informed precision dosing of vancomycin in children 3 months to 18 years of age using Australia-wide data. Antimicrob Agents Chemother 2026. DOI: 10.1128/aac.01840-25  Harwood K, et al. Refining Absolute GFR Estimation in Children: Development and Validation of the GROW-GFR Equation. Front Pharmacol 2026. Minichmayr IK, et al. Model-informed precision dosing: State of the art and future perspectives. Adv Drug Deliv Rev 2024. DOI: 10.1016/j.addr.2024.115421Starp J, et al. Towards model-informed precision dosing of intravenous linezolid: a multicentre external evaluation of pharmacokinetic models in critically ill adults. CMI 2026. DOI: 10.1016/j.cmi.2025.08.032Wilkins AL, et al. Individualized vancomycin dosing in infants: prospective evaluation of an online dose calculator. Int J Antimicrob Agents 2023. DOI: 10.1016/j.ijantimicag.2023.106728 US National Kidney Foundation for different equations for kidney function estimates as related to drug dosing – estimating kidney function for medication related decisions: Frequently Asked Questions | National Kidney Foundation

This week Bruce takes a deep dive into Pascal's Wager. If there is even a small chance that God exists, and believing could lead to an infinite reward while disbelief risks an infinite loss, then believing is the rational choice... right?Blaise Pascal was one of the founders of probability theory back in the 17th century. Even aside from the potentially problematic theological point his famous wager makes, it also raises important questions about probability theory, Bayes' theorem, and even modern Bayesian epistemology. How do we handle situations where the potential payoff is enormous—or even infinite?Pascal's Wager is also the thread we pull that unravels Bayesian Epistemology, while also showing why Bayesian Reasoning is really part of Critical Rationalism.⁠⁠⁠⁠⁠⁠⁠⁠⁠Support us on Patreon⁠⁠⁠⁠⁠⁠⁠⁠

Hi everyone Dexmedetomidine, is both a difficult word to pronounce correctly in casual conversation but also a useful pharmacological tool in many different situations. This week I am joined by one of our fellows Dr Han Lu to discuss dexmedetomidine, it’s history, it’s basic pharmacology and then more specifically it’s use in different clinical situations where it may be useful in obstetrics and gynaecology. Thanks Han for a great Tuesday talk and a very informative podcast! References Han’s powerpoint talk Effectiveness of dexmedetomidine on patient-centred outcomes in surgical patients: a systematic review and Bayesian meta-analysis. BJA September 2024 Crowe. G, et al Perioperative Applications of Dexmedetomidine. ATOTW 469 — Perioperative Applications of Dexmedetomidine April 2022 Douglas MS, Soloniuk LJ, Jones J, Derderian R, Baker C, Stier G. Intravenous dexmedetomidine use in obstetric anesthesia: a focused review. Int J Obstet Anesth. 2025 May;62:104345. doi: 10.1016/j.ijoa.2025.104345. Epub 2025 Feb 13. PMID: 40090158 Atipamezole – alpha 2 agonist reversal agent used in veterinary practice

Jordi Visser is a veteran macro investor with 30+ years of experience and the author of the VisserLabs Substack. In this conversation, we break down why the AI trade already bottomed, how bitcoin fits into the future of markets, and why companies like Figure Technologies are outgrowing legacy banks like JPMorgan. We also discuss inflation, the AI IPO wave with Anthropic and OpenAI, and how AI is reshaping everyday life — from raising kids to running a household.====================Arch Public is an agentic trading platform that automates investment strategies across Stocks, Commodities, ETFs and Crypto. Whether you're rotating into AI & Gold, allocating to the S&P 500, or accumulating Bitcoin, Arch Public executes your plan 24/7 without ever taking custody of your assets or funds. Sign up today at https://www.archpublic.com, and start your FREE automated trading strategy! ====================GalaxyOne is a financial technology platform built for people who want their cash working harder. Open an account with promo code POMP and deposit $10,000 to earn a $3,000 bonus. See site for promotion details → https://go.galaxy.app/HMiq/p57n69yy Galaxy Premium Yield is an investment note issued by Galaxy Digital LP and guaranteed by Galaxy Digital Holdings LP. It is not a bank deposit, is unsecured, and is not FDIC or SIPC insured. U.S. accredited investors only. Cash deposits held at Cross River Bank, Member FDIC. Securities products are not FDIC insured, not bank guaranteed, and may lose value. GalaxyOne Crypto is not FDIC or SIPC insured. Terms apply.====================0:00 - Intro0:50- Why stocks keep climbing despite AI fears2:41 - Jordy's AI portfolio & calling market bottoms7:05 - Inflation report & the Bayesian mindset11:27 - Figure Technologies & the tokenization boom19:47 - Why JPMorgan and legacy banks are at risk24:18 - Bitcoin is the "S&P 500 of ten years from now"26:14 - Anthropic, OpenAI & the AI IPO wave31:25 - AI goes mainstream: his wife & raising kids in the AI era42:43 - ICE deportations & the inflation debate47:47 - Politics, protests & the AI data center fight55:50 - Jordi's weekly videos & how you can help him

Summary In this episode, Andy sits down with Owen Fitzpatrick, a psychologist, speaker, and author of Inner Propaganda: Leading Hearts and Minds through Turbulent Times. Owen has spent close to 30 years studying how beliefs form and change, interviewing people everywhere from North Korea to Rwanda to Afghanistan. His thesis is unsettling: our brains do not simply take in facts and reach objective conclusions. They build a story we then experience as reality. Owen and Andy work through what that means on real projects. You'll hear how a warning from a colleague can quietly harden into a conviction about a teammate, and how Bayesian reasoning gives you a way out. You'll learn Owen's five types of truth, how to tell courageous conviction from dangerous denial, and what leaders can actually make stable when they cannot promise a stable outcome. Owen also explains why pushing harder for buy-in is often the very reason people resist, and how an antifragile identity helps teams face uncertainty like AI without denial or panic. If you're looking for a fresh way to think about belief, influence, and leading through turbulent times, this episode is for you! Sound Bites "We all live in that world where we think we're the one person that isn't the victim of propaganda, and my point or my thesis is we're all victims of our own inner propaganda." "We're not necessarily just convinced by others. We convince ourselves." "Because I think when we say, 'I'm no good at something,' we lock ourselves into it." "Well, if you're not great at communicating with people, get great." "Most of the time our beliefs just create a sort of a reality for us, and that reality can help us or harm us." "So our brains are prediction machines." "Whenever we talk about belief, believing in your ability to succeed in the future is critical if you want to succeed in the future, but that doesn't mean you deny the present." "It, it's not that we think negatively, it's we believe negatively, and we see the world through those lenses." "So I think we want to be able to challenge our beliefs and build a bit more and get more comfortable with uncertainty away from the table, but when we're at the table with our team, that's when we bring the certainty." "We like the idea that we are making this decision of our own free will." "And when you look at what a belief is, a belief is an idea we feel certain about, and that word feel is the most important word of that sentence." "It's okay to believe less in certain things. It's okay to believe more in certain things. It's okay to believe better." Chapters 00:00 Introduction 02:29 Start of Interview 02:40 The Belief About Himself Owen Took Too Long to Update 05:56 A Sweaty Debate Speech and What Came After 08:28 Not Good at Something Is a Skill Gap, Not an Identity 09:45 The Belief Growth Mindset 12:55 The Sandra Story: When a Warning Becomes a Conviction 14:07 Why the Brain Craves Cognitive Closure 17:15 Using Bayesian Reasoning to Loosen a Belief 21:00 Melanie Perkins and Alan Mulally: Conviction or Denial? 21:36 The Five Types of Truth 26:24 Just Because It Is Your Truth Does Not Make It True 28:41 Where Objective Truth Actually Belongs 31:47 What "Leadership Is Propaganda" Does Not Mean 34:45 Why Change Management and AI Adoption Stall 36:30 Why Owen Chose Propaganda Over Self-Talk 37:57 Do I Believe It Because It Makes Me Look Good? 38:49 Answering the AI Question Without Denial or Hype 40:15 Building an Antifragile Identity 44:27 Certainty Is Contagious, and When to Change Course 45:20 What Leaders Can Make Stable in a Volatile Period 47:45 Control What You Can, Influence What You Can, Accept the Rest 51:01 When Pushing for Buy-In Creates the Resistance 52:45 You're Asking the Wrong Question About Persuasion 54:30 Reading the Person Before You Make the Ask 57:30 Helping Kids Hold Strong Beliefs Without Contempt 1:01:46 End of Interview 1:02:15 Andy Comments After the Interview 1:05:09 Outtakes Learn More You can learn more about Owen and his work at InnerPropaganda.com. For more learning on this topic, check out: Episode 370 with Chantel Prat. One of the smartest, clearest, and funniest books on the brain, and why all of it matters for how you lead. Episodes 59 and 60 with Cathy Davidson. She explains how the brain science of attention changes everything. Episode 32 with Brad Kolar. A look at the direct implications of neuroscience for leadership. Owen's TED Talk, available here on YouTube. Chat with PMeLa You can chat directly with PMeLa, the podcast's AI persona, to get episode recommendations and answers to your project management and leadership questions. Visit PeopleAndProjectsPodcast.com/PMeLa to chat with her. Join Us for LEAD52 I know you want to be a more confident leader–that's why you listen to this podcast. LEAD52 is a global community of people like you who are committed to transforming their ability to lead and deliver. It's 52 weeks of leadership learning, delivered right to your inbox, taking less than 5 minutes a week. And it's all for free. Learn more and sign up at GetLEAD52.com. Thanks! Thank you for joining me for this episode of The People and Projects Podcast! Talent Triangle: Power Skills Topics: Leadership, Belief, Persuasion, Influence, Uncertainty, Change Management, Growth Mindset, Psychological Reactance, Buy-In, Artificial Intelligence, Resilience, Project Management The following music was used for this episode: Music: The Fantastical Ferret by Tim Kulig License (CC BY 4.0): https://filmmusic.io/standard-license Music: Funny by Frank Schroeter License (CC BY 4.0): https://filmmusic.io/standard-license

Today's clip is from episode 160, featuring Vaden Masrani. In this conversation, Vaden explores the tension between Bayesian statistics and Bayesian epistemology, and why he sees them as fundamentally different.He explains why Bayesian epistemology can run into problems when trying to explain where hypotheses themselves come from, and argues that an emphasis on finding supporting evidence can encourage confirmation bias rather than genuine scientific inquiry. He also discusses Hempel's paradox, Popper's idea of falsification, and why these philosophical problems don't necessarily undermine Bayesian statistics itself.Full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free): Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

All Else Equal is taking a short break over the summer and will be revisiting some of our favorite conversations from last season. The show will be back with new episodes soon.  Whether it be in politics, public health, or corporate finance, why are people more likely to interpret facts or data in a way that fits their preconceived notions about the world as opposed to searching for the fundamental truth?   A new paper from the Harvard Business School called, Sharing Models to Interpret Data (by Joshua Schwartzstein and Adi Sunderam) studies the propensity for people to adopt interpretations to data based on their community's beliefs, and why this can lead to less accurate conclusions. Hosts and finance professors Jonathan Berk and Jules van Binsbergen are joined by the paper's co-author  Adi Sunderam, who is a professor of corporate finance at Harvard Business School, a research associate at the National Bureau of Economic Research, and a co-editor of the Journal of Finance.  The conversation covers the complexity of Bayesian updating and how the process is improperly deployed in today's thinking, not only in corporate decision-making but also on a sociological level. They also discuss Sunderam's model for explaining how people interpret data, why people are more likely to fall into group-belief dynamics, and if there are any interventions that would lead to better decision-making.   Read Adi Sunderam and Joshua Schwartzstein's paper:  Sharing Models to Interpret Data   Find All Else Equal on the web:   https://lauder.wharton.upenn.edu/allelse/   All Else Equal: Making Better Decisions Podcast is a production of the UPenn Wharton Lauder Institute through University FM.  Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Scientific Sense ® by Gill Eapen: Prof. Harald Uhlig is Professor of Economics at the University of Chicago. His research focuses on macroeconomics, monetary economics, financial markets, and Bayesian time series analysis, particularly at the intersection of macroeconomics and financial economics. Please subscribe to this channel:https://www.youtube.com/c/ScientificSense?sub_confirmation=1

This week on the Excess Returns Weekly Wrap, Jack Forehand and Matt Zeigler break down the AI capital spending boom, the risk that data center investment is crowding out housing and other parts of the economy, and what that means for markets. Featuring Richard Bernstein, David Rosenberg, Tian Yang, and Brent Donnelly, the episode covers AI CapEx, GDP growth, inflation, the K-shaped economy, AI ROI, and why rationality and Bayesian thinking matter more than raw intelligence for investors and traders.Topics coveredWhy the AI and data center boom may be misallocating capital away from housing and infrastructureWhat the dot-com bubble taught Richard Bernstein about investing where capital is scarceWhy AI related spending is approaching half of business CapEx while ex-AI investment is shrinkingHow today's K-shaped economy differs from the broad economic boom of the late 1990sThe difference between AI's contribution to GDP growth and its share of total GDPTian Yang's Kalecki-Levy framework for understanding spending, savings, income, and economic resilienceWhy a pullback in hyperscaler CapEx could weaken the spending and income loopWhy AI return on investment is so difficult to measure and how the profit pool could broaden beyond hardwareBrent Donnelly on why rationality and flexibility matter more than credentials or raw intelligenceWhy persistent bearishness can become a major investing mistakeHow Bayesian thinking, position sizing, and changing your mind help investors stay in the gameTimestamps00:02 Rich Bernstein and David Rosenberg reunite and this week's lineup04:10 The dot-com lesson: what happens when capital floods one sector08:15 AI CapEx, inflation, and why today's economy is different from the 1990s13:58 Kalecki-Levy: how spending and savings are keeping growth resilient18:03 AI CapEx concentration, productivity, and the uncertainty around ROI22:21 Brent Donnelly on why rationality beats intelligence26:21 Strong opinions, flexibility, and Bayesian thinking30:33 What traders and market makers can teach long-term investorsLearn more about the Excess Returns podcast network:https://excessreturns.coNo information discussed in this podcast should be construed as investment advice. Securities discussed may be held by the hosts and guests, their firms or their clients.

Your marketing engine is generating leads, but how do you know which ones are actually worth handing to sales? That is the question at the heart of this episode, and it is one that gets harder to answer the bigger your funnel gets.My guest is Nick Panayi, a technology marketing veteran who most recently led marketing at Inovalon. If you listened to our episode with Nick last year, you know he has spent his career at the intersection of marketing and AI. This time he is back to walk me through something he built with his partner Chris Marin: an AI-powered lead scoring system that moved his team's MQL-to-opportunity conversion rate from 10 percent to 17 percent in a single quarter. That is a 70 percent improvement in lead conversion.We get into why traditional lead scoring has not really changed since marketing automation platforms first showed up, and why that single-dimensional approach creates false positives and quietly erodes trust between marketing and sales. Nick breaks down how his team combined ten different signals, from individual behavior to ICP alignment to lookalike modeling, into one dynamic score built on a self-improving Bayesian model.We also talk about the part I think matters most for actually getting sales to act: pairing every score with a plain-language, AI-generated synopsis that tells the rep exactly why the lead matters and how to open the conversation.If you are a CMO or VP of marketing trying to get sales to actually believe in your leads, this episode gives you a real blueprint, along with Nick's honest take on what it takes to get started even if you do not have a full AI team in-house.Key Topics Covered"(00:00)" Welcome and introduction"(05:00)" Nick's background "(06:00)" Why traditional lead scoring has not changed in years "(09:00)" The specific problems with lead scoring"(12:00)" Multivariate AI platform scoring"(15:00)" Turning complex scoring data into a single AI generated synopsis "(17:00)" Why recency and lead quality together drive better conversion"(19:00)" How the project actually got built"(21:00)" The result"(23:00)" A real example of a counterintuitive lead score"(25:00)" The value of being able to reconstruct a full deal story"(27:00)" Nick's one piece of advice for a CMO who wants to get started next week"(31:00)" Adam's five key takeaways from the conversationIf you are interested in discussing this or any other topic, let's have a chat.  Reach out to me directly to schedule a no-obligation discussion. This isn't a sales call, but rather an opportunity to talk through your questions and challenges.Follow me on LinkedIn.Subscribe to The Healthtech Marketing Show on Spotify or watch us on YouTube for more insights into marketing, AI, ABM, buyer journeys, and beyond!Find all of our episodes on the Health Podcast Library.Thank you to our presenting sponsor, HealthcareNOW, 24/7 expert shows, interviews, and podcasts, powering healthcare leaders with innovation, policy, and strategy insights.

Today's clip is from episode 160, featuring Vaden Masrani. In this conversation, Vaden lays out a sharp critique of Bayesian epistemology - the roughly hundred-year-old philosophical tradition, popular in some Oxford-adjacent circles, that treats subjective probability estimates as legitimate even when there's no data behind them.Vaden's core objection: doing Bayes' theorem on numbers you made up in your head is like fitting a regression line to an empty scatter plot - the math looks rigorous, but there's nothing underneath it. He argues this "math-washing" can trick people into thinking a decision is well-informed simply because it's dressed up in probability language, when frequentists and data-driven Bayesians alike would say the same thing: no data, no model.Get the full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free): Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

******Support the channel******Patreon: https://www.patreon.com/thedissenterPayPal: paypal.me/thedissenterPayPal Subscription 1 Dollar: https://tinyurl.com/yb3acuuyPayPal Subscription 3 Dollars: https://tinyurl.com/ybn6bg9lPayPal Subscription 5 Dollars: https://tinyurl.com/ycmr9gpzPayPal Subscription 10 Dollars: https://tinyurl.com/y9r3fc9mPayPal Subscription 20 Dollars: https://tinyurl.com/y95uvkao ******Follow me on******Website: https://www.thedissenter.net/The Dissenter Goodreads list: https://shorturl.at/7BMoBFacebook: https://www.facebook.com/thedissenteryt/Twitter: https://x.com/TheDissenterYT This show is sponsored by Enlites, Learning & Development done differently. Check the website here: http://enlites.com/ Dr. Wataru Toyokawa is Unit Leader at the Computational Group Dynamics (COGNAC) Collaboration Unit at RIKEN CBS, Tokyo, and Visiting Scientist at Center for Advanced Study of Collective Behaviour at the University of Constance. His research focuses on the computational underpinnings and eco-evolutionary implications of human social learning and their relationships with group decision-making and collective behavior. Using computational modelling and Bayesian statistical methods, coupled with online real-time behavioral experimentation, he is quantitatively approaching human social behavior and group dynamics. He also uses mathematical models to study population dynamics and evolutionary games. In this episode, we start by talking about social learning: what it is, and how it is studied. We discuss why we learn from others, when we copy others, and how we learn from others despite our individual differences. We then talk about collective decision-making, and discuss the madness and wisdom of crowds, and the social learning strategies that regulate the wisdom and madness of interactive crowds.--A HUGE THANK YOU TO MY PATRONS/SUPPORTERS: PER HELGE LARSEN, BERNARDO SEIXAS, ADAM KESSEL, MATTHEW WHITINGBIRD, ARNAUD WOLFF, TIM HOLLOSY, HENRIK AHLENIUS, ROBERT WINDHAGER, RUI INACIO, ZOOP, MARCO NEVES, COLIN HOLBROOK, PHIL KAVANAGH, SAMUEL ANDREEFF, FRANCIS FORDE, TIAGO NUNES, FERGAL CUSSEN, HAL HERZOG, NUNO MACHADO, JONATHAN LEIBRANT, JOÃO LINHARES, STANTON T, SAMUEL CORREA, ERIK HAINES, MARK SMITH, JOÃO EIRA, TOM HUMMEL, SARDUS FRANCE, DAVID SLOAN WILSON, YACILA DEZA-ARAUJO, ROMAIN ROCH, YANICK PUNTER, CHARLOTTE BLEASE, NICOLE BARBARO, PAWEL OSTASZEWSKI, NELLEKE BAK, GUY MADISON, GARY G HELLMANN, SAIMA AFZAL, ADRIAN JAEGGI, JOÃO BARBOSA, JULIAN PRICE, HEDIN BRØNNER, FRANCA BORTOLOTTI, GABRIEL PONS CORTÈS, URSULA LITZCKE, SCOTT, ZACHARY FISH, TIM DUFFY, SUNNY SMITH, JON WISMAN, WILLIAM BUCKNER, LUKE GLOWACKI, GEORGIOS THEOPHANOUS, CHRIS WILLIAMSON, PETER WOLOSZYN, DAVID WILLIAMS, DIOGO COSTA, ALEX CHAU, CORALIE CHEVALLIER, BANGALORE ATHEISTS, LARRY D. LEE JR., OLD HERRINGBONE, DAN SPERBER, ROBERT GRESSIS, JEFF MCMAHAN, JAKE ZUEHL, MARK CAMPBELL, TOMAS DAUBNER, LUKE NISSEN, KIMBERLY JOHNSON, JESSICA NOWICKI, LINDA BRANDIN, VALENTIN STEINMANN, ALEXANDER HUBBARD, BR, JONAS HERTNER, URSULA GOODENOUGH, DAVID PINSOF, SEAN NELSON, MIKE LAVIGNE, JOS KNECHT, LUCY, MANVIR SINGH, PETRA WEIMANN, CAROLA FEEST, MAURO JÚNIOR, TONY BARRETT, NIKOLAI VISHNEVSKY, STEVEN GANGESTAD, TED FARRIS, HUGO B., JORDAN MANSFIELD, CHARLOTTE ALLEN, PETER STOYKO, DAVID TONNER, LEE BECK, PATRICK DALTON-HOLMES, NICK KRASNEY, RACHEL ZAK, DENNIS XAVIER, CHINMAYA BHAT, RHYS, ALEX MACLEOD, HAIDAR, JULIEN PORCHER, ROBERT SUNDSTRÖM, JON STEWART, AND JENNY M!A SPECIAL THANKS TO MY PRODUCERS, YZAR WEHBE, JIM FRANK, ŁUKASZ STAFINIAK, TOM VANEGDOM, BERNARD HUGUENEY, CURTIS DIXON, THOMAS TRUMBLE, KATHRINE AND PATRICK TOBIN, JONCARLO MONTENEGRO, NICK GOLDEN, CHRISTINE GLASS, IGOR NIKIFOROVSKI, PER KRAULIS, ADAM HUNT, ANTHONY DI LORENZO, AND JOÃO BARBOSA!AND TO MY EXECUTIVE PRODUCERS, MATTHEW LAVENDER,SERGIU CODREANU, AND GREGORY HASTINGS!

Brent Donnelly joins Matt Zeigler to explain how professional traders build a durable edge through risk management, trading psychology, probabilistic thinking, and creative market analysis.Drawing from his new book, Trade Outside the Box: Advanced Thinking for Professional Traders, Brent breaks down why trading strategies decay, why rationality beats intelligence, how to avoid risk of ruin, and how lessons from poker, behavioral finance, and real-world experience can improve decision-making.Trade Outside the Box: Advanced Thinking for Professional Tradershttps://amzn.to/4h9bi3eBrent Donnelly on Xhttps://x.com/donnelly_brentSpectra Marketshttps://www.spectramarkets.comTopics covered:Why fundamentals, technical analysis, behavioral finance, and quantitative methods are necessary but not sufficient for trading successHow traders can develop an edge by connecting markets to poker, psychology, biology, auto racing, and video gamesWhy profitable trading strategies decay as more investors discover and copy themHow changing volatility regimes force traders to adapt their style and avoid becoming a one-trick ponyWhy mismatching a long-term investment thesis with a short-term stop loss can destroy a good ideaHow trading journals and P&L data help separate normal variance from a broken processWhy the house money effect can make traders more reckless after large gainsWhy rationality, flexibility, and expected value matter more than credentials or raw intelligenceHow Bayesian thinking helps traders update probabilities and fight confirmation biasThe difference between independent thinking and blind contrarianismWhy avoiding risk of ruin, protecting family and health, and defining success beyond money are essential to a sustainable trading careerTimestamps:00:00 Introduction to Brent Donnelly and Trade Outside the Box04:00 Why smart analysts often produce fully priced trade ideas08:00 Poker discipline and avoiding boredom trades12:00 How lead-lag correlation trading lost its edge16:35 Matching a trade's stop loss to its time horizon21:00 What trading data reveals about win rates and expected value25:00 The house money effect and the danger of overearning29:00 Why rational traders beat smarter traders33:00 Strong opinions weakly held and Bayesian updating37:00 Curating a balanced diet of bullish and bearish information41:00 Using creativity and outside disciplines to find market edge45:11 Avoiding risk of ruin and the lessons of Jesse Livermore50:29 The Serenity Prayer and focusing on what traders can control55:00 Choosing family and health over markets59:00 Why your first thought may not be your ownLearn more about the Excess Returns podcast network:https://excessreturns.coNo information discussed in this podcast should be construed as investment advice. Securities discussed may be held by the hosts and guests, their firms, or their clients.

Today's clip is from episode 162, featuring Chris Krapu. In this conversation, Chris explains why Bayesian thinking remains surprisingly valuable in today's AI landscape - even when the models themselves aren't explicitly Bayesian.Rather than uncertainty estimation, Chris highlights a different advantage: Bayesian training provides a deep intuition for concepts like priors, sampling, rejection sampling, and high-dimensional geometry, making it much easier to understand and apply modern AI research. He also discusses why Bayesian methods are becoming increasingly relevant for evaluating agentic AI systems, where complex workflows and limited evaluation data make hierarchical models and sensible priors especially powerful.Get the full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free): Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work!

Two theoretical physicists helped design landmark surveys of 1,600+ scientists on the deepest unsolved questions in physics — and found almost no consensus at all. From string theory's shockingly low support to physicists admitting their models run on "belief," this episode exposes the faith hiding inside science. Adam Frank (University of Rochester astrophysicist and astrobiologist) and Niayesh Afshordi (Perimeter Institute / University of Waterloo cosmologist, co-author of the APS Physics Magazine "Big Mysteries" survey) join Brian to unpack what physicists actually believe versus what they can prove. We cover: - Why string theory pulled a shockingly small share of the vote against loop quantum gravity - What a Bayesian "prior" reveals about every scientist's hidden beliefs - Why one branch of physics quietly became unfalsifiable - How sociology and "tastemakers" can hijack scientific consensus - Why AI might become the field's unlikely savior. There is no sane statistical analysis that doesn't have a prior — that's your belief. Timestamps: 00:00 – Why scientists owe the public real answers 04:00 – Splitting time between research and outreach 07:55 – The survey that "rankled" Brian Keating 09:18 – Are physicists secretly just like Spock? 12:05 – Are we living in an anti-scientific age? 16:56 – Why most scientists refuse to go public 21:36 – Should "belief" ever appear in a survey? 23:12 – Is string theory really 21st-century physics? 26:20 – When sociology hijacks scientific consensus 28:39 – The hidden "prior" behind every experiment 33:01 – String theory's shockingly low vote count 39:03 – Four levels of belief, from data to faith 41:39 – Inside the 1,675-physicist mystery survey 46:12 – Kingmakers, tastemakers, and physics cliques 54:12 – Is advanced tech indistinguishable from magic? 59:23 – Could AI become physics' long-awaited savior? 1:01:47 – What's next for Adam and Niayesh ———

Support & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome workTakeaways:Q: How does putting a Gaussian process on unknown coordinates fix noisy location data in mineral prospecting?A: In mining and geostatistics, the classic Gaussian process model, known there as kriging, assumes you know exactly where each sample was taken. Chris' project broke that assumption on purpose: the recorded coordinates for each core sample were only accurate to within a rough radius. By treating the true locations as latent variables and putting a Gaussian process over them jointly with the measurements, the model could still reconstruct the underlying gold-concentration field, even though the exact sampling locations were never known precisely. It's a demonstration that Gaussian processes can absorb structural uncertainty that looks, at first glance, like it should make the problem impossible.Q: What is "Poverty Bayes," and what did it cost to train a two-million-parameter Bayesian model?A: Poverty Bayes was Chris' experiment in seeing how cheaply a large Bayesian model could be trained using modern cloud infrastructure. He fit a hierarchical logistic regression with close to two million parameters, using PyMC's Hamiltonian Monte Carlo on a single A100 GPU rented through Modal, a serverless platform that deploys a Python script straight to GPU hardware with almost no setup. He'd originally guessed it would cost around five dollars, the price of a Big Mac, but the real bill came in an order of magnitude lower. A model that would take a Gibbs sampler weeks to run, and that once required a research lab's dedicated GPU, now costs pocket change and a few minutes of setup.Q: What's the current bottleneck in Bayesian-at-scale tooling?A: Chris argues the software has largely caught up: PyMC's JAX backend and NumPyro make GPU-accelerated Bayesian modeling work out of the box for most problems. What's missing is common knowledge. Companies are clearly running large Bayesian models in production, but the results stay behind corporate firewalls. Chris' proposal is a community benchmark effort: which frameworks handle a million-parameter Markov random field on a given GPU out of the box, since this kind of expensive, slow-running benchmark is a poor fit for standard CI pipelines but valuable for the field to know.Chapters:22:57 When does GPU acceleration actually pay off for a Bayesian model?26:33 What did it cost to train a two-million-parameter model on Modal?30:36 What happened when Chris asked 200 different LLMs to flip a coin?34:50 Where do Bayesian ideas show up in the agentic AI systems Chris builds at Nvidia?40:16 Are statisticians being made obsolete by large language models?41:19 How does putting a Gaussian process on unknown coordinates fix noisy data in mineral prospecting?58:05 What is Chris looking forward to working on next?Thank you to my Patrons for making this episode possible!Links from the show here

Today's clip is from episode 161, featuring Luigi Acerbi. In this conversation, Luigi explains one of the biggest engineering bottlenecks facing transformer-based probabilistic models—and how his group found a way around it.The core challenge is that many inference models treat data as an unordered set, making them naturally permutation invariant. That's statistically elegant, but computationally painful: every time a new data point arrives, the model has to recompute attention over the entire dataset from scratch, preventing the kind of KV caching that makes modern language models so efficient.Luigi walks through his team's solution: a hybrid architecture that keeps the original context fully set-based while introducing a causal-attention buffer for newly arriving data. The result is dramatically faster inference- up to 100× faster in some settings - opening the door to applications like reinforcement learning, active data acquisition, and, ultimately, Luigi's long-term vision of a foundation model for Bayesian inference.Get the full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work

Can you find a real edge in prediction markets? Simon hunts the arbitrage — and calls Netflix cheap. In this week's WorldWideMarkets: •

We've spent a long time bashing Bayesian epistemology on the podcast. Unfortunately, this is sometimes taken to mean that we bash all things Bayesian. But our God-fearing listeners will be pleased to learn that we do not bash all things that the Reverend Thomas Bayes stood for. Bayesian statistics, when appropriately separated from epistemology, has many things to teach us. And instead of taking our word for it, we're bringing on an expert in this field: Alex Andorra, a research scientist at Meta and host of the Learning Bayesian Statistics podcast. The Gods Smile. Check out Alex's website and the Learning Bayesian statistics podcast. We discuss 538 and election forecasting Sports analytics and why Messi is objectively the best What is Bayesian statistics? What is the difference between Bayesian and frequentist statistics? What is the allure of Bayesian statistics? Are priors best thought of as beliefs? How do critical rationalists make decisions? References Learning Bayesian Statistics: https://learnbayesstats.com/ Silver Bulletin: https://www.natesilver.net/ Vaden's blog post on decision making: https://vmasrani.github.io/blog/2020/vaden_second_response/#what-other-options-to-decision-theory-are-there # Socials Follow us on Twitter at @alex_andorra, @IncrementsPod, @BennyChugg, @VadenMasrani Come join our discord server! DM us on twitter or send us an email to get a supersecret link Become a patreon subscriber here. Or give us one-time cash donations to help cover our lack of cash donations here. Click dem like buttons on youtube Send us an email because you have - no, because you want to - over at incrementspodcast@gmail.com.Special Guest: Alex Andorra.

Support & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome workTakeaways:Q: What is Variational Bayesian Monte Carlo (VBMC) and how is it different from Bayesian optimization?A: VBMC borrows the machinery of Bayesian optimization but aims at a different target. Bayesian optimization fits a Gaussian process surrogate to an expensive function and uses it to hunt for the optimum. VBMC instead treats the log-posterior as the function to model, evaluates it at a few carefully chosen points, and keeps the whole reconstructed shape rather than just its peak. That gives you the full posterior, not a single best-fit value. Where MCMC might need tens of thousands to millions of evaluations, VBMC often reconstructs a good posterior approximation from a few hundred, which matters when each evaluation is slow.Q: When should you reach for PyVBMC, and when is it the wrong tool?A: Two symptoms tell you PyVBMC might help. First, speed: if a single evaluation of your log density takes on the order of a second, running MCMC over tens of thousands of evaluations becomes painful, and PyVBMC's few-hundred-evaluation budget pays off. Second, dimensionality: because it leans on a Gaussian process surrogate, it works well up to roughly 10 to 15 parameters and degrades beyond that. If your model already runs fine in Stan or PyMC, you do not need it. It shines for expensive, low-dimensional models common in science and engineering, where you are modeling a process rather than composing nice distributions.Full takeaways hereChapters:00:18:13 What is Variational Bayesian Monte Carlo (VBMC) and how does it differ from Bayesian optimization?00:30:21 When should you use VBMC versus BADS in practice?00:31:20 What is Bayesian Adaptive Direct Search (BADS) and how does its hybrid optimization strategy work?00:39:18 What are neural processes, and why are transformers a natural neural process architecture?00:45:54 What is the Amortized Conditioning Engine (ACE) and what problem does it unify?00:55:42 What do PriorGuide and the new autoregressive buffer paper solve for amortized inference?01:02:03 How does the new autoregressive buffer speed up predictions in transformer probabilistic models?01:06:11 What is Luigi Acerbi's vision for a foundation model for inference?01:09:26 What is ALINE and how does it add active data acquisition to amortized inference?01:12:43 How does Luigi Acerbi connect LLM agents, Bayesian decision theory, and the nature of intelligence?01:18:44 For a PyMC, Stan, or NumPyro user, where should you start with VBMC, BADS, or BayesFlow?Thank you to my Patrons for making this episode possible!Links from the show here

Sam Harris speaks with Siddhartha Mukherjee about the science of cancer. They discuss the updated edition of The Emperor of All Maladies, whether cancer is one disease or many, why prevention is so hard, inflammation and air pollution as carcinogens, the myth that cell phones cause cancer, liquid biopsies and Bayesian reasoning, immunotherapy and CAR T cells, drug pricing, the promise of AI in drug discovery, the state of American medical science, and other topics. If the Making Sense podcast logo in your player is BLACK, you can SUBSCRIBE to gain access to all full-length episodes at samharris.org/subscribe.

Send us Fan MailIn this Journal Club, Ben and Daphna dig into two new papers on PDA management in our smallest patients. First, the SMART-PDA pilot RCT from Souvik Mitra and colleagues, which uses comprehensive hemodynamic screening to selectively treat high-volume shunts in infants born before 26 weeks, and whose striking Bayesian signal for reduced pulmonary hemorrhage and NEC stopped the trial early. Then a companion JAMA Network Open comparative effectiveness study across four pharmacotherapy regimens. Along the way, Ben shares hemodynamics pearls from his Montreal training: why left ventricular output, LA:Ao ratio, and transductal velocity matter more than PDA diameter alone.----Selective early medical treatment of the patent ductus arteriosus in extremely low gestational age infants: a pilot randomised controlled trial (SMART-PDA). Mitra S, Hebert A, Castaldo MP, Disher T, El-Naggar W, Dhillon S, Alhassen Z, Koo J, Katheria AC, Hyderi A, Kumaran K, Ting J, Surak A, Larocque J, Pepper D, Hornberger L, Makoni M, Weisz DE, Jain A, Bacchini F, Cameron-Nola AJJ, Hatfield T, Dorling J, McNamara PJ, Thabane L.Arch Dis Child Fetal Neonatal Ed. 2026 May 18:fetalneonatal-2026-330462. doi: 10.1136/archdischild-2026-330462. Online ahead of print.PMID: 42150872Pharmacologic Therapies for Patent Ductus Arteriosus in Extremely Preterm Infants. Mitra S, Jain A, Ting JY, Ben Fadel N, Drolet C, Abou Mehrem A, Soraisham AS, Jasani B, Louis D, Lapointe A, Dorling J, Khurshid F, Hyderi A, Kumaran K, Toye J, Harabor A, Weisz DE, Stavel M, Morin A, Bhattacharya S, Lalitha R, Afifi J, Augustine S, Castaldo MP, Hatfield T, Su YC, Shah PS; Canadian Neonatal Network Investigators.JAMA Netw Open. 2026 Jun 1;9(6):e2617477. doi: 10.1001/jamanetworkopen.2026.17477.PMID: 42262753 Free PMC article.Support the showAs always, feel free to send us questions, comments, or suggestions to our email: nicupodcast@gmail.com. You can also contact the show through Instagram or Twitter, @nicupodcast. Or contact Ben and Daphna directly via their Twitter profiles: @drnicu and @doctordaphnamd. The papers discussed in today's episode are listed and timestamped on the webpage linked below.Enjoy!

In this episode, Professor Paola Cinnella - Professor of Fluid Mechanics at Sorbonne University and Director of the Sorbonne Cluster for Artificial Intelligence (SCAI) - joins Neil to discuss her path from classical fluid mechanics and high-order numerical methods into uncertainty quantification, Bayesian methods, data-driven turbulence modeling and AI for Science.Paola has built a career at the intersection of CFD, compressible and turbulent flows, dense gas dynamics, uncertainty quantification, robust optimization and machine learning. We discuss academic careers, dense gases, RANS uncertainty, AirfRANS, surrogate modeling, scientific publishing, education in the age of AI, and the idea of the "centaur scientist".Key topicsFluid mechanics, CFD and high-order schemesDense gases, real-gas effects and expansion shockwavesUncertainty quantification and Bayesian methodsRANS turbulence-model uncertaintyAirfRANS and CFD datasets for machine learningTurbulence modeling vs surrogate modelingScientific publishing and ML-for-CFD standardsSCAI and AI for ScienceEducation, ChatGPT and centaur scientistsPapersQuantification of model uncertainty in RANS simulations: A review - Heng Xiao, Paola Cinnellahttps://doi.org/10.1016/j.paerosci.2018.10.001Discovery of Algebraic Reynolds-Stress Models Using Sparse Symbolic Regression - Martin Schmelzer, Richard P. Dwight, Paola Cinnellahttps://doi.org/10.1007/s10494-019-00089-xBayesian estimates of parameter variability in the k-epsilon turbulence model - W.N. Edeling, P. Cinnella, R.P. Dwight, H. Bijlhttps://doi.org/10.1016/j.jcp.2013.10.027AirfRANS: High Fidelity Computational Fluid Dynamics Dataset for Approximating Reynolds-Averaged Navier-Stokes Solutionshttps://arxiv.org/abs/2212.07564Data-driven turbulence modeling - Paola Cinnellahttps://arxiv.org/abs/2404.09074Direct numerical simulations of supersonic turbulent channel flows of dense gases - Luca Sciacovelli, Paola Cinnella, Xavier Gloerfelthttps://doi.org/10.1017/jfm.2017.237LinksPaola Cinnella named Director of SCAIhttps://scai.sorbonne-universite.fr/news/paola-cinnella-new-directorSCAIhttps://scai.sorbonne-universite.fr/Paola Cinnella - HAL publicationshttps://cv.hal.science/paola-cinnellaPaola Cinnella - Google Scholarhttps://scholar.google.com/citations?hl=fr&user=wBRA0JAAAAAJERCOFTAC SIG 54 - Machine Learning for Fluid Dynamicshttps://www.ercoftac.org/special_interest_groups/54-machine-learning-for-fluid-dynamics/master-of-science-internships/Chapters00:00 Podcast intro00:39 Introducing Prof. Paola Cinnella03:28 Conversation begins03:56 How Paola found fluid mechanics07:09 Moving from Italy to France08:37 High-order schemes and compressible flows09:30 Building an academic career12:06 Dense gases and uncertainty quantification15:16 Expansion shockwaves and real-gas effects19:17 Returning to Paris and academic mobility24:52 Academia, passion and persistence27:51 Bayesian methods and turbulence uncertainty30:47 Learning statistics across disciplines33:07 LearnFluidS, AirfRANS and CFD datasets36:33 Skepticism and physics in ML turbulence modeling40:41 Could ML lead to a universal turbulence model?42:59 Turbulence models, surrogate models and RANS45:03 Why LES alone cannot solve optimization47:15 Multi-fidelity modeling49:08 What Computers & Fluids looks for in ML-for-CFD papers54:05 CFD metrics vs machine-learning metrics57:13 Overselling, publication pressure and quality62:22 SCAI and AI for Science66:07 Cross-disciplinary AI for Science69:26 Education in the AI era72:44 Critical thinking and AI outputs78:15 AI as a companion, not a replacement81:42 AlphaFold and the future of discovery83:43 Training centaur scientists85:11 Closing thoughts

Aaron Brown is an author and risk management professional, formerly the Chief Risk Officer at the hedge fund AQR. Aaron's recent works are titled Wrong Number: How to Extract Truth From a Blizzard of Quantitative Disinformation and The Poker Face of Wall Street. Greg and Aaron discuss why quantitatively flawed studies still persist today. Aaron argues that the central problem is not just incompetence or conspiracy but a macro phenomenon he calls tribalism, combined with the diffusion of responsibility across authors, reviewers, journals, and journalists. He discusses examples, including an NTSB “Chinatown bus” study, a USAID mortality claim, a Chunnel fire-risk study, an observational marijuana/heart-attack paper, and a study claiming 40% of COVID deaths were caused by evictions and later cited in courts and legislation. They contrast academia's weak incentives with finance and gambling, where betting forces accountability, and Aaron describes the empirical Bayesian approach he prefers using base rates and evidence. *unSILOed Podcast is produced by University FM.* Episode Quotes: There are consequences to publicizing bad research. [34:45] I think most researchers are careful not to let the university press office get ahold of their bad study. They're careful not to go out and give interviews on it. The ones who forget that, they're the ones who cause the problems and get caught. I mean, not many people do get caught, and the consequences of getting caught are pretty low, but it can happen. You lose professional credibility, and that's extremely important, you know? That's really the be all and end all for most researchers I know, is what their peer researchers think of them. And that's where you get hurt. In fact, you get hurt even for getting publicity for your good work, you know? There still is a real feeling in a lot of sciences that the guy in the headline is not a real scientist. Why are people so easily misled by quantitative information? [05:32]  It's been documented over and over in lots of different ways, that most published research findings are false, and yet nobody seems to care.  Aaron discusses the promise and pitfalls of Bayesian reasoning.  [57:28] You don't have to go all the way to Bayesian to know that what they're doing in the journals is wrong. The journal, the frequentist, the Fisher classical hypothesis testing, the gold standard, double-blind control trials—those things are just wrong. And you don't have to go all the way to Bayesianism. You can just say, "Okay, we can just show mathematically that those don't work." Show Links: Recommended Resources: National Transportation Safety Board (NTSB) Tribalism United States Agency for International Development (USAID) Evaluating the impact of two decades of USAID - Lancet Study Channel Tunnel Fires Ronald Fisher Intergovernmental Panel on Climate Change (IPCC) Outlive: The Science and Art of Longevity National Bureau of Economic Research (NBER) Francesca Gino Robin M. Hogarth Harrison White Bayesian Statistics UnSILOed 584: David Zweig - Examining School Closure Policies During the Pandemic Guest Profile: LinkedIn Profile Reason Profile Wikipedia Profile Guest Work: Amazon Author Page Wrong Number: How to Extract Truth From a Blizzard of Quantitative Disinformation Red-Blooded Risk: The Secret History of Wall Street The Poker Face of Wall Street Financial Risk Management For Dummies Fischer Black and the Revolutionary Idea of Finance A World of Chance: Betting on Religion, Games, Wall Street Wrong Number with Aaron Brown YouTube Series Google Scholar Page Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

What if a materials lab on Earth could screen a hundred alloys a day with little input from the scientist? Taylor and Andrew sit down with Joseph Krause, CEO and co-founder of Radical AI, to dig into what it takes to build a self-driving lab and why most of the field is still missing the hard part. From discovering that flagship SEM and XRD instruments ship with no real data access (and rebuilding their entire OS around the workaround), to MATRIX — their multimodal vision-language model that hones in on a target property in roughly 20 experiments — Joseph walks through the technical bets that got them here. He explains why they're using language-model embeddings to teach Bayesian optimization what "28% titanium" actually means, why "scientific intuition" has to be measured as a delta between human and AI annotations, and why Radical is going all the way to manufacturing instead of licensing compositions — because the real IP, and the only training data that matters, lives on the production floor. Check out Radical AI here [LINK] This episode of the Materialism Podcast is sponsored by Momentum Transfer. Visit their website for more details about their measurement services. [LINK] The Materialism Podcast is sponsored by Materials Today, an Elsevier community dedicated to the creation and sharing of materials science knowledge and experience through their peer-reviewed journals, academic conferences, educational webinars, and more. [LINK] Thanks to Kolobyte and Alphabot for letting us use their music in the show! If you have questions or feedback please send us emails at materialism.podcast@gmail.com or connect with us on social media: Instagram, Twitter. Materialism Team: Taylor Sparks, Andrew Falkowski, & Jared Duffy.

Can you even negotiate fraud? This episode on Fraudish we discuss how negotiation and fraud are more related than you would think. I am joined by Martin Medeiros about fraud, negotiation, and AI. Martin is an attorney (I know, another one!) who helps organizations build value by treating intellectual property as a strategic asset.In the episode, Martin describes his career building Fortune 500 tech negotiation teams. We discuss his data-driven negotiation approach using game theory and Bayesian updating. And of course, we discuss using LinkedIn as attorney marketing. Connect with Martin: https://www.linkedin.com/in/martin-medeiros2/Persuasion Lab: https://thepersuasionlab.com/Buckley Law: https://www.buckley-law.com/martin-medeiros/

Support & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome workTakeaways:Q: What's the difference between Bayesian statistics and Bayesian epistemology?A: Bayesian statistics uses Bayes' theorem on actual data: you put a prior over parameters, combine it with a likelihood, and the data is allowed to tell you your model is wrong. Vaden loves it. Bayesian epistemology, in his tongue-in-cheek phrase, is "Bayesian statistics minus the statistics" - taking Bayes' theorem as a general account of how anyone should reason under uncertainty, including about events where there is nothing to count. The first is falsifiable and grounded; the second, he argues, lets people attach authoritative-sounding numbers to pure belief.Q: Why is it a problem to put a probability on a one-off future event like human extinction?A: Because there are no statistics behind it. Vaden's trigger example is Toby Ord's The Precipice, where a data-derived probability (supervolcanoes per millennium) is placed side by side with a probability of extinction-by-superintelligence that came from no data at all. His reaction is the statistician's first instinct: where are the numbers coming from, and what could ever make them come out differently? A subjective degree of belief is fine as a hunch. The trouble starts when it is communicated as though it were an objective, data-grounded frequency.Q: What does Vaden Masrani actually like about Bayesian statistics?A: The freedom to encode domain knowledge as a prior and have the result respect common sense - estimating an average human height, you can rule out zero and a hundred feet before seeing a single measurement. But the part he keeps stressing is falsifiability: you fit the model, compare it to data, and the data can tell you the model was bad. That contact with reality is exactly what makes the statistics legitimate and what the epistemology lacks. On Bayesian-versus-frequentist for engineering problems, he says he has no dog in the fight -- both are useful, and any working statistician uses both.Full takeaways hereChapters:00:24:01 What's the difference between Bayesian statistics and Bayesian epistemology?00:33:12 How can Bayesian epistemology lead to bad real-world decisions?00:36:36 Is Bayesian or frequentist statistics better for real-world problems?00:39:31 What is the problem of induction, and how does Bayesian epistemology try to solve it?00:43:50 What are the main logical problems with Bayesian epistemology?00:48:40 What is Popper's critical rationalism, and how does falsifiability fit in?00:52:31 How does critical rationalism work when you can't run a clean experiment?01:15:03 Why should you treat criticism as a gift, even when it hurts?01:19:54 How do Stoicism and equanimity help you handle criticism?01:23:19 Why does critical rationalism apply to everyday life, not just science?Thank you to my Patrons for making this episode possible!Links from the show here

Thomas Ahle wants Normal Computing to be the Lovable for chip design: type your intent, and a swarm of agents carries it from design through optimisation, formalisation and verification to tape-out. To get there, his team at wrote their own open-source Verilog simulator, 580,000 lines in 43 days, because commercial EDA verifiers run about $10,000 per core and there are no decent open-source compilers to build on.That sets up the question Tim keeps pressing: if an agent can produce a chip design, a proof, or a working program, how do you actually know it is correct? Passing 70% of tests is not the same as being right, and a single fabricated bug can cost a company a fortune. They dig into ProgramBench (rebuild a program from its tests, roughly 0% success), the difference between structure and competence, and the "understanding debt" you take on when nobody reads the code.From there: auto-formalisation in Lean and the AlphaProof trick of training on prove-or-disprove; why there is no single true representation of a spec (Petri nets, TLA+, Erik Curiel's "math does not represent"); and thermodynamic computing, where Normal Computing's CN101 chip is built so that its physical noise *is* the computation, settling a stochastic differential equation in hardware to invert a matrix. Plus Bayesian uncertainty, specialisation, the Chomsky hierarchy, AI slop, and whether performance is all that matters.Recorded in Zurich.Disclosure: Normal Computing paid our production and travel costs for this show. We retained full editorial control. They did not see the video before publication, and we did not show it to them or discuss it with them beforehand.---TIMESTAMPS:00:00:00 Meet Thomas Ahle: the Lovable for chip design00:03:41 Why hardware needs formal verification00:06:36 Ten thousand dollars per core and a six-month agent run00:07:40 Rebuilding programs from tests: ProgramBench and zero percent00:12:15 Structure vs competence: can you learn a program from behavior?00:15:27 Continual learning, abstraction, and Claude as an ecosystem00:23:17 Autoformalization and the AlphaProof trick00:29:31 No single true representation: specs, Petri nets and TLA+00:34:43 Thermodynamic computing: when noise is the computation00:37:32 Bayesian uncertainty in the age of token streams00:41:12 Hybrid compute: vibe-coding loops, binaries and Stockfish00:44:44 Co-design, central-AI apps and API pricing00:49:45 Chain of thoughtlessness and the Chomsky hierarchy00:53:40 AI psychosis, slop and the broken social contract00:57:34 Typing it yourself, teamwork and performance vs competence---REFERENCES:person:[00:00:10] Thomas Ahlehttps://thomasahle.comorganization:[00:00:27] Normal Computinghttps://normalcomputing.com/paper:[00:11:21] ProgramBench: Can Language Models Rebuild Programs From Scratch?https://arxiv.org/abs/2605.03546[00:31:55] Autoformalizing Memory Device Specifications with Agentshttps://arxiv.org/abs/2605.00058[00:35:20] Thermo AI and the Fluctuation Frontierhttps://arxiv.org/abs/2302.06584[00:36:40] Thermo Comp System for AI Applicationshttps://arxiv.org/abs/2312.04836[00:37:05] Thermodynamic Linear Algebrahttps://arxiv.org/abs/2308.05660[00:44:50] An efficient probabilistic hardware architecture for diffusion-like modelshttps://arxiv.org/abs/2510.23972tool:other:[00:01:00] Building an Open-Source Verilog Simulator with AI: 580K Lines in 43 Dayshttps://normalcomputing.com/blog/building-an-open-source-verilog-simulator-with-ai-580k-lines-in-43-days[00:02:55] Normal Computing Announces Tape-Out of the World's First Thermodynamic Computing Chip (CN101)https://www.normalcomputing.com/blog/normal-computing-announces-tape-out-of-worlds-first-thermodynamic-computing-chip[00:32:02] DRAMBench: Autoformalizing DRAM Specifications with Timed Petri Netshttps://www.iese.fraunhofer.de/blog/drambench-autoformalizing-dram-specifications/---ReScript: https://app.rescript.info/share/ff9684a112ab37744096adaeb097a263

Would You Hire Your Own Graduate? What if the true measure of a university wasn't its rankings, campus amenities, or brand recognition—but the quality of the young adults it sends into the world?In this episode of CEO Blindspots®, UATX⁠ President ⁠Dr. Carlos Carvalho discusses a bold question facing higher education today:Are colleges producing graduates who are truly prepared for life, leadership, and work?Dr. Carvalho shares how UATX is challenging conventional assumptions about higher education through merit-based admissions, rigorous intellectual inquiry, open debate, and meaningful real-world experience. We explore why employers increasingly seek graduates who can think critically, communicate effectively, navigate disagreement, and contribute from day one.Whether you're an employer searching for future leaders, a parent evaluating college options, or a student determined to make the most of your education, this conversation offers a fresh perspective on what higher education can be designed to achieve.CEO Blindspots® Podcast Guest: ⁠Dr. Carlos Carvalho⁠Dr. Carlos Carvalho is the President of the ⁠University of Austin⁠. Prior to taking on this role, he spent 15 years as a professor at the University of Texas at Austin's McCombs School of Business, where he held the La Quinta Centennial Professorship and founded the Salem Center for Policy. A native of Brazil, Dr. Carvalho earned his doctorate in statistics from Duke University and has also taught at the University of Chicago Booth School of Business. His research focuses on Bayesian statistics in complex, high-dimensional problems with applications ranging from economics to genetics to public policy. At ⁠UATX⁠, he is leading a bold effort to build a new university that stands for American principles and academic excellence.

    Few careers in military medicine trace an arc as wide as that of CAPT (Ret) Kimberly Elenberg, DNP, RN. In this episode she sits down with WarDocs to map a journey that began as an ROTC cadet who joined because she saw students rappelling down a building in Philadelphia, and that has since carried her from the bedside at Walter Reed Army Medical Center to the role of principal investigator on a Carnegie Mellon University team competing in the DARPA Triage Challenge. Along the way she changed uniforms, disciplines, and altitudes of responsibility, but never lost the thread that ties it all together: people first, and the relationships that make hard things possible.     CAPT (Ret) Elenberg describes how early mentors shaped her. Colonel Graham showed her that putting people first is a practice, not a slogan. Major McGee backed her instinct for innovation, and as a young nurse on Ward 51 she built one of the first patient education centers in a military treatment facility, learned to set up networks and hardware, and pursued nursing informatics before the field was common. She recounts moving to research at NIH, where her work on TPA for clearing central line catheters was later adopted as best clinical practice, and her decision to volunteer as an EMT and medic so she would understand field medicine as well as hospital medicine.       From there the conversation follows her into the U.S. Public Health Service, where after 9/11 the Surgeon General asked her to help build the nation's deployable response teams from concept to operation, training them in real communities facing real crises. She explains how anthrax and zoonotic disease drew public health into agriculture and food security, how her long relationship with Carnegie Mellon's Auton Lab began with a bus trip and a phone call, and how that mathematical grounding in probabilistic modeling resurfaced when she was asked to model the effects of policy during COVID and, later, to track military security assistance flowing to Ukraine.     The episode closes on the present and the future: autonomous triage payloads that can read a casualty's physiological state without touching them, robotic snakes that might pack non-compressible hemorrhage, swarms of drones and ground robots that find the wounded and feed the right information to the right echelon. Throughout, CAPT (Ret) Elenberg returns to her core lessons — trust your chain of command, define what success really looks like, build on small wins, and never limit yourself to your military occupational specialty. From an orphanage and a food-service background to teaching at the National Defense University, hers is a story about doors held open and relationships that endure. Chapters (00:54-07:11) From Rappelling Cadet to Innovating Army Nurse (07:11-16:48) Building the Nation's Public Health Response Teams (16:48-22:24) Biosurveillance Modeling COVID and Ukraine Aid (22:24-32:32) The Power of Relationships Across a Career (32:32-37:37) Autonomy Confidence and Knowing When to Explore (37:37-51:33) The DARPA Triage Challenge and Lessons That Last Chapter Summaries (00:54-07:11) From Rappelling Cadet to Innovating Army Nurse The guest traces her start as an ROTC cadet drawn in by students rappelling down a Philadelphia building, her commissioning as an Army nurse, and her first duty station at Walter Reed Army Medical Center. Early mentors, including Colonel Graham and Major McGee, taught her that people truly come first and backed her instinct for innovation. On Ward 51 she built one of the first patient education centers in a military treatment facility while teaching herself websites, networking, and nursing informatics.   (07:11-16:48) Building the Nation's Public Health Response Teams Her NIH research on TPA for central line catheters was later adopted as best clinical practice, and she volunteered as an EMT and medic to learn field medicine. After moving to the U.S. Public Health Service for family stability, she answered the Surgeon General's call following 9/11 to build the nation's deployable response teams from concept to operation. Anthrax and zoonotic disease pulled public health into agriculture and food security across the federal enterprise.   (16:48-22:24) Biosurveillance Modeling COVID and Ukraine Aid Tasked to advise on detecting events and discerning intent, she leaned into probabilistic modeling and a long relationship with Carnegie Mellon's Auton Lab that began with a bus trip and a phone call. As Director of Population Health at the Defense Health Agency she modeled total force fitness, then was asked to model the effects of policy during COVID rather than the disease itself. The work forced coordination across agencies, departments, and services on a scale not seen since World War II.   (22:24-32:32) The Power of Relationships Across a Career Describing herself as an introvert, she explains why relationships are the engine of accomplishment, recalling a Ranger literally pushing her up a mountain during advanced camp after a car accident. Those bonds endured and resurfaced decades later in Texas during the DARPA Triage work. She recounts retiring out of Poland after 28 years, where she stood up a secure network to coordinate 26 non-doctrinal partners supporting aid to Ukraine.   (32:32-37:37) Autonomy Confidence and Knowing When to Explore She makes the case for military service as a path to clinical autonomy and the chance to think, decide, and do research that civilian roles often do not allow. She reflects on how to know when to pursue a new opportunity: trust your chain of command, negotiate and listen when you are the one in charge, and act on principles of doing no harm. Confidence, she says, means not being afraid to fail.   (37:37-51:33) The DARPA Triage Challenge and Lessons That Last She gives a plain-language tour of her team's autonomous triage work — payloads that read physiological state without touching a casualty, visual reasoning models tempered by Bayesian rigor, and platforms that deliver the right information to each echelon. Using a DoD-wide tobacco policy as a case study, she explains the art of the doable and building success on small wins. She closes with advice on confidence, integrity, and holding doors open for the next generation.   Take Home Messages Cross disciplines to scale care: The greatest gains often come from teaming up outside your own specialty. Pairing clinical insight with engineering, informatics, and operations lets a single provider extend capability and capacity far beyond what one profession can deliver alone. People first is a practice, not a slogan: Leaders who genuinely put people first earn the trust that makes hard missions possible. The example of a leader who recognized her team while facing her own serious illness shows that the principle is proven in action, not in words. Relationships are the engine of accomplishment: No one knows everything, and progress depends on the people willing to push you up the mountain. Networks built early endure for decades and can be called on when the mission needs them most. Define what success really looks like: Insisting on the perfect outcome can stall progress entirely; agreeing on the art of the doable moves the mission forward. Real success is often a series of small wins that build on one another over time. Confidence means not being afraid to fail: Growth lives outside the comfort zone, and everyone fails sometimes. Acting with honesty, integrity, and your best effort each day — then trusting tomorrow brings another chance — is what builds lasting confidence. Episode Keywords military medicine, Army nurse, military nursing, WarDocs, military medicine podcast, public health service, USPHS, DARPA Triage Challenge, autonomous triage, battlefield medicine, combat casualty care, Carnegie Mellon University, Auton Lab, nursing informatics, biosurveillance, COVID modeling, population health, Defense Health Agency, Walter Reed, military innovation, medical robotics, drone medicine, military mentorship, veteran leadership, military medical research Hashtags #MilitaryMedicine, #WarDocs, #ArmyNurse, #PublicHealth, #BattlefieldMedicine, #DARPA, #MilitaryInnovation, #VeteranLeadership   Biography Dr. Kimberly Elenberg, a retired USPHS Captain, is the Director of Data and Mission Partner Sharing at ECS. A distinguished leader in biosurveillance and emergency response, she applies data science to enhance national security. Notably, she served as the incident response commander for modeling and analytics for the Secretary of Defense COVID Task Force. Previously, as a principal scientist at Carnegie Mellon University, she advanced autonomous systems for biosurveillance. Dr. Elenberg consistently bridges theoretical research with practical healthcare delivery, leveraging her clinical expertise and military discipline to safeguard public health. Her exceptional contributions have earned her several highly prestigious awards, including the 2022 Defense Superior Service Medal, the 2022 USPHS Distinguished Service Medal, and the 2020 National Emergency Preparedness Award for her outstanding operational acumen.         Honoring the Legacy and Preserving the History of Military Medicine The WarDocs Mission- WarDocs exists to honor the legacy of Military Medicine, preserve its history, and inspire every generation — across all Services, Corps, and Ranks — to serve with excellence and pride. Through mentorship, coaching, and education, we equip those considering, entering, and serving in military medicine with the knowledge, connections, and community they need to thrive. We celebrate Who we are, What we do, and, most importantly, How we serve Our Patients, the DoW, and Our Nation.   Find out more and join Team WarDocs at https://www.wardocspodcast.com/ Check our list of previous guest episodes at https://www.wardocspodcast.com/our-guests Subscribe and Like our Videos on our YouTube Channel: https://www.youtube.com/@wardocspodcast Listen to the “What We Are For” Episode 47. https://bit.ly/3r87Afm   WarDocs- The Military Medicine Podcast is a Non-Profit, Tax-exempt-501(c)(3) Veteran Run Organization run by volunteers. All donations are tax-deductible and go to honoring and preserving the history, experiences, successes, and lessons learned in Military Medicine. A tax receipt will be sent to you. WARDOCS documents the experiences, contributions, and innovations of all military medicine Services, ranks, and Corps who are affectionately called "Docs" as a sign of respect, trust, and confidence on and off the battlefield, demonstrating dedication to the medical care of fellow comrades in arms.     Follow Us on Social Media Twitter: @wardocspodcast Facebook: WarDocs Podcast Instagram: @wardocspodcast LinkedIn: WarDocs-The Military Medicine Podcast YouTube Channel: https://www.youtube.com/@wardocspodcast    

This week, we dive deep into the world of data, decision-making, and uncertainty with Dale Nesbitt, a lecturer at Stanford and principal at Arrowhead Economics. Drawing on his unique upbringing in a mining town, Dale Nesbitt shares how witnessing raw data collection firsthand shaped his perspective on what it really takes to make informed decisions—hint: it's not just about having more data.Together, we explore the pitfalls of relying solely on data for critical choices, the importance of understanding probability and risk, and why data-gathering itself is often a noisy and imperfect process. From commodity pricing and speculation in oil markets to the real-world impact of data-driven decisions in healthcare, Dale Nesbitt reveals why true analytic power comes from combining rigorous analysis, sound judgment, and the right kind of data—not just more of it.Join us as we challenge myths around "data-driven" decisions, unpack lessons from COVID-era data science, and discover why wisdom of the crowd, probability, and a healthy respect for uncertainty are key to navigating our data-rich world.LinksDale's LinkedIn profile -https://www.linkedin.com/in/dale-nesbitt-b574a83a/Watch on YouTube -https://www.youtube.com/watch?v=USOKgv1avHoTime Stamps00:00 Growing up in a mining town05:44 Data as the New Crude Oil07:31 Estimating and Understanding Stochastic Processes12:49 Impact of Strait of Hormuz Closure14:19 Challenges of AI in Economics17:05 Betting on events and elections21:43 Bayesian analysis and hydroxychloroquine data23:28 Understanding data and judgment26:38 Analyzing data for better decisions

Today's clip is from Episode 158 featuring Stefan Radev. In this conversation, Alex Andorra and Stefan break down a core argument from their paper: Bayesian statistics has never been more computational than it is now, and simulation is the thread that ties the whole workflow together.Stefan parcellates the Bayesian workflow into four stages, and this clip covers the first two. Stage one is model specification, where the workflow community has long recommended prior predictive checks. You can do this informally, just running simulations from your model and eyeballing whether the output meets your expectations, or formally, à la Michael Betancourt, by pushing your model's high-dimensional output through a transformation into a low-dimensional, interpretable space and checking it against reality. The punchline: a surprising number of models can be discarded before you've even seen real data, yet Stefan notes these checks remain underused in practice.Stage two is model verification, where the question shifts to whether your inferences are well calibrated. This is the territory of simulation-based calibration and parameter recovery studies, classic tools that have always carried a steep computational price. You simulate thousands of synthetic datasets and run inference on every single one, which is exactly why these checks are so often skipped in papers, even though doing one well can be a contribution in its own right.Here's where amortized simulation-based inference changes the math entirely. Checks that used to take days now take seconds, and instead of laboriously running inference dataset by dataset, you get millions of posterior samples essentially for free. The calibration checks that the field has always known it should be doing finally become cheap enough to actually do.Get the full discussion hereSupport & Resources→ Support the show on Patreon→ Bayesian Modeling Course (first 2 lessons free)Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work

Yes, I mean whale privilege, aka fat women who demand attention. Lots of topics today, deadlifts, bayesian curls, wild meals and questions, solid rants. SUMMER SWOLE SPECIALS: https://summerswole.com