Podcasts about Adjacent

A guest podcast has entered the feed! Zach and Jess did a musical episode of the incredible Erin Whitehead and Seth Morris' podcast "College Town" and now it's here JUST FOR YOU!And It Was Called Jello: Don't ghost us this week, Kevin, have we got a show for you: Beebo and Jan chat… and SING? Joined by poetry student Kaylee Ryan Everheart (Jessica McKenna) and Mammet Valley legend Darien Chance (Zach Reino), hear songs about stunningly hot ghosts, rat tail wishes, living in the slash, mine shaft collapses, memory lapses and more!Featuring Dana Wickens on improvised drums and Brett Morris on improvised guitar(s)!Leave a voicemail for College Town at www.cbbworld.com/callcollegetown or email them at collegetownpod@gmail.com and they'll respond on the show!See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Anatol Lieven of the Quincy Institute for Responsible Statecraft analyzes the multi-faceted, global strains of the Russia-Ukraine war. He discusses how Ukrainian attacks on Russian civilian-adjacent sites like Sochi and the Wildberries warehouse aim to demoralize Russia, though historically such strategic bombing campaigns fail. Meanwhile, Russia is betting on a war of attrition, weaponizing Black Sea grain blockades to squeeze Ukraine's economy and strain global food supplies. Additionally, Ukrainian drone strikes on Russian refineries are triggering a worldwide energy crisis, while Western nations like the UK face economic paralyzation as they struggle to balance heavy defense commitments with severe domestic recessions. (1)

Joe Marrese says he was camp adjacent in reference to summer break and attending camps. Joe and Aaron pick the winner of the Most Relaxing Things Bracket. Joe makes it known that there is tar in the sand of the beaches around Southern California. All this and much more on the 268th episode of Joe Code! Be sure to subscribe, rate, & review the pod wherever you listen Write Joe an email: joecodepodcast@gmail.com Support Joe Code at www.patreon.com/joecodepocast

Kennedy Caughell joins Kyle to discuss her debut album, Just the Beginning, which highlights a variety of stellar musicals in which she has appeared, including Wicked, 9 to 5, and Waitress. Also on the album is a cover of "Being Alive."To pre-order the album you can click here. Follow Kennedy on Instagram by clicking here.Check tour dates for Hell's Kitchen by clicking here. Send feedback to puttingittogetherpodcast@gmail.comKeep up to date with Putting It Together by following its social media channels.Patreon: https://www.patreon.com/puttingittogetherpodcastTwitter: https://twitter.com/sondheimpodcastInstagram: https://www.instagram.com/sondheimpodcast ★ Support this podcast on Patreon ★

Front CFO & COO Meredith Finn joins CJ to explain why finance leaders are taking on more of the operator's job. They break down the CFO-COO dual mandate, how Front funds new bets inside a mature SaaS business, why AI requires more coordination than simply handing everyone new tools, and how finance teams should think about the growing cost of AI.—SPONSORS:Brex is an intelligent finance platform with AI-powered workflows that enforce expense policies at the point of sale, match receipts automatically, and reduce month-end close from weeks to hours. Thousands of companies, including Anthropic, Coinbase, and DoorDash, already run on Brex. Stop asking A-level finance talent to do B-level admin work. Learn more at https://www.brex.com/metricsAnrok is the sales tax platform that watches your exposure everywhere, automates compliance, and flags risk before it turns into a surprise back-tax letter from a state you've never set foot in. Companies like Anthropic, Notion, and Vanta already trust Anrok to stay ahead of rules that move faster than any spreadsheet can. Talk to a sales tax expert for a personalized exposure estimate at https://www.anrok.com/rtnRightRev is a revenue recognition platform built for the AI economy, helping finance support usage-based pricing, credits, hybrid contracts, seats plus consumption, and whatever commercial model comes next. It gives product teams the freedom to keep innovating without outdated revenue systems slowing them down. Learn how RightRev can help at https://rightrev.com/cjPulley is an equity management platform that lets you issue options, model dilution, and complete 409As without your cap table turning into a spreadsheet disaster. Founders raising, hiring, and scaling use Pulley to keep equity clean and stay focused on building. Learn more or request a demo at https://pulley.com/mostlymetricsRillet is an AI-native ERP built for modern finance teams that want to replace NetSuite and close faster. With revenue recognition, close management, multi-entity support, and native Stripe and Salesforce integrations, Rillet helps scaling companies run their finance stack in one place. Hundreds of teams, including Windsurf and Mercor, use Rillet to make the zero-day close real. Book a demo at https://www.rillet.com/cjMaximor is an autonomous finance platform that runs order-to-cash, procure-to-pay, the close, cash management, and reporting on self-learning agents instead of a dozen disconnected tools. One PE-backed customer posts 98% of transactions directly to its ERP, with the remaining 2% routed to a human for review. You pay for outcomes, not seats. See it at https://www.maximor.ai/—LINKS: Mostly Talent: https://mostlymetrics.typeform.com/to/cLTxtAsNGuest: https://www.linkedin.com/in/meredithfinn1/Company: https://front.com/CJ: https://www.linkedin.com/in/cj-gustafson-13140948/Mostly metrics: https://www.mostlymetrics.com—RELATED EPISODES:Adam Swiecicki - CFO of Ripplinghttps://youtu.be/JyGJVpmZNacAurélien Nolf - CFO of Navanhttps://youtu.be/siwzJSRXAvY—TIMESTAMPS:0:00 The CFO/COO Dual Mandate1:45 Welcome to Run the Numbers2:52 Front's In-House Podcast Studio3:41 Meredith's Investment Thesis on Front7:21 The "Portfolio Career" Analogy8:35 What She Had to Unlearn as an Operator13:13 Books on Strategy vs. Execution16:17 When a CFO Should Also Be COO18:27 What the COO Hat Actually Covers20:20 Where the Two Roles Conflict25:42 Budgeting for AI Token Costs29:16 Using Internal Usage as a Benchmark35:40 Funding a Series A Bet Inside a Series D Company38:04 Sizing Bets: Core, Adjacent, Moonshot40:04 The Secret to Re-Accelerating Growth41:46 Going Deeper, Not Wider43:18 What the Pricing Overhaul Taught Her47:30 Why Customer P&L Matters50:11 Lightning Round Begins51:32 Advice to Her Younger Self52:33 Front's Finance Tool Stack53:21 Craziest Expense Ever#RunTheNumbersPodcast #CFOLife #FinanceLeadership #AIinFinance #SaaSGrowth

First metatarsophalangeal joint arthrodesis is a reliable treatment for painful end-stage degenerative, inflammatory, and post-traumatic conditions and severe deformities of the first ray. Although long-term functional outcomes are well documented, the incidence and relevance of degenerative changes in adjacent joints of the medial column after first metatarsophalangeal joint (MTP1) fusion remain uncertain. In conclusion, MTP1 arthrodesis provides excellent long-term satisfaction and functional outcomes with low revision rates. Although radiographic adjacent-joint degeneration, particularly involving the interphalangeal joint, may occur over time, these changes do not appear to influence clinical outcomes.

Josh only wants one thing for his birthday, and it's disgusting

Richard is the co-founder of ⁠1of10⁠, a research platform built by YouTube strategists, and his team has quietly been behind the scenes for some of the biggest channels on the platform—helping creators accumulate over 2 billion views through a repeatable, data-backed system. In this episode, Richard walks through his complete four-phase ideation system—audience identification, outlier research (using five distinct methods), idea remixing, and validation—and backs every step with real examples. We talk about what happens when the wrong audience floods your channel, why creators should double and triple down on formats that work, and how a single title change took one creator's video from 10,000 views to 150,000. He also shares data from 300,000+ YouTube outliers on the ideal title length (hint: shorter than you think) and where the sweet spots are for video duration across different niches. ⁠Save 20% on 1of10 using code JAY20⁠ ⁠Schedule a 1of10 Strategy Call⁠ ⁠Full transcript and show notes⁠ *** TIMESTAMPS (01:12) Where 80-85% of YouTube success comes from (01:50) Phase 1: Audience (03:19) When should you start a fresh channel instead of pivoting? (04:09) The danger of going viral with the wrong audience (05:40) Phase 2: Research (07:37) Format vs. Interest Topic (08:00) Method 1: Inside your own channel (10:52) Tripling and quadrupling down (12:33) Method 2: Inside your niche (13:45) Method 3: Adjacent niches (16:00) Method 4: Outside your niche (17:37) The "Japanese Rule" format (20:56) Method 5: External inspiration (22:07) Phase 3: Remixing (23:00) Escalation, inversion, and interest topic replacement (24:10) Viral vectors: concepts that work across all niches (25:28) Phase 4: Validation (27:00) Optimal video duration by niche (30:45) Why long videos are making a comeback (31:39) Total Addressable Viewership (34:36) Titles: Fear, Curiosity, and Desire as the three core drivers (37:17) Data: Title Length (37:51) Three methods for generating title angles (42:11) Thumbnails: Composition and Elements (45:11) It's never too late: title/thumbnail changes (46:10) Live demo: 1of10 thumbnail generator (48:10) The full 1of10 workflow *** RECOMMENDED NEXT EPISODE → ⁠#282: David Altizer — How to Make Great Thumbnails (For Non-Designers)⁠ *** ASK CREATOR SCIENCE ⁠Submit your question here⁠ *** WHEN YOU'RE READY

In this episode Patrick watched some stuff and Andy watched some other stuff.

In this presentation, Shaykh Arif Abdul Hussain examines how doctrines now central to mainstream Shi'i theology and piety — such as wilāya takwīniyya, the Imams' knowledge of the unseen, and tawassul — sit within a broader Qur'ānic definition of ghuluww. He traces this to an inconsistency in how verification standards are applied across different genres of Imāmī literature, and argues that applying the Imams' own criteria consistently exposes these ghulūw-adjacent idioms.

Watch the full episode on YouTube:We first covered Baseten last year when DeepSeek mania was at peak hype. Now they have raised a monster $13B round and become one of the new cohort of AI Infra decacorns that are (with Nvidia, Intel, and the semis complex) chief beneficiaries of the Inference Inflection. We return to Baseten at the peak of the 2026 edition of Open Weights debate. Ali has published a viral breakdown of Kimi K3:And since you last saw him, Philip has spoken at AI Engineer and written the definitive book on Inference Engineering spotted all over SF:Three years ago, inference engineering barely existed as a category.Today, it is one of the most critical disciplines in AI. Inference engineering inherently tackles a different question than standard model training: “How do you turn those weights from training into a product that is fast, reliable, and affordable at scale?” Focusing on these creates an entirely new optimization problem.In one recent GLM-5.2 experiment, quantizing more of the model actually preserved its benchmark quality while increasing throughput by 20%, because the errors introduced in different layers could cancel each other out.Inference is no longer just the final step after training. It is becoming its own engineering discipline, with its own research problems, infrastructure, and increasingly specialized roles.In this episode, Baseten's Philip Kiely and Ali Taha join swyx and Vibhu to explain what actually happens after a new open model is released and what it takes to turn “we generated a token” into a fast, reliable, production-ready API.We go deep on cache-aware routing, disaggregated prefill and decode, quantization, speculative decoding, KV-cache movement, model parallelism, GPU kernels, and the race to make frontier models up to 10× faster. Philip and Ali explain why inference optimizations can still produce gains of 20%, 100%, or even 200%; how quantization errors can cancel one another out; why identical weights can behave differently across clusters; and how Baseten grafted a Kimi vision encoder onto GLM-5.2 without changing the underlying language model.The conversation then expands beyond LLMs into NVIDIA Dynamo, mega kernels, Rubin, AI-specific chips, local inference, video generation, diffusion versus autoregressive models, and the enormous compute barrier to generating coherent long-form video. Finally, we explore the convergence of training and inference, continual learning through persistent KV cache, and the emerging loop where models help optimize the infrastructure that runs them.We discuss:* What happens when a 200,000-token request enters an inference system* Cache-aware routing and reusing previously computed KV cache* Why prefill and decode are increasingly handled by different GPUs* When dedicated deployments become cheaper and more reliable than shared APIs* How speculative decoding uses a smaller model to accelerate a larger one* Tool calling, structured outputs, and what LLMs actually do* What it takes to support a new open model on day zero* Grafting Kimi's vision encoder onto GLM-5.2* Retrofitting inefficient model layers with components from other architectures* Why models sometimes collapse into repeating the same token* How hardware, kernels, and race conditions create nondeterministic failures* Preserving model fidelity while making inference faster* How quantization errors can cancel each other out* Why inference optimizations still deliver gains of 20%, 100%, and 200%* How optimized serving can make a model up to 10× faster* NVIDIA Dynamo, KV-aware routing, and distributed model serving* Speculative decoding the speculative decoder* Why local AI is about making models less dumb while data-center AI is about making them less slow* Tensor, expert, and pipeline parallelism across GPUs* Hardware-aware model design, auto-tuning, and the case against mega kernels* Rubin and why inference is becoming a systems problem* Whether modern GPUs are evolving into programmable AI ASICs* Why enormous models like Kimi K3 require GB300-class hardware* Why open-source video generation still trails Veo, Kling, and other closed models* The quadratic attention bottleneck behind long-form AI video* Autoregressive video, real-time generation, and compounding quality drift* Why future video systems may combine autoregressive and diffusion architectures* Training for inference and inference for training* Continuous post-training, deployment, evaluation, and improvement loops* How GLM-5.2 helped optimize the kernels serving GLM-5.2 itself* Why faster networking could unlock dramatically faster decoding* Continual learning, KV-cache compaction, and persistent model memoryShow Notes* How to build a day-0 API for Kimi K3* 22580: From GPT2 to Kimi3, ExplainedPhilip Kiely* LinkedIn: https://www.linkedin.com/in/philipkiely* X: https://x.com/philipkiely* Inference Engineering: https://www.baseten.co/inference-engineering/Ali Taha* LinkedIn: https://www.linkedin.com/in/aliestaha/* X: https://x.com/waterloointernTimestamps00:00:00 Introduction and the 200K-Token Prompt00:03:18 Dedicated Deployments, Speculative Decoding, and Tool Calling00:11:26 Launching Production-Ready Open Models00:19:06 Model Retrofits, Failure Modes, and Nondeterminism00:28:22 Quantization and Canceling Errors00:32:15 The Race to 10× Faster Inference00:40:48 Dynamo, Speculation, and Local vs. Data-Center AI00:50:18 Model Parallelism, Auto-Tuning, and Mega Kernels01:00:55 Rubin, GPUs vs. ASICs, and Custom AI Chips01:10:03 Giant Models and the Limits of GPU Memory01:12:42 AI Video, Quadratic Attention, and Autoregressive Generation01:21:47 Audio, Images, and Diffusion Models01:27:32 Training, Self-Optimizing Models, and Continual Learning01:40:06 Closing ThoughtsTranscriptIntroduction: Baseten, Waterloo Intern, and Inference EngineeringSwyx [00:00:00]: Okay, we're here in the studio with Philip, old friend from Inference Engineering, the book, as well as Baseten and everything that you've done, you and I have done before, as well as Ali. Welcome.Ali [00:00:15]: Pleasure to meet you.Swyx [00:00:15]: Waterloo intern.Ali [00:00:16]: Waterloo intern, always.Swyx [00:00:17]: When did you get “Waterloo intern” as a handle?Ali [00:00:19]: As a handle? Oh.Ali [00:00:20]: I think the rebranding happened mid-March. When I saw it was open, I was like, “I have to take it. Up for grabs.”Philip [00:00:26]: The problem is that Ali is really good at his job and is not gonna be an intern much longer.Philip [00:00:30]: So we have to figure out who's gonna get the handle.Ali [00:00:33]: Well, I'll pass the torch over to the next intern.Swyx [00:00:34]: Oh, okay. It can be, like, you just pass it to another Waterloo grad.Ali [00:00:37]: To another Waterloo intern. No, bruh.Philip [00:00:39]: Yeah.Ali [00:00:39]: Intern.Swyx [00:00:40]: Intern, yeah.Ali [00:00:40]: And no.Philip [00:00:41]: You gotta get an intern from Waterloo.Ali [00:00:42]: Yeah, I've gotta get an intern from Waterloo.Swyx [00:00:44]: Right.Ali [00:00:44]: But they have to follow the path.Swyx [00:00:45]: Oh, it could, but it could come from Baseten, so it's like whoever Baseten gets from Waterloo.Ali [00:00:48]: Right.Swyx [00:00:49]: Has the title of Waterloo.Ali [00:00:50]: It stays in the ecosystem.Philip [00:00:51]: Exactly.Ali [00:00:52]: Halfway through the internship, you either get it or you're out.Philip [00:00:55]: You should also do, like, a big graduation ceremony where you change the handle.Ali [00:00:59]: Just say it.Philip [00:00:59]: For everybody.Swyx [00:01:00]: You guys are good at ceremonies, clearly. We had a nice launch of the book, very successful. But before we get into all that, I wanna start off with a fun question for you. Okay, you're an expert inference engineer. What happens when I send a long query, say two hundred thousand tokens into Baseten's inference? What's the process of query through GPU model routing, balancing, all that? What is all the stuff that we don't think about?Long Context Requests, KV Cache, and Cache-Aware RoutingPhilip [00:01:26]: With a long query specifically, the first thing that I'm gonna ask is, “Have you sent me this query before, or at least part of it?” and I really hope you have, because it's gonna be a lot easier for me and a lot cheaper for you. So the first thing that we're gonna look at is some cache-aware routing, where we're going to see, we probably have a number of instances, a number of replicas up serving whatever model you're hitting. We want to send this one to something with, number one, available prefill workers, and number two, ideally some cached input already there so that we can skip prefill on at least part of these two hundred thousand tokens. If you're doing two hundred thousand tokens, it's probably coding or a multi-turn agent or something where you would expect to have that cached. If you don't, we're gonna have to send it to a prefill worker. We've at least on certain models disaggregated prefill and decode, so you're going to have one set of GPUs that's solely going to process the input, create the KV cache, and get you your first token, and then that's going to be passed over to a separate set of GPUs, which is going to run decode. We're going to iteratively make those tokens. We're probably going to have some speculator model in front of that. I'm going to assume that you're doing coding, and because of that, our speculator model, which assumes you're doing coding, is gonna have a high draft token acceptance rate. If I'm wrong and you're asking me to summarize every Harry Potter book, it's gonna be slower. And then we stream that output to you and account for it, charge you, a couple of pennies and say, “Hey, would you like to send another one?”Swyx [00:03:04]: Except Baseten doesn't charge by pennies.Philip [00:03:07]: Well, yeah, we charge. I'm assuming that we're talking about the public model APIs. If you are setting up a dedicated deployment, then yeah, it's not pennies.Public APIs vs. Dedicated DeploymentsSwyx [00:03:18]: Yeah, one of the key differentiators when I was talking with Baseten initially was that people who want very high volume just need to rent by the box, ‘cause then it's up to you to figure out how to saturate the box.Ali [00:03:31]: And more often than not, it's, like, way cheaper if you're pushing, like, millions of tokens per hour, if you just pay per hour instead of pay per token.Philip [00:03:37]: Yeah, they do. I think that we've increasingly seen a lot of demand for the pay per token APIs, just because everyone wants to try open models, and then once they find a use case that's really sticky, then they move over to dedicated.Swyx [00:03:51]: Is there a best practice on when it's time to swap over?Philip [00:03:54]: Couple reasons. Yeah, reliability, that's a big one, right?Ali [00:03:57]: Like, if they have a very specific use case, they want you to train something specifically for them, like they want their own spec dec, for instance, for their own traffic.Swyx [00:04:04]: Spec dec is speculative decoding.Speculative Decoding and Custom SpeculatorsAli [00:04:05]: Speculative decoding, yeah.Swyx [00:04:07]: You have to explain.Ali [00:04:07]: Sorry. Like, speculative decoding is like, if you have a huge model, right? And so the model is going to be generating one token at a time every single turn, every single forward pass. So we attach, like, this little, like, parasite, like this layer that goes on top of the model, and this model just has to predict. It does three very fast autoregressive forward passes, and it will predict, like, three certain tokens, and then you do one forward stage over the entire original model in order to see if those predictions were correct or not, and then you accept them or you reject them. Now, this draft model is traffic specific, so if you, like, Philip said, if you're summarizing Harry Potter books, I can train exclusively that draft model on Harry Potter books, and I can guarantee you that I'm gonna accept the three tokens every single time. And so with that case, I increase your decode speed. I wouldn't be able to provide this to you if you're a shared endpointSwyx [00:04:53]: YeahAli [00:04:53]: ‘cause I have no idea if you're doing Harry Potter, if you're doing coding, if you're doing English. We don't know. Also, there was a thing in the book that mentioned that if they really cared about a specific threshold, chapter four, I think. Do you remember that?Philip [00:05:06]: Yeah. The things that you can do is you can set a specific, like, batch sizing, a specific, like, parallelism strategy if you're trying to optimize for, like, throughput versus latency. You can. Maybe a NVFP4 quant doesn't pass your benchmarks and you wanna run a model at higher precision, you could do that. There's just a bunch of reasons why you might wanna have your own endpoint and the biggest one, of course, just being, like, you don't have to deal with someone else doing a hundred million tokens of benchmarking traffic at the endpoint when you happen to be trying to serve your users.Swyx [00:05:40]: Yeah. I think one thing that is. That is a classic journey. Like, it's people is asking the, what happens when you type Google into the browser. Tool calling, is that just, you're generating JSON or is there more complication beyond that?Tool Calling, JSON, and Structured OutputsAli [00:05:58]: Certain customers that we have, they have their own post-trained models, and so they demand a tool calling that's not just, like parse a file or go find the weather. It's something that's very specific and you have to do post-training on this. And if the post-training on the model is not good or if the quantization after the post-training to get the inference to be fast, the model will struggle reading the JSON file and reading the tool calling. But it doesn't require its own like sandbox. It's not like it's going to use that tool calling to like escape a sandbox or like it doesn't have to be contained. It can just be a normal dedicated deployment. The challenge with tool calling more and more seems to be that the companies want certain tool calling which is a very sensitive thing to train. And because you're dealing with all of the JSON outputs, if it doesn't like close the end of the request in a very certain manner, you end up with a model that did the tool calling and like the thinking and so as a result of that, it didn't see the result and just hallucinated the result as it decoded. That seems to be the most challenging thing with tool calling, not really the sandboxes model.Philip [00:06:56]: Yeah, that's a challenge on the training side and then on the inference side, there's work that you can do to scope the possible output. So we published this at this point close to two years ago, the solution to this problem which is you make a state machine and you use that to constrain the output to a specific format. So this is the structured output problem. If you remember backSwyx [00:07:27]: Yeah, the specific grammar is,Philip [00:07:29]: Yeah, exactlySwyx [00:07:30]: GML had this thing.Philip [00:07:31]: Yeah. So it's like the old-school “make sure this is only JSON”, return only JSON orSwyx [00:07:38]: YeahPhilip [00:07:38]: Grandma's gonna die type of prompts.Swyx [00:07:39]: Is it BNF grammar? At some point OpenAI had released a thing that was like, yeah, if you want to constrain your output, write BNF grammar, back as NOR.Philip [00:07:47]: In our inference system, it's just a specified output format. And you get the guarantee that your output's gonna be structured along that format. And so applying that to tool calls can like help cut down on. You can still call the wrong tool or call no tool. It doesn't solve the certainty problem but it at least solves the output structuring problemSwyx [00:08:10]: YeahPhilip [00:08:10]: Within tool calls.Swyx [00:08:12]: And MCP is just another form of tool, right.Philip [00:08:14]: Yeah, exactly.Swyx [00:08:15]: As far as there's no special thing there.Philip [00:08:16]: The thing I'm always like explaining to people is the LLM is not capable of doing anything. It's only capable of making suggestions of what to do and then if those suggestions are formatted in a certain way and applied to a system that knows what to do with them, then an action occurs.Swyx [00:08:32]: Yeah. Part of the fun stuff is, this is solved outside of tool calling too. Like in an agent loop if the output is not correct or you're right, like reasoning, tool calling was done in the reasoning trace, just be like, “Oh, I don't know what to do. Let me just try again.” And it might get there after a few tries. And on your point of training, sometimes this is harder in smaller models, so you don't have the same exact quality outputAli [00:08:56]: Right.Swyx [00:08:57]: When you just swap from a big model, right?Ali [00:08:59]: Yeah. I will say that, before, I think we need to go back to inference engineering proper.Ali [00:09:04]: But, I had expected that something would replace JSON because it's hard to stream JSON ‘cause JSON must be complete and you must have open and close brackets and everything. So it's hard to parse something or validate something while it's being streamed. So people invented all sorts of things that are like, I forget the name of some of these alternatives, but it's something like TOML, something like YAML. But JSON seems to be dominant still.Philip [00:09:30]: The JSON outputs aren't that long, right? Like you could have a long-- ‘cause tool calls also contain the arguments in them and perhaps for a certain tool you might pass like a very long argument. But my impression of the median tool call is that it's a relatively small number of tokens, right? So I would expect that speculators are generally fairly good at something as formatted as JSON. And so you would have like a pretty fast decode step there and that the streaming wouldn't be as valuable, but maybe I'm wrong about that.Ali [00:10:02]: I think you're also bounded by the software or that the model is gonna integrate with if the software is built with JSON for the tool calls or if the company that you'- if your customer says that this is how our software works and our tools are interfaced with JSON, you can ask them to like, change their software and say like, “Yeah, this is gonna be better for the model.” but like with the right training shouldn't be that much of a difference. Also more profitable if it outputs more tokens probably.Swyx [00:10:25]: Depends on your business model.Swyx [00:10:27]: It really depends. But I will say that, as a writer with like experience a lot with generated output, I do try to move from text to JSON text which is very long JSON, right? Like there's paragraphs in every field because I'm trying to structure it, right?Philip [00:10:44]: Right.Swyx [00:10:44]: I want you to first make factual statements, then make opinions then make bullet point summaries, have dates, have entity references have your sources for references, all these things. Anyway, so these are things that like I think people who really experiment with structural output have to really care about. But, let's, let's recurse up the stack a little bit. Before we started recording, you mentioned something really cool, which is that there's a lot of engineering that-- inference engineering that goes on when a new model provider releases a new model, right? So let's call it GLM-5.2, Kimi K3. I had previously assumed, especially if it's like, well, GLM 5 to 5.1 to GLM-5.2, like that you've supported them before. Is it that much work?What It Takes to Support a New Open ModelAli [00:11:26]: It's a lot of work.Swyx [00:11:28]: Yeah. Okay. So like, a lot of people, all you guys, right whenever a new model launch like, people rush to say like, “Oh, Hugging Face supports this, Fireworks supports this, Spacetime supports this,” and I'm like, “Yeah, of course we support it.” But what goes into that? What goes intoPhilip [00:11:40]: I think it's more than just support it too, right? It benefits the consumer a lot. Like I think it was with Kimi K2.5 or GLM-5.2 the latest, there was an inference war, right? X provider is at 90 tokens a second. The next day we're at 150. The nextSwyx [00:11:55]: I kinda kicked that off with the GLM-5.2.Swyx [00:11:58]: I wrote a Twitter article about. It got like half a million views,Ali [00:12:02]: Based on being numberSwyx [00:12:03]: YeahAli [00:12:04]: Or it's for something else.Swyx [00:12:05]: Yeah. Which,Ali [00:12:06]: Oh my GodSwyx [00:12:07]: Which then got everyone really excited about, hey, how can we, bend tracks a little bit further and,Philip [00:12:14]: There's a difference between support the model, as in I can make a token out of this model, and support a model, as in I have a production-ready API from this model.Philip [00:12:26]: Getting to the point of I can make a token out of this model is not that hard because generally the, open source inference engines, vLLM, SGLang of the world oftentimes even receive weights ahead of time, maintainers do, or the people making the model merge PRs to ensure support. So you generally can, just get it working on the standard open source stack without too much pain in most cases. The challenge is, every inference company is gonna have own proprietary stack. Some open source components, some in-house stuff. And for any arbitrary model, there's going to be some new stuff. Sometimes you get lucky, like K, two five to two six was, like, pretty similar.Quantization, Speculators, and Production ReadinessAli [00:13:16]: Yeah. It was pure continued post-trainingPhilip [00:13:18]: YeahAli [00:13:18]: If I remember correctly.Philip [00:13:19]: Even in those cases, there's still stuff you have to do. You have to redo the quantization work. You're taking the model from. Generally, these models are not released in NVFP4, and we want them to be in NVFP4 for maximum Blackwell compatibility. So we have to perform that quantization, and, calibrate the quantization to make sure that we're not causing any regression in the model's intelligence. And then we also have to train the speculator, as we've talked about. Generally, we have. We have ZDR, zero data retention on our model APIs, so we don't know exactly the traffic that people are sending us, but we know what's popular. We know that coding use cases are popular. We know that agents, agentic use cases are popular. So we can get public data sets that are representative of that traffic and train general speculators. Now, with speculators today, you need to train the speculator using the base model itself because you're getting hidden states out of the model from running inference on these specific prompts, and that is the training data you use to create the speculator. So there's that process which you need the real model weights for. And then there's of course just the process of, standing up all the infrastructure behind it, loading all this stuff, testing it. And then when there's a new model with a newer architecture, I think that, like, the DeepSeek models tend to be the most challenging as they have, like, the most novel architectural stuff going on, model after model. But every new model has something. Kimi K2 had. Oh, sorry, GLM-5.2 hadAli [00:14:53]: Sparse attention.Philip [00:14:54]: Yeah,Ali [00:14:54]: YeahPhilip [00:14:54]: the DSA.Ali [00:14:55]: Right. Which is brought from DeepSeek.Philip [00:14:57]: Yeah. AndAli [00:14:59]: So you can copy-paste then?Philip [00:15:01]: It kindAli [00:15:01]: I don't know how this works.Philip [00:15:02]: So, like we had to, like, build support for that into our runtime. And you're right, like it is really interesting the way that all of these open source labs borrow from each other. For example, like GLM-5.2 doesn't have vision. So something that, Haley, a guy on our team, if we could take a look at this, he, like, grafted the Kimi vision encoder onto GLM-5.2.Retrofitting Vision into GLM-5.2Ali [00:15:27]: We'll be training the projector.Philip [00:15:28]: Exactly. So if you think about, like, the encoder, there's the encoder, which is the part that looks at the image and turns it into latent information, and then there's the projector which likeAli [00:15:38]: You can say latent space. It's okay.Philip [00:15:41]: And then there's the projector that maps it onto, the model itself, and then there's the model weights. You don't wanna mess with the model weights because you run a chance of making the model dumber at something else for the purpose of giving it vision. So instead, Haley started with just a projector, which is only a handful of millions of parameters.Ali [00:16:02]: That would be, yeah.Philip [00:16:02]: Yeah.Ali [00:16:03]: Can you show the training one?Ali [00:16:04]: Like the way it groksPhilip [00:16:05]: YeahAli [00:16:06]: Very interesting.Philip [00:16:06]: And maybeAli [00:16:07]: That right therePhilip [00:16:07]: Maybe Ali, you should take it from here. You've got a betterAli [00:16:10]: Ooh, double the sandPhilip [00:16:11]: Understanding of this than I do.Ali [00:16:11]: Yeah. You can see, like, he. The way he trained this is really cool. At the beginning, he was training it using just like, “Here's a picture of a mountain. Can you describe what's in this mountain?” And that caused it just like the first, learning walls. Like here you can see this all we're trying to teach it is to translate the encoded. Like it's already taken the encoder from Kimi K. It's taken the image. It'Philip [00:16:31]: Yeah. FrozenAli [00:16:31]: FrozenPhilip [00:16:32]: With adapter.Ali [00:16:32]: Exactly.Philip [00:16:33]: Yeah.Ali [00:16:33]: So the brain is frozen and the eyes are frozen. It's just we're tryingPhilip [00:16:37]: AlignAli [00:16:38]: Interconnect between the eye and the brain, right? So the projector. And so you take the tokens and then he's like, “Oh, can you describe what's in this image?” And he's like, “Oh, it's a mountain,” or it's a person or it's a human, whatever the case is. But that didn't cause complete understanding. So he changed it such that every image was associated with a data set of questions. Like, does this image have a white male? Does this image have birds in the top corner? Does this image have a scientist in it? All of that stuff. And it would have to answer questions correctly. And using not just training on describing an image, but being able to answer question, another question, answer over time. Like you can see the grokking, which is like genuinely insane, that retrofitting vision into a large LLM can learn to that extent. And even for images that it doesn't perform well on, for instance, if you ask it a picture of like Stephen Hawking, “Who is this?” Maybe it doesn't get it, but it will say something like, “This is Albert Einstein.” Like it still understandsPhilip [00:17:25]: Close enoughAli [00:17:26]: That this is a scientist who is a man who has, some significant achievements, all that stuff. So that's like really cool.Philip [00:17:32]: Yeah. So, we've covered Hao Tian before, who the author of the LLaVA paper that did this, a while ago. And I think that's very foundational work for anyone who hasn't done vision work before.Ali [00:17:41]: Same with the CLIP and MetaCLIP, where you go from just captioning to building out questionsPhilip [00:17:47]: RightAli [00:17:47]: Off the image and how much better you can get performance.Philip [00:17:50]: Right. Right. Right. Yeah. But what's, what's so exciting about this is if you look at a model like this. Now, this is a little bit more of a research project. It's not. It got to 56% on MMLU Pro, I think. So not quite frontier. But if you're running this model, you haven't suffered any loss on your GLM-5.2 quality. If you don't have an image, it'll just behave exactly the way it used to. And ultimatelyAli [00:18:14]: Which in the inference code you literally do not include the other part, right?Philip [00:18:18]: Yeah. You would just skip the encoder if you don't have an image input.Ali [00:18:22]: Okay.Philip [00:18:22]: Just confirming.Philip [00:18:23]: YeahAli [00:18:23]: Does it affect a lot on the overall inference side? Like you're not adding much, you're adding a very small vision encoder. These are typically likePhilip [00:18:30]: They're super fineAli [00:18:31]: Less than a billion parameters, right?Philip [00:18:32]: Yeah. It's, - There's a little bit less standardization among vision encodersSwyx [00:18:37]: YeahPhilip [00:18:37]: So the support matrix can be a little bit, sparser. But overall, yeah, it's a pretty, it's a pretty minor component of the overall system. And ultimately what you get out of the system is all of a sudden you have Kimi Vision, GLM weights, and DeepSeek attention all in one model.Open Source Model Grafting and Franken-MergesPhilip [00:18:56]: And that's, I think, a lot of the power and beauty of open source, is that you can take all of these different components and combine them together into a system that's better than anyoneSwyx [00:19:05]: YeahPhilip [00:19:05]: Can be individually.Swyx [00:19:06]: People used to say that you would also do Franken-merges where you would take likePhilip [00:19:10]: YeahSwyx [00:19:10]: Layers from each model.Swyx [00:19:11]: Does anyone do that anymore?Ali [00:19:13]: Well, to your point previously when you were mentioning like, the work that goes into supporting a model when it first comes out, like GLM-5.2 or MiniMax M3 or whatever the case is. Sometimes you do have to like, you do have to switch out some things. Like, for instance, the MiniMax M3 head uses full attention, and with full attention you end up with this like insane bottleneck in spec dec ‘cause you're doing auto-regressive token generation for three tokens, and you're doing this like N squared over all of the tokens that are in your sequence. Your KV cache is like very large because it's not sparse, it's not top K. So we find it better to like, okay, we're gonna replace this, we're gonna replace this layer with a layer from another model that's using like GQA, for instance. And then just with the right training, you can get it to have the same acceptance rate. So it is very possible to retrofit layers from other models and very much needed. If a layer is like inefficient, the training just becomes the challenge, like how do you ensure that you train it properly? Which again to your earlier point is like the mesh between training and inference. As in like you need very good training in order to do fast inference. That's like, I feel like more and more becoming true.Swyx [00:20:21]: Yeah. Anything else on the support side when you say like get it to fully production ready?Loop Detection, Race Conditions, and Non-DeterminismPhilip [00:20:26]: Yeah. I think that there's also a question of just, we can test a model to a pretty extensive degree, but we're trying to get it out quickly and then you see a bunch of other people test it and you get interesting results. There was an issue with, GLM briefly where we had some like mode collapses where it would just output the same token over and over again for certain prompts on certain temperatures. Like once you expose an endpoint to the real world, there's going to be, so many more varieties of things given to it that you're able to, discover and patch things. So it's not just a, day zero process, it's then like for the first week, for the first month, if a model remains popular, like how do you both fix bugs and then continue to push the envelope on performance?Ali [00:21:21]: What do you mean you don't want your model outputting S?Swyx [00:21:24]: Is there loop detection on that stuff, by the way? It still happens like quite a lot, which is surprising.Ali [00:21:30]: We have like we, in our endpoint, like if a model was to output the same token like four plus times, we just cut the generation. We say like, “Oh, sorry, this-- Like try again,” or like we will reprocess the request. ‘Cause we know then, like if it, like if, yeah, it's four times the same token, it's probably collapsed.Swyx [00:21:45]: Yeah. Is there a way to opt out in case I really want that?Ali [00:21:48]: You want that?Ali [00:21:50]: I think there's a way that we have to handle it. I'm not exactly certain, but I feel like in certain models, like when they output something like you can imagine, like a table for instance, and so they want, they wanna draw like 12 dashes and 12 dashes. Yeah, I think there's a way for that to happen. I think we only do it on certain tokens. Like we exclude certain special characters.Swyx [00:22:07]: Yeah.Ali [00:22:07]: So we only do it on like certain like S is the most common almost. GLM-5.2Swyx [00:22:11]: OhAli [00:22:11]: And I think it was DSV 4 as well. Like you'd just have like looping issues where like you literallySwyx [00:22:17]: ItAli [00:22:17]: Just have like S.Swyx [00:22:18]: Yeah. Is there a special, something special about S? No, just randomlyAli [00:22:21]: It just seems to be the one token involved.Swyx [00:22:23]: Yeah. And it'Philip [00:22:24]: Is thereSwyx [00:22:24]: And it's only temperature 0Ali [00:22:27]: NoSwyx [00:22:27]: Even at other temperaturesAli [00:22:27]: Even at like 0.9 or whatever, it will still, it will still collapse.Swyx [00:22:30]: That's weird, right?Ali [00:22:30]: It's, it is an inference problem to be honest, like a software problem. Like oftentimes, the image you run will-- like NVIDIA will release an image for instance, and if we will upstream the changes from their latest TensorRT-LLM image into our stack, we'll find that it fixes it. Or oftentimes this will only happen in an inference engine that you're using like SGLang. But if you were to switch to vLLM, that isn't the case. So it seems to be like an extremely like deterministic software issue and not really a model issue. It's not like a weights problem. Like I'- we'll say like, “Oh, it's a problem with the quant. We did PTQ wrong,” right? But that isn't, that doesn't make sense because the same weights used with a different inference engine does not repeat the problem. And sometimes it's, the kernels that are being used in the backend have like these very subtle sometimes race conditions, where if you were to use this model hosted on one cluster, you will never get this problem.Swyx [00:23:19]: Oh my God.Ali [00:23:19]: But if you host it on a different cluster, you will. And the reason is the KV cache transfer from a node to node in that one cluster is using a slower interconnect than the node to node in another cluster. So that exposes the race, whereas in another cluster it doesn't. So then you end up just like, okay, this model is not gonna be hosted on this cluster. We're gonna host it on, another cluster because that cluster exposed that problem. But then it ends up with like, okay, is it the software? Is it the model weights or is it the hardware?Swyx [00:23:42]: There is a thing about this with temperature 0 still not being deterministic, right?Ali [00:23:46]: Right.Swyx [00:23:46]: Mostly because of hardware. Even at temperature 0 same model, you won't always get the same output.Swyx [00:23:52]: Even-- But I'm surprised by the race condition one because, I thought PyTorch was a graph that like guarantees that you at least, execute things in the right order.Ali [00:24:02]: Well, yeah, true. Like I'm not, I'm not saying that there is. Like well, you have things like PTL optimizations where like you can start a kernel before the end of the previous kernel, and that's like ‘cause you want to do that because there'sSwyx [00:24:12]: It's like pipeliningAli [00:24:12]: Expense. Exactly.Swyx [00:24:13]: Yeah.Ali [00:24:13]: But it'- But you don't do it cleanly. Like you overlap a little bit of the execution. No, it is very possible that the kernel itself, like that one block that is supposed to be running in this instance of time, that kernel itself has a race condition. For instance, like a missing barrier. Like often if you're designing a kernel and you want it to make it to be very fast, if you don't test it extensively, you'll, you'll have certain threads access data points from registers before they've been written to by other threadsSwyx [00:24:36]: YeahAli [00:24:36]: For example, because like your barrier is wrong or your synchronization was wrong. But yeah, like the testing itself is very difficult in those like, andSwyx [00:24:42]: And there's no like borrow checkerAli [00:24:45]: What does that mean?Swyx [00:24:46]: Like Rust. Like the. If you're trying to have like memory safety It sounds like a comparable problem.Ali [00:24:52]: Well, yes, but you're working in CUDA, right, NVIDIA GPUs. Like- You just need a higher level language like modular Maybe that's what modular is supposed to do. I don't know.Quantization Quality and Vendor FidelityVibhu [00:25:00]: How do you see keeping quality of the model? So you talked about all these steps of, okay, you gotta do quantization, train your own speculative decoderAli [00:25:07]: RightVibhu [00:25:07]: Run on different hardware. Looking at other model providers, okay, you kicked off a inference speed race on the consumer end. What goes into keeping quality the same across them, right? Sure, you can run benchmarksAli [00:25:22]: YeahVibhu [00:25:22]: But, like, how do you determine how much quantization are there standards? What goes intoPhilip [00:25:27]: There's a few things on quality. Most inference optimizations are lossless. KV caching, for example. You are just recomputing or preventing recomputing the same values. Speculation, of course, if a draft token is wrong, it gets rejected. The main lossy optimization is quantization. And that really comes down to, number one, data format, number two, which parts of the model you choose to quantize, which layers, and number three, like doing a lot of calibration on the quantized weights, to ensure that you're preserving all the outliers. There's other tricks that you can do, though. A big one is long context, ‘cause one thing you asked at, right at the beginning is, “Oh, what's gonna happen if I send a 200,000 token request in?” So with a long input sequence, you need to, store a lot more information. You need to process a lot more tokens. And so even if a model has a context of a certain length, you might, as an inference provider, choose to build an API with a shorter context length, and of course a full length one as well. Because if someone doesn't need the full million token context, for example, you can get them better performance. I don't know if that's exactly like quality of the model. The way that I think about quality is to what degree are we faithfully serving the original model? If you think of a golden implementation of a model that performs exactly the way the model is designed to perform, I think of quality as how close are we getting to that, 100% fidelity of the model.Philip [00:27:13]: You can also, of course, think about quality from the training side and how do you push yourself past 100%. But when I think about purely inference optimizations, it's getting faster while staying as close to that 100% fidelity mark as possible. And certainly our standard internally is that, like you should not be able to tell the difference between our API and a, official API. I think Kimi in particular does a good job of vendor benchmarking hereAli [00:27:41]: YesPhilip [00:27:41]: Where they haveAli [00:27:42]: They released an actual vendor benchmark.Philip [00:27:43]: Exactly, yeah.Ali [00:27:44]: ‘Cause they accused, some people, Amazon? There was some provider that was not doing very well on Kimi's benchmark.Philip [00:27:50]: Yeah.Philip [00:27:51]: So, with Reflect we probablyVibhu [00:27:52]: This was a long time ago, right?Philip [00:27:54]: No.Ali [00:27:54]: Yeah, like threeVibhu [00:27:55]: They alsoAli [00:27:55]: Four, five months agoVibhu [00:27:57]: This also happened with, I don't remember which model, but they pulled out quite a few, and then they started a whole chart about this. It might have beenPhilip [00:28:03]: Kimi Vendor Verifier.Ali [00:28:04]: Yeah.Philip [00:28:05]: Yeah.Ali [00:28:05]: Yeah, ‘cause you, ‘cause you'd be pissed, right? Like if you'Philip [00:28:07]: Yeah.Ali [00:28:07]: If like if I'm a consumer and I'm using like Amazon's endpoint for instance, and I've used Kimi and I'm like, “Oh my God, like this is bad,” I'm not gonna say, “Oh, Amazon quantized the model in a bad way.” I'm gonna say, “Oh, Kimi sucks.” Right?Philip [00:28:17]: Yeah.Ali [00:28:17]: So it seems like that makes sense.Philip [00:28:19]: Yeah, they care. They care.Vibhu [00:28:21]: Justifiably.Ali [00:28:21]: Yeah, justifiably.Vibhu [00:28:22]: This is probably a stupid question, but just checking, has anything improved from main quantization?Philip [00:28:28]: Yeah.Vibhu [00:28:28]: Like, is quantization always strictly worse?Ali [00:28:30]: Well technicallyVibhu [00:28:32]: NoAli [00:28:32]: It's a lossy. QuantizationPhilip [00:28:33]: YeahAli [00:28:33]: Is a lossy, it's a lossy implementation.Philip [00:28:36]: Speed improvesVibhu [00:28:36]: Speed improves.Ali [00:28:37]: It the number, likeVibhu [00:28:38]: No, I' always look for inverse scaling laws.Philip [00:28:40]: Yeah.Ali [00:28:40]: Yeah.Vibhu [00:28:40]: This is something I learned from Noam Brown, where like things that normally act in one direction sometimes do.Philip [00:28:45]: Well, technically when you run a benchmark, because these models are deterministic, sometimes your,Ali [00:28:52]: YeahPhilip [00:28:52]: NVFP4 quant is like, two basis points higher than yourAli [00:28:56]: No, it's noise. It's noise.Philip [00:28:57]: Yeah, exactly. I'm like, yeah, it's, it's within. That's why I always say within margin of error.Philip [00:29:01]: And I stopped saying that because everyone assumes that what is, well, within some margin of error, we're barely inside of that to the worst, so we're saying. But yeah, sometimes it's just like, gives you a higher output score. But like Ali said, that's noise. To my knowledge, you're not necessarily making the results better. You're just trying to, again, like keep your fidelity as close to 100% to the original model.Layer Selection, KL Divergence, and Better QuantizationAli [00:29:27]: There is, to your point, research that we did on MP. I don't know if you are able to pullPhilip [00:29:31]: YeahAli [00:29:32]: A tweet we did. One of our research interns, Joshua, I think it's a tweet on how we have 20% better quantized GLM-5.2 than NVIDIA. Essentially what we found throughout like this month research is, okay, quantization is a lossy. It's. You're compressing the data from, occupying 16 bits to occupying, four bits, for instance. And so you're losing some information, and you're trying to minimize that. And so when I say that I'm gonna quantize the model, my job becomes how do I find the layers that I can quantize, and how to find the layers to not. For instance, with image models, I don't quantize modulation layers, and I don't quantize out projections because those two are. Like out projection is what you see as the user. Modulation is what the model sees or understands. Right, exactly. And so to his paper, do you have the. It doesn't have the. Yeah. It's a long paper. I don't know if I can findVibhu [00:30:25]: If there's a part to search or it's probably in the thread.Ali [00:30:28]: It's probably in the thread.Vibhu [00:30:29]: Yeah.Ali [00:30:29]: But the long and the short is it is very possible that quantizing more of the model makes the results. Like if I have a model that I quantize layers one, five, and 10, and another model where I only quantize layers one and It is possible that the model in which I quantized more information is going to perform better because the quantization errors have canceled out. And so what Joshua showed in his mathematical proof where he had like a verifier in, is that you can predict which layers are going to have quantization errors that will cancel out with each other, and you choose to quantize those layers. And so the result of doing this mathematical quantization is you end up with a model that's 20% more quantized than another provider, so you get 20% more throughput of it because there's more layers than running an NVFP4, and your quality is better than that other quant because the layers that you chose to quantize have their errors cancel out, like one layer skewed to the right one layer skewed to the left, one layer skewed to the right. Your final logits distribution is more similar to the original distribution of the model, so you have better fidelity. And so the way we proved this was with KL divergence. So instead of just scoring on the benchmarks, we scored the KL divergence between the logit distribution of the quantized model and the logit distribution of the original full precision model, and we showed that with this technique we get. If your probability distribution on the logits which token it wants to select is more of the same as the original model, you're probably gonna end up staying true to the original model. So yeah, so it seems like previously before this, it seemed like the industry was, well, the more you quantize, the worse it's gonna be, ‘cause the more loss you introduce. That's not exactly, not necessarily true. So yeah, doesn't improve it, but can cancel out.Philip [00:31:57]: I think it might be this, but reminds me a good bit about pruning where you can prune off certain layers.Philip [00:32:03]: But very interesting. Didn't know this was a whole paper you guys put out.Ali [00:32:06]: It's. Fun fact, it was originally 72 pages, this paper, and then we decidedPhilip [00:32:11]: WowAli [00:32:11]: We can't tell. We couldn't release it. So it's now 45.Swyx [00:32:15]: Still 39 pages, so very substantive. We talked about evals and all these things and, like what's possible in terms of speedup? Like it's like probably like the numberInference Speedups and BenchmarkingSwyx [00:32:25]: Thing that people do wanna care about, and it's something that you wrote about in your post. Like official API is 70 tokens per second, and you push it up to 90. Is that like a normal thing?Philip [00:32:36]: So what's cool about working in inference, the reason that I think inference is going to be a useful place to do engineering for a long time, is that if you look at highly optimized domains like, say, finance, if you're in finance, you measure how much better you got in basis points. It's like, “Oh, I got five basis points better, like twentieth of 1% better,” that's huge news because everything is so optimized. When we publish optimizations, it's 20%, it's 100% it's 200%. So there's still probably like a lot further to go, honestly. Like you'll, you'll know that inference is pretty much solved when researchers start publishing about how they got 1% faster at something.Swyx [00:33:19]: Which by the way, because I am from the finance background, in the ‘70s, that was the margin at the time. When you did quantitative finance research, you would findAli [00:33:27]: And like 20%, tens of percent.Swyx [00:33:29]: That's. Yes.Philip [00:33:29]: Yeah.Swyx [00:33:30]: And now it'Philip [00:33:31]: Tiny fractionsSwyx [00:33:32]: For those people interested, look up Andrew Lo's paper. He had a really interesting illustration of quant, stat arb, distribution, narrowing down from like those kinds of 20% differences in the ‘70s, down to nothing today, which is very cool.Philip [00:33:48]: Exactly, and we're at the beginning of the same type of thing. Now benchmarking is hard. I think anyone will tell you that, and benchmarking provider speeds is hard because there's so many variables that go into it. What hardware are you using? How much load do you have on the system? What's the exact nature of the prompts and input and output sequence lengths? All that stuff. But overall, when you start stacking these improvements, you're looking at multiples. You can look at it. The most common form, of course, is TPS, tokens per second, which is bad naming by us in the industry, ‘cause there's two tokens per second. There's tokens per second, the throughput number, and the latency number.Ali [00:34:31]: TTMT, yeah.Philip [00:34:32]: Like total tokens per second out of the, out of the GPU as a throughput number. Most people only care about tokens per second as the latency number, which we should call ITL, intertoken latency, but we don't.Philip [00:34:44]: Anyway, so you can imagine a standard API without many optimizations for a 1 trillion parameter model operating somewhere in the 30 to 50 tokens per second range for reasonable traffic profile. And we generally see the goal of, pushing to 10X that. But, not necessarily day zero, but by stacking enough optimizations, if you have, say like four optimizations, each of which doubles performance. Or sorry, three optimizations, each of which doubles performance, then you stack that up, that's an 8X gain. That's the order of magnitude that we're working with in this space. We're trying to make things substantially faster, not just go from like 70 to 90.Swyx [00:35:38]: Are you saying you've. You have done that?Philip [00:35:40]: So let's say you have as a reasonable baseline, 30 or 40 tokens per second. You can achieve 10X that. So like on GLM-5.2, if you run it unquantized, perhaps on H100s even, and you're just using an off-the-shelf inference engine with no particular optimizations, no speculator, nothing extra around like KV routing, no disaggregation, you're, you're probably, yeah, looking at that like 30 to 40. You think that's like a reasonable baseline?Swyx [00:36:12]: Right. Right.Philip [00:36:12]: To get to something like 10X, there's a lot of trade-offs that you're making. If we're running at more like a 300, 400 tokens per second range, you are using the best hardware possible. You have a optimized speculator. You have done all of your quantization work. You are Seeing a pretty high cache hit rate. You are running with a reasonably small batch size and a parallelism configuration that is tuned for latency versus throughput, but it is possible. So the spreads that you see if you, like, go on artificial analysis or you go on OpenRouter and you look at, the worst provider to the best provider, oftentimes can hit that range. 10X is of course very aggressive. It's oftentimes maybe more of a four to six times improvement. But that's the performance that makes us really excited, is when we can get these huge gains, not just go from 70 to 90 tokens.Stacking Optimizations: NVFP4, Speculation, and DisaggregationAli [00:37:19]: It's also, like, hardware dependent. Like, ifPhilip [00:37:20]: YeahAli [00:37:20]: If you have a thing where you're serving it on just, like, a node of H100s and then you throw, like, you shard the model across, like, four nodes of B200s. Like, you can definitely increase the speed with just throwing more hardware at it. Like, normalizing for the same exact hardware and the same number of GPUs.Philip [00:37:35]: Yeah. Then you're looking at, like, a two to 4X improvementAli [00:37:38]: Right. RightPhilip [00:37:38]: Depending on the inference optimizations. So yeah, it's. Some of it's, what's the call, and some of it's who's the driver.Vibhu [00:37:46]: If you break down the two to 4X, say the example is run GLM-5.2Ali [00:37:51]: YeahVibhu [00:37:51]: On B200sAli [00:37:53]: YeahVibhu [00:37:53]: Single node, right? What's, like, the cost trade-off for effort to get, like, the last bit of juice out versus what should people just think of, right?Ali [00:38:01]: Spectre quantization. Yeah.Vibhu [00:38:03]: Spectre quantization.Ali [00:38:04]: That's, that's, that's like 95%. LikeVibhu [00:38:06]: And how far does that get you? And how easy is that for the average person to do? So say right I wanna throw the weights of GLM-5.2 on a node of B200s, how easy is it to find speculative decoder- decoder model or already quantized model? How much work goes into it?Philip [00:38:23]: If you're doing it up front, it's quite a lot of work. If you're doing it today, there's going to be people who have published things that you can just, you can just grab some NVFP4 weights. You can grab a speculator. Yeah, if we're thinking about, like, what are the 2Xs we're stacking, going from, BF16 to NVFP4 is, it's not quite a 2X, right? It's like. I think it's about, like, 30 to 40%, from 16 to 8, and then another 30 to 40% multiplied from, 8 to 4. So that doesn't quite get you a 2X, but, like, roughly a 2X. Speculator, roughly a 2X. Disagg on top of that if you're able to get enough hardware and put enough traffic through it, another roughly a 2X. And then you add in some, double-digit percent increase from having just a better runtime with, the latest kernels and stuff behind it. And that's how it stacks up.Ali [00:39:21]: YeahPhilip [00:39:21]: So building each of those, like, building the, quantized weights is, for someone who really knows what they're doing, hours to days of work. Building the speculator, again, like, hours to days of work. And the, disagg setup, hours to days. Well okay, but like once you haveAli [00:39:39]: Once set up. Once set up. YeahPhilip [00:39:40]: Yeah, getting disagg working for the first time, I'm saying, of course, is very difficult.Philip [00:39:44]: The marginal implementationAli [00:39:48]: Like, if you're just grabbing, like if you are a person, like just a normal consumer who has access to, like, a node of B200s and you're wondering, “How can I just host it myself?” You don't need to quantize the model yourself. There's always gonna be, like, an open source quantized checkpoint. NVIDIA's gonna push one out if no one else does. You. Usually, the providers will have their own spec dec that they've trained as well. You don't need to train your own spec dec. You can just use that as well.Philip [00:40:09]: Yeah. Like, GLM-5.2 has its own MTP.Ali [00:40:13]: Right. Right.Vibhu [00:40:14]: What's multi token prediction?Philip [00:40:15]: Yes.Ali [00:40:16]: I'm justVibhu [00:40:16]: Can you explain that?Ali [00:40:16]: I'm just an expert.Ali [00:40:18]: I can do it for you in case I get it wrong?Vibhu [00:40:20]: No.Vibhu [00:40:21]: Yeah, you should correct if we're wrong, but their multi-token prediction can be used for self-speculative decoding.Ali [00:40:27]: I'm not sure. I'm not gonna correct that.Vibhu [00:40:28]: Okay. I'm semi-confident in thatAli [00:40:30]: Okay. YeahVibhu [00:40:30]: But someone can check. But it's useful to paint the story of, okay, not just the average person, but say a company wants to switch from serverless inference I wanna throw this up on. I wanna rent some GPUs, throw it up. These are the steps you take to do significantly faster than just put it behind vLLM.Ali [00:40:48]: Right.Vibhu [00:40:49]: I was waiting for a mention of Dynamo.Vibhu [00:40:51]: I feel like, that's supposed to be the baseline that you measure against.Dynamo, KV Routing, and Disaggregation ToolkitsPhilip [00:40:55]: I would think of Dynamo as less of a box system and more of a toolkit for building with. So when we talk about doing aware routing, when we talk about doing KV offloading, when we talk about doing, PD disaggregation, Dynamo fundamentally is. By the way, Dynamo is an open source library from NVIDIA.Ali [00:41:17]: We've done a pod with KylePhilip [00:41:18]: OkayAli [00:41:19]: Kyle Cranin.Philip [00:41:19]: Cool. So then your listeners know then that it supports all the different inference frameworks. And it is multi hardware, which is interesting.Ali [00:41:28]: But it's just a router, it's not like an optimizer layer.Philip [00:41:30]: Yeah. All it does, like, what Dynamo is good at, it is a library for moving information around your cluster, around your hardware. So if you have, KV cache on one place and you need it to be somewhere else, Dynamo coordinates NIXL for you to move that around.Philip [00:41:49]: That doesn't mean that, like, out of the box, you just say, “Pip install Dynamo,” and then you get, like, a massive performance speed up. It's more of a developer toolkit.Ali [00:42:01]: Yeah. I would have said it would. It comes with a set of defaults that you can then swap out.Philip [00:42:06]: It does. If the industry at large, I think, was, like, rolling out all of these deployments, standard, then I think it would be, like, a credible baseline. But, we've got to, we've got to benchmark against, like, what we're seeing in the wild.Speculative Decoding Methods: Medusa, EAGLE, n-Gram, and Spec-SpecVibhu [00:42:23]: I did wanna talk a little bit more about PD disagg, because that is probably, like, number three after quantized and speculative decoding. In your book though, I was just gonna pull out the book.Philip [00:42:31]: Yeah.Vibhu [00:42:32]: Like section 522 on Medusa, 523 on EAGLEPhilip [00:42:35]: YeahVibhu [00:42:36]: 524 on gram.Philip [00:42:37]: It's 55, would be disaggregationAli [00:42:42]: Yeah. Well, no, I just wanted to dwell a little bitPhilip [00:42:44]: YeahAli [00:42:44]: The other. Like, so what do you choose to include? What do you choose to not to include? Because there was all these other techniques.Philip [00:42:51]: Yeah.Ali [00:42:51]: Are these still relevant? Because I think they came out, like, a year and a half ago maybe.Vibhu [00:42:55]: Medusa is quite old.Philip [00:42:56]: Yeah, Medusa's old.Ali [00:42:58]: It was old.Vibhu [00:42:58]: But is it in the book as a good, here'sPhilip [00:43:01]: BaselineVibhu [00:43:01]: Baseline vanilla understand it?Philip [00:43:02]: Like you should know this.Vibhu [00:43:03]: Like I read the paper, I'm like, “ it makes so much sense.”Philip [00:43:05]: Yeah.Philip [00:43:05]: So with the book, I had a couple goals. One was to give people just a working vocabulary for the space as a whole, and the other was to give them some intuition about how each of these techniques works. As I mentioned in my AI Engineer talk, which is the first public addendum to this, the speculation space has moved much faster than everything else. So yeah, even at the time that I wrote the book Medusa, I very much included as a way for people to understand how the space evolved rather than what the most modern technique is. And now of course, there's DFlash, dSpark. There's, there's newer techniques even than EAGLE, although EAGLE is still very commonly used.Ali [00:43:51]: SpecSpecta.Philip [00:43:52]: Yes. Speculative decoding.Vibhu [00:43:54]: What canAli [00:43:56]: Oh, it's a paper by Tri Dao and it's like, it's doing speculative decodingVibhu [00:44:00]: HuhAli [00:44:01]: For the speculative decoder.Philip [00:44:02]: Oh, in spec- oh my God.Ali [00:44:02]: It's literally just an another. It's like, yeah, that's the most simple way to explain it, and it seems like he got trivial speed ups there. But it seems that the complexity with training, it's almost like in our mind at least, it's almost as complex as training GANs. Like it's like a very delicate balance and oftentimes you, it's just but yeah, it's literally speculative decoding on speculative decoding.Vibhu [00:44:21]: Speculative.Ali [00:44:22]: Yeah. We saw this paper.Vibhu [00:44:24]: It's interesting, right?Ali [00:44:24]: Yeah.Vibhu [00:44:24]: I wouldn't even expect it to be very particular to train, I wouldAli [00:44:29]: Right.Vibhu [00:44:29]: The naive part of me is like, okay, train speculative decoder.Ali [00:44:32]: But like, and it makes sense, like the whole idea of speculative decoding is you. It's like, it's like almost like the iPhone auto predict version but for a normal model, right? Like you're just, you're just, generating three tokens and you're like, okay, I'll do prefill on them. And so you save those three turns for your original model. Now your speculative decoder is doing three turns of auto regression, so why not just have an even smaller model?Ali [00:44:53]: The other question there is what are the size of speculators? So say forPhilip [00:44:58]: Right. It's like a billion parameters.Ali [00:45:01]: Like for MiniMax, it's. Yeah. It's like one layer. It's like one 60th of the original model usually.Philip [00:45:06]: Yeah. I think we should do a paper when we get back to the office.Philip [00:45:10]: SpeculativeAli [00:45:11]: SpeculativePhilip [00:45:11]: Decoding.Ali [00:45:13]: No, it's, it does seem like how, when do you stop? But then it also seems like if you're able to train spec-spec decode for instance, right? Like if you're able to have a small model that is accurately predicts what the intermediate speculator is gonna predict, that is able to predict what the original target model's gonna predict, then why not just use that smallest model directly, right?Vibhu [00:45:34]: Yeah. This isAli [00:45:35]: Like it seems likeVibhu [00:45:35]: Adjacent to the routing problem.Ali [00:45:36]: Right.Vibhu [00:45:36]: Yeah.Ali [00:45:36]: Right.Philip [00:45:37]: The thing with speculators is one of the practical constraints on using them is that you do have to run a small model on the same hardware that you're running the big model on. There is a orchestration and resource competition problem inherent in that, and that is one of the constraints on speculation in general, is that draft tokens cost resources to create and cost software complexity to manage. And so if you have like infinitely recursive speculators, you add in quite a bit of that complexity on the actual implementation within the inference engine as well, not just in the training process.Vibhu [00:46:17]: I was gonna say, I would wonder if you could do similar, like distillation and pruning of, it's the same thing, it's just a model. Can we not just distill a lot of the weights, quantize the speculator, out of my domain? The question that also comes up is, this is all for big server workloads, right? How much of this applies to, say I have this MacBook, I wanna run Gemma really efficiently. Similar problems, not the same?Local AI vs. Data Center InferencePhilip [00:46:45]: Pretty different. I talked to Selo, about this on his podcast a couple weeks ago. The difference between inference engineering for the data center and for production workloads versus inference engineering for local AI, is that we start with fundamentally like different constraints and different goals. With local AI, it's how do I fit this model onto my hardware and then make it less dumb? And with data center influence, it's how do I load this model and then make it less slow? And we care about less dumb, and they care about less slow. But the local AI inference engineering ecosystem, I think has a lot for us to learn from in the data center space. They are experts in various forms of quantization, including dynamic quantization that we just don't touch, in the pruning, in the distillation, in the, layer removal. There'Ali [00:47:42]: Layer removal matters less.Philip [00:47:43]: Yeah. There'Ali [00:47:44]: No one loves pruning really.Philip [00:47:45]: Yeah. Well, but the, but they doVibhu [00:47:46]: Which is surprising, right? But that's, that's a whole different thingPhilip [00:47:48]: Just to fit something on the laptop.Ali [00:47:50]: Right.Philip [00:47:50]: So yeah, it's a, it's an interesting, it's an interesting space. Not necessarily that like their techniques make sense for us to do in the data center, because we have different resources and different goals, but more that the process as well as the openness of that field is something to, admire.Ali [00:48:12]: Yeah. Like to your point, like, certain optimizations that would. Like for instance, Turbo Quantum Sharper, like it made such huge hype on that and we did like a whole deep dive on Twitter and like said, what is it? How does it work? Why is it good or not? And it took off and it was implemented on local devices because your memory bandwidth is so slow on like a MacBook, for instance. But try putting the same thing on like an NVIDIA GPU on a B200 Turbo quant would not be. Like, it would not be used. Like, NVIDIA - Like, NVIDIA made it clear that this is not a good optimization, and we've seen it firsthand where the overhead of doing dequantization, quantization of, in the kernel itself with turbo quant kernel, each end is much slower than the time that you save from doing the bandwidth. ‘Cause on the B200s, you have like 3.5 terabytes per second. You don't need decrease the storage that much. You don't need to do, FP4 KV cache. You don't need to use a requant. There's, there's, there's better optimizations to be made. But on Edge devices, it's extremely important, it's extremely useful. So, seems to be, like, different optimizations there, but then they're all uniquely combined with like all you wanna quantize the model, you wanna do speculative decoding, like certain common prefixes with bothPhilip [00:49:18]: Principles.Ali [00:49:19]: Yeah, exactly. Exactly. Exactly.Philip [00:49:20]: They also do a lot of work on, model parallelism, especially over, heterogeneous topology, where you have, some sparks and they are wired together with, Ethernet, DGX sparks.Ali [00:49:35]: Yeah, this is the Exo Labs guys.Philip [00:49:36]: Yeah. You have, a nu

Welcome to the latest episode of Harmonious World, where I interview musicians about how their music helps make the world more harmonious.I met drummer Roan Anderson while at the Jazzahead! 2026 Trade Show and Festival in Bremen. We chatted then and it was great to have a proper conversation for this episode in the cafe at Kings Place in London.Roan won the BBC Scotland Jazz Musician of the Year competition in 2025 and has already - just two years out of the Glasgow Conservatoire - been playing with lots of exciting new ensembles. There's clearly great things to come in his musical future.Thanks to Roan for allowing me to play extracts from his two latest singles - Swept Away and Emotionally Intelligent - alongside our conversation.Get in touch to let me know what you think!Thank you for listening to Harmonious World. Please rate, review and share: click on the link and subscribe to support the show.Don't forget the Quincy Jones quote that sums up why I do this: "Imagine what a harmonious world it would be if every single person, both young and old, shared a little of what he is good at doing."Support the showRead reviews of albums and gigs and find out more about me at hilaryseabrook.co.ukFollow me on instagram.com/hilseabrookFollow me on facebook.com/HilarySeabrookFreelanceWriterFollow me on twitter.com/hilaryrwriter

A fun look back at what some of the other 90's sitcoms looked like.  Did you watch any of these shows?Coach1989-1997 Season 7 ep 19JohnsonwreckersDrew Carey Show 1995-2004 Season 3 ep 26From the Earth to the MoonAlf 1986-1990 Season 2 ep 2Somewhere over the re-runRoseanne1988-1997, returned in 2018,  The Conners 2018-2025Season 6 ep 4A stash from the past 

Andy watched The Fantastic Mr. Fox, and Patrick watched Split Second.  One of those movies is better than the other.

Join us on a brief but fun side quest with the horror adjacent film THE RAVEN (1963) from Roger Corman! Loosely based off the Edgar Allan Poe poem, and featuring horror icons like Vincent Price, Boris Karloff, and Peter Lorre, the flick is a fun albeit silly time at the movies. Context setting 00:00; Synopsis 23:21; Discussion 31:14

Vince is back from the Trojan War and he has more tales of Secret Mormon Wives. How could we ever do it without him? By the way, Swayze would've played Odysseus's dad, Laertes (neither of them were in the movie)

What if I told you that most jobs in wildlife, conservation, and environmental science aren't directly in research? Would you be shocked? Well, it's true! And that's exactly what I talk about in this episode of the Fancy Scientist Podcast. When most people think of wildlife work, they think of jobs in wildlife research. Those are jobs where you pose questions or objectives, and then collect, process, and analyze data to address them. But when you compare these jobs to all of the others in the realm of wildlife, conservation, and environmental sciences work, they are actually in the minority. I am calling these non- ”traditional” jobs wildlife research-adjacent work or even just wildlife-adjacent jobs, even though they relate to conservation and the environmental sciences, and some are definitely in the wildlife field. This was a topic suggested by members of my Getting a Job in Wildlife Biology Facebook group a couple of years ago because there is strong interest in them, so I thought it would be the perfect topic to take a deep dive into for the podcast because it is a big one. In fact, it's WAY too much to cover in just an episode. This is by no means a full list, and there's not enough time to go into them in-depth. For example, in my group mentoring program, the Successful Wildlife Professional, we spend at least 30 minutes for each job and organization type. I'll walk you through the major categories: data analytical and processing roles like GIS, statistics, lab work, and computer science; communications, outreach, fundraising, and education jobs; wildlife health careers, rehab, and vet work; zookeeping; more environmental science roles in water quality, forestry, and agriculture; corporate sustainability positions; and outdoor recreation jobs. But first, before you even consider a wildlife-adjacent job, I ask you to question WHY you're pursuing one of these jobs in the first place. Is it because it is truly a job you want? Or is it because you feel like you can't get the job you really do want? In my experience working with individuals seeking these kinds of jobs, it's usually the latter, and if that's the case, I encourage you to go after what it is you really want because this is a fixable problem. I identify the two big reasons that people can't get wildlife jobs in the first place and what you can do about them. Finally, I'll leave you with this: wildlife jobs have a reputation for being competitive and low-paying, but that's not universal, especially when you consider adjacent jobs. In a recent podcast interview with Patrick Rainey of Ducks Unlimited, we talked about some wetlands restoration jobs that are in demand and also well-paying. So if getting a high salary is important to you, you might want to explore wildlife-adjacent work. Specifically, we cover:What "wildlife-adjacent" work means and how it differs from traditional research careersThe two biggest reasons people fail to land wildlife jobs, and why that matters before you consider an adjacent pathWhen pursuing a wildlife-adjacent job is a smart move even if it's not what you really want, and how to prevent it from becoming a trapAnalytics jobs, including GIS, statistics, lab work, and computer science rolesCommunications, outreach, fundraising, and education jobs, including science writing and ecotourismWildlife health careers like rehabilitation and vet workZoo and aquarium roles outside of research Environmental science jobs in water quality, forestry, agriculture, and moreMuseum workCorporate sustainability roles, even within companies you wouldn't expect to have a conservation focus, and how this can indeed make a big difference for wildlife Outdoor recreation jobsWhy some wildlife-adjacent fields are actually in high demandAnd more!Jump links:2:21 What Are Wildlife Research-Adjacent Jobs?4:25 Common Job Search Mistakes16:19 Data & Analytics Jobs21:39 Communications & Education Jobs28:19 Wildlife Rehab & Vet Careers30:51 Zookeeping & Aquariums35:11 Forestry, Agriculture & Museums37:05 Private Sector and Corporations37:47 Final Takeaways Dream of being a wildlife biologist, zoologist, conservation biologist, or ecologist? Ready to turn your love of animals into a thriving career?

I created Do I Know You? because growing up with a famous sibling comes with a whole lot that nobody talks about... and Dana brought the jokes AND the wisdom!!Dana is a standup comedian, woman who does it all, and yes... Miles Teller's big sister. But five minutes in, you'll forget all about that last part.We talked about the wild whiplash of "celebrity adjacent" life (private jets to studio apartments, anyone?), the terrifying reality of teenage superfans, and why she left a career in finance and accounting to chase her standup dreams. We also got into the real stuff: bombing on stage, learning to take a note without losing yourself, "friendship pruning," and why we believesnothing in your life experience is ever wasted.There were laughs (duh), a little TMI about Top Gun Maverick, and a whole lot of "it's never too late." Come sit with us!!! xoxoxoxoFOLLOW DANA AT @DanaTellerFOR FUNSIES:Instagram: @jordierae_Show Instagram: @doiknowyoushowTiktok: @jordierae_LET'S TALK BUSINESS:

Welcome back ladies and gentlemen to the Woody Allen Retrospective Podcast!   Before we get into this month's adjacent title, we've got quite a bit of podcast business to get through.   Today is pretty much the final episode of what we'd call the second era of the show, well, aside from next month where we will be discussing the latest Woody Allen book - that will TRULY be the final part of the second era of the show.   The first era was the early years with Simon, when we were still figuring out the format, the editing, using the very ugly Microsoft paint looking artwork and, quite frankly, we were still figuring out what the podcast even actually was doing long term.   This second era has been the James Daniel Walsh years, where the show found a new rhythm and started the Woody Adjacent Project.   ...and now we're getting ready for the third and final act of the podcast.   From October we begin the Woody Allen RE-retrospective, where we'll be going back through every single Allen film or off cut odd special release ala non woody directed movie etc   We will be hitting you with fresh conversations, improved production and everything we've learned from doing this podcast for so many years.   Then, when that final retrospective is complete, that will bring this podcast itself to a fitting end.   The podcast feed will still be active but myself and james will likely be starting a new podcast about film and television (to be decided later)   We really want to make this last run of the show the best version the podcast it possibly can be.   As part of the new final era, we've got brand-new podcast artwork, a dedicated YouTube channel, a new email address and a number of other updates coming as we freshen everything up for the road ahead.   Today, we also explain why August is going to work a little differently moving forward.   August is always a hectic month for me personally, with family birthdays, holidays and a lot of podcast admin happening behind the scenes, so moving forward, August will essentially become our annual pre-recorded month.   There will still be an episode in August, but it may be a solo recording by me Don, something James and I have cobbled in advance, or a different type of episode that gives us the space to work on anything else woody related - at the very least it will be a Q&A episode / AMA type episode.. we'll figure it out later   Putting all that aside.. we also make a small request to our listeners.   James has been using the same microphone for around ten years, going all the way back to outside work he was doing on his manic expression animated movie,   Before we begin the final retrospective go around I would really love to upgrade his audio gear.   The microphone I've picked out for him is the Elgato Wave 3, which should give him a substantial improvement in sound quality and help us make this final run through on Woody's filmography sound as professional as possible.   This request is not aimed at our existing Patreon supporters, who already do a great deal for the podcast and have our genuine thanks. This is simply an opportunity for anyone else who enjoys the show and has ever wanted to support us directly help out.   Whether that means making a small one-off contribution, becoming a new Patreon or even donating suitable equipment to us directly, every bit of support would help us get James the updated microphone he deserves.   ==========================================================   James also shares some very exciting news of his own.   He has already begun an ambitious fifteen-month rollout of new creative work, with fresh material being released every single week and, at times, several releases within the same week.   By the time this episode reaches you, his stories Bloodshot, The Last Spike and Faded should all be available through Amazon, alongside his original song Constellation, which can be watched for free on the Manic Expression YouTube channel.   other links to James's work, the microphone fund and the various ways to contact or support us will be included below.   SUPPORT JAMES HERE    >>>>>>>>>>>    https://linktr.ee/manicexpression1980   ==================================================   That all said, we now return to the Woody Allen Adjacent Project with a listener-requested film that we've wanted to discuss for quite some time :)   Good Luck to You, Leo Grande is the 2022 comedy-drama directed by Sophie Hyde, written by Katy Brand and starring Emma Thompson and Daryl McCormack.   The film follows Nancy Stokes, a retired teacher and widow who decides that there are certain experiences she has missed during her life and may finally be ready to pursue.   What follows is an intimate, largely two-person film built around a series of meetings between Nancy and Leo Grande   In this sex dramadey we consider the strengths of contained, dialogue-driven films, the natural chemistry between the two leads, and the question at the heart of every episode in this series: What exactly makes thisWoody Allen adjacent?   Thank you, as always, for reading, listening, supporting the show and joining us as we prepare to enter the third and final era of the Woody Allen Retrospective Podcast.   That was one hell of a long read aye :P - we really do hope you have a great rest of your summer!   ==================================================   For full cast details, user reviews, and more background on the film, check out:  

It was a heavy week of content for Andy and Patrick...

Send us your Florida questions!Stephen Cabebe, of Fit for the Magic, joins the podcast to talk about how to enjoy Florida's natural spaces without withering in the heat — even in the summer months.Stephen's Top Five Disney-Adjacent Things To DoPaddle the canals in Winter Park (Dinky Dock)West Orange Trail (And Cathy's travel article about Oakland Nature Preserve along the trail)Rock Springs at Kelly ParkHorizon West Regional ParkPoke HanaMills 50 District (Mills Market)Links We MentionedFit2RunRobinson PreserveFort De SotoVeggie GardenSampaguitaFlatlands Bar & Grill Support the showQuestion or comment? Email us at cathy@floridaspectacular.com.Subscribe to The Florida Spectacular newsletter, and keep up with Cathy's travels at greatfloridaroadtrip.com. Keep up with Rick at studiohourglass.blogspot.com and get his books at rickkilby.com.Find Cathy on social media: Facebook.com/SalustriCathy and everywhere else as @CathySalustri; connect with Rick Facebook.com/floridasfountainofyouth, Bluesky (@oldfla.bsky.social), and IG (@ricklebee).NEW: Florida landscape questions — Send us your Florida plant questions and we'll have an expert answer them on the show! Use this link!

World Cup Final adjacent topics, Mets and Yankees weekend recap

See more at >>https://nemosnewsnetwork.com/andrew-tate/ Andrew Tate's 15 yo GF>> -https://old.bitchute.com/video/sNZhMBIUQHvx/If you appreciate the work we do and wish to support us, you can donate here >> https://www.nemosnewsnetwork.com/donateBitchute – Where We Don't Have To Watch Our Mouths!Click Here For Exclusive Deal and Remove all ads and secure your privacy!https://www.bitchute.com/affiliate/dustinnemos

Andy watches Sheep Detectives, and Patrick talks about Backrooms

Get out your gators and lets see some weird birds, Nerd Talk+ travels to Florida to drink in the morning and make podcasts and trouble. Please be advised that the editing for this is rough and some dodgy stuff gets through and you're gonna live and so are we (maybe). Thanks or your support, enjoy the bonus episode! nerdtalkplus.com

See omnystudio.com/listener for privacy information.

Emergency episodes call for Emergency Rooms. Mal and Josh both try out their Vince impressions to fill the void left by his absence.

Andy and Patrick both watched Supergirl...and they have thoughts.

The episode prioritizes the operational and exit-planning risks associated with MSPs lacking formal contracts. According to Amy , approximately half of MSP business owners operate without managed service agreements (MSAs), a decision that frequently results in reduced business valuation during sale negotiations. Both James and Amy emphasized that the absence of documented agreements is commonly flagged by buyers as a significant risk, often resulting in a devaluation of the acquired customer relationships. This exposes small and mid-sized providers to continuity risks, particularly where customer retention and service transferability are not contractually secured. Further details outlined by Amy indicate that reluctance to implement contracts stems from concerns about client reactions, particularly in longstanding relationships. She observed that auto-renewal clauses and periodic, non-intrusive contract updates can streamline compliance and reduce friction. A personal account highlighted that, out of numerous customers, only one refused to sign an agreement, and this isolated case did not lead to client loss but necessitated risk pricing adjustments. James Kernan advised that contract clarity—covering terms, automated payments, and built-in annual price adjustments—should be positioned as a value to both parties, reinforcing operational stability and predictability. Adjacent topics addressed contemporary service risks such as the proliferation of shadow AI applications and exposure to business email compromise. Amy reported discovering over 150 unmonitored AI-powered apps at client sites, emphasizing these as vectors for data exfiltration and compliance gaps. The Guards Cybersecurity Statistics report was cited, identifying business email compromise and social engineering as persistent attack methods, while underscoring that modern threat actors often bypass traditional privilege escalation in favor of capturing identity credentials and tokens. The operational focus for MSPs was advised to shift toward email, AI governance, and identity protection rather than legacy device vulnerabilities. For MSPs and IT service providers, the main takeaways involve reassessment of contractual practices and a heightened approach to governance and risk management. Documented client agreements are necessary not only for valuation at exit but also for protection against operational disruptions and liability. Simultaneously, providers are urged to implement discovery and control mechanisms for AI use and to refresh security postures in line with current attack methods. The importance of establishing relationships with specialized advisors, attorneys, and alternative financing partners was also articulated, illustrating the multi-layered risk landscape that management teams must navigate for business resilience.Show title: “How to Start a MSP” 1. How to Start a MSP- Resources: www.itspu.com 2.    New article from Third Tier: Taming Shadow IT before it tames you https://www.thirdtier.net/2026/06/21/taming-shadow-ai-before-it-tames-you/ 3.    Guardz released a new cyber security statistics report: https://guardz.com/blog/security-awareness-statistics-msps-cant-ignore/ 4.    Thoughts about whether an MSA is needed? Send them through the website: www.smbcommunitypodcast.com 5.    Anthropic Partner Program - https://www.anthropic.com/news/services-track-partner-hub   Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

The term “stealthing” is making headlines after Maine Democrat and Senatorial Candidate Graham Platner has been accused by an ex-girlfriend of repeatedly removing condoms without her consent during sex. We did a deep dive on the subject and were fascinated and frustrated by the lack of recognition from lawmakers in this country about NCCR: Nonconsensual condom removal. Several countries around the world have strict, criminal code on stealthing, categorizing it as a “rape adjacent” crime. Right now, only a handful of states allow victims to pursue civil damages, but not a single jurisdiction in this country considers stealthing a crime.See omnystudio.com/listener for privacy information.

The term “stealthing” is making headlines after Maine Democrat and Senatorial Candidate Graham Platner has been accused by an ex-girlfriend of repeatedly removing condoms without her consent during sex. We did a deep dive on the subject and were fascinated and frustrated by the lack of recognition from lawmakers in this country about NCCR: Nonconsensual condom removal. Several countries around the world have strict, criminal code on stealthing, categorizing it as a “rape adjacent” crime. Right now, only a handful of states allow victims to pursue civil damages, but not a single jurisdiction in this country considers stealthing a crime.See omnystudio.com/listener for privacy information.

The term “stealthing” is making headlines after Maine Democrat and Senatorial Candidate Graham Platner has been accused by an ex-girlfriend of repeatedly removing condoms without her consent during sex. We did a deep dive on the subject and were fascinated and frustrated by the lack of recognition from lawmakers in this country about NCCR: Nonconsensual condom removal. Several countries around the world have strict, criminal code on stealthing, categorizing it as a “rape adjacent” crime. Right now, only a handful of states allow victims to pursue civil damages, but not a single jurisdiction in this country considers stealthing a crime.See omnystudio.com/listener for privacy information.

The term “stealthing” is making headlines after Maine Democrat and Senatorial Candidate Graham Platner has been accused by an ex-girlfriend of repeatedly removing condoms without her consent during sex. We did a deep dive on the subject and were fascinated and frustrated by the lack of recognition from lawmakers in this country about NCCR: Nonconsensual condom removal. Several countries around the world have strict, criminal code on stealthing, categorizing it as a “rape adjacent” crime. Right now, only a handful of states allow victims to pursue civil damages, but not a single jurisdiction in this country considers stealthing a crime.See omnystudio.com/listener for privacy information.

Another bye week, another bi week. This time: EVEN MORE BISEXUAL. Alicia and Mal are beloved guests on the network, so here's a bonus helping of them both. Listen to the end to meet your Swayze Scale (SWAYZE SCALE) needs. EVEN MORE Bring It On content is available at Patreon.com/nooneworkhere, where we watched ALL the Blu Ray bonus features, including the director's commentary (alone. he watches it alone), and a second conversation with our producer, Bud, after his first-ever viewing of this classic bisexual film.

Pirates have stolen supplies vital to the Rebel Alliance. Our intrepid band of heroes have been tasked with getting the supplies back.Come check out all the people that made this possible!https://www.recklessattack.comWant to hang out with other fans?Join us on Discord!Like the show? Support us on Patreon! https://www.patreon.com/recklessattack

Writer Ed Brubaker joins the show to talk about his long collaboration with Sean Phillips and the recently released Five Gears in Reverse. Brubaker discusses the status of the Criminal show, the state of TV, why Sean Phillips fits him so well, how they push each other, his favorites from their run, what fuels their new projects, Criminal as a story engine, Five Gears in Reverse's origins, the fusion of their approaches, the development of its cast, the roots of Unfinished Tales, operating in their own space, how their books have sold of late, and more.

DescriptionA shift I am making in the podcast is NOT only to speak with speakers about the art and science of preaching, I am also looking to interview thought leaders ADJACENT to preaching and teaching. Well today, my guest Julianne Stanz, is both! She is a gift to the Church and I'm so grateful to get a chance today to hear beautiful stories from the heart of the Church: from her first time meeting Pope Leo to sharing what, in her opinion, has led to such growth in the Church, and also where ‘the puck is headed', if you will. Julianne Stanz is a storyteller, bestselling author, and encourager whose heart for ministry has inspired and equipped audiences across the globe. With more than two decades of distinguished service in church leadership, both nationally and internationally, she has partnered with dioceses, bishops, clergy, and lay leaders to advance the mission of the Church. She currently serves as the Director of Outreach for Evangelization and Discipleship at Loyola Press and as a consultant to the United States Conference of Catholic Bishops (USCCB), providing strategic guidance on evangelization, missionary discipleship, and pastoral leadership. She is the author of several acclaimed books, including her latest: The Catholic Parent's Survival Guide: How to Answer Your Children's Toughest Questions about Faith (all published by Loyola Press). Julianne is married to Wayne, a Catholic school teacher, and together they are raising three children.Check out Julianne's website right here: www.juliannestanz.comAnd here's the ink to her latest book, “Catholic Parents Survival Guide.” Check it out!LinksFor more information about the Better Preach Podcast visit: www.ryanohara.org/betterpreachBetter Preach Podcast is now on YouTube. Here's a link to the channel.Check out Ryan's FREE course on “sharing your faith as a Catholic.”Follow Ryan on Instagram, Twitter, YouTube, or FacebookJoin the Better Preach email list.

Link Up w/The Morning Sickness Digitally All Over:Instagram: @hms_98_official, @bosskupd, @bretvesely, @dickToledoX/Twitter: @HMSon98, @DickToledo, @bretveselyFacebook: @HMSKUPDYouTube: @hmspodcast9320, @98kupdRequest/Call in/Wakeup Song line:(IN AZ) 602.585.9800More HMS: holmbergpodcast.com, 98kupd.comEmail: dtoledo@98kupd.com, bvesely@98kupd.com, bbogen@98kupd.comSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Link Up w/The Morning Sickness Digitally All Over:Instagram: @hms_98_official, @bosskupd, @bretvesely, @dickToledoX/Twitter: @HMSon98, @DickToledo, @bretveselyFacebook: @HMSKUPDYouTube: @hmspodcast9320, @98kupdRequest/Call in/Wakeup Song line:(IN AZ) 602.585.9800More HMS: holmbergpodcast.com, 98kupd.comEmail: dtoledo@98kupd.com, bvesely@98kupd.com, bbogen@98kupd.comSee Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Al updates Jerry on all the latest Sports-Adjacent stories

Welcome back to the Woody Allen Retrospective Podcast for another episode of Woody Allen Adjacent!   As we get ever closer to the re-retrospective coming this October 2026 we carry our Adjacent series and this month it's James's pick he's chosen Stanley Tucci's possibly overlooked 1998 comedy The Impostors—a loving homage to the golden age of slapstick, screwball comedies, and classic double acts like Laurel & Hardy and the Marx Brothers.   Packed with an incredible ensemble cast including Oliver Platt, Alfred Molina, Steve Buscemi, Tony Shalhoub, Lili Taylor, Isabella Rossellini, Billy Connolly and even a brief appearance from Woody Allen himself, this is one of those films that has quietly built a passionate cult following over the years… while leaving plenty of other viewers completely baffled.   Along the way, Don discovers that he may never have truly understood what defines a farce until finally sitting down with this film, while James makes the case that Hollywood simply doesn't take creative swings like this anymore. Whether The Impostors is a forgotten comedy gem or an ambitious experiment that doesn't quite come together… it certainly gave us plenty to talk about.   Thanks as always for listening!   ==========================================================   Links mentioned in this episode:   James Content Rollout Announcement Trailer - https://www.youtube.com/watch?v=OEFovWooMuA   Stanley Tucci's The Impostors (1998) Interview - https://freshairarchive.org/segments/actor-director-and-writer-stanley-tucci-imposters   ==================================================   For full cast details, user reviews, and more background on the film, check out:  

The episode's principal focus is on managing employee compensation requests within MSP operations, specifically addressing scenarios in which employees request significant raises. The discussion highlights the need for MSP leaders to weigh current market rates and organizational financial constraints in response to such requests. Both collaborative problem-solving—engaging employees in discussions regarding revenue generation—and performance-based compensation structures were identified as practical strategies for balancing employee retention with business viability. Supporting this, the speakers emphasized transparent conversations with employees about the company's billing structure, the sources of revenue, and the tradeoffs required to increase compensation. Performance-based incentives, such as commissions or bonuses tied to measurable business milestones (e.g., revenue targets, net new business), were noted as effective mechanisms for aligning individual rewards with organizational profitability. The discussion also noted risks when negotiation dynamics become adversarial, underscoring the importance of maintaining professionalism in both management and staff behaviors. Adjacent topics included the increasing relevance of AI proficiency for workforce retention and the emerging terminology in the sector, such as “Managed Intelligence Provider” (MIP). It was observed that employees lacking AI skills face elevated layoff risks, a factor attributed to shifting operational priorities and the need for continuous learning. While new acronyms like MIP may gain industry attention, participants agreed that most services related to intelligence and data management could be incorporated within the MSP framework, suggesting minimal structural change for service providers. For MSPs and IT leaders, the session underscores the necessity of clear compensation policies, ongoing staff development—especially in AI—and cautious positioning during merger or acquisition discussions. Risk mitigation includes frank financial communication with staff, investments in workforce upskilling, and careful negotiation to avoid undervaluation during business transitions. Performance incentives and clear cultural alignment are practical steps to balance employee satisfaction with organizational sustainability.1. MSP Question of the week: My employee wants more money- what should I do? 2. A single decision that triples some workers layoff risk – Tech workers who regularly use artificial intelligence tools are far less likely to be laid off compared to colleagues who use AI less frequently, Bloomberg reports, citing new Gallup research. It puts infrequent users' layoff risk at 18%, tripling the 6% layoff risk for frequent users. Earlier this year, researchers tracked how often 23,000 employed and displaced workers used AI 3. The RISE of the Managed Intelligence Provider 4. M&A: How do I find the right buyer? Amy's Book: https://amzn.to/4dSYOcR 5. Tales from the Field: Merger conversation – make sure it's a culture fit and be fair with the price (no lowballing)   Mastermind Event – July 30-31st, 2026 Register here: https://portal.kernanconsulting.com/mastermind-event Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

WarRoom Battleground EP 1034: Leading German AfD-Adjacent Youth Newspaper Suppressed By WhatsApp 24 Hours After Coming On WarRoom

Comedians Mike Goldstein and Lizzy Hoo talk all things grass season and try to solve the bone-chilling mystery of 'Why Were There Loose Coins in the Shower?' Watch in disgust as the butcher the pronunciation of the names of several international tournaments, and then decide once and for all which sport is harder to play: Tennis or Golf? The answer, will shock you. It definitely shocked me.It's Pod Laver Arena! The only podcast that's been asked to record on the lawn of the White House (we declined the offer)! AusOpen.comiHeartApple PodcastsSpotifyYouTubeSee omnystudio.com/listener for privacy information.

A significant regulatory development affecting MSPs was discussed: the US Government ordered the suspension of access to the Fable 5 and Mythos 5 AI models by any foreign national, including those inside and outside the United States, citing national security authorities. As reported in the Anthropic company statement, this directive forced abrupt discontinuation of these AI tools for all customers, regardless of business impact. The decision resulted in the sudden loss of access to custom-built AI applications and business process automation tools that MSPs and their clients had integrated into daily operations. Immediate disruptions included the cessation of SEO and analytics engines, RMM automation, and bespoke backup solutions that relied on the now-restricted AI platforms. Further clarification showed these suspensions are tied to concerns about potential backdoor access, disputed by Anthropic but acted upon due to US Government findings. Additional context revealed Amazon—the largest investor in Anthropic and a direct competitor—alerted federal authorities to the supposed vulnerability. Stock prices of competitive AI offerings in China reportedly rose by 48% following the announcement, indicating market reactivity to perceived US regulatory risks. The episode underscored that AI model dependencies—whether managed internally by MSPs or by third-party vendors—can introduce sudden continuity hazards if access or legal standing is rapidly altered. Adjacent discussions evaluated operational models for MSP service offerings. Contrasting perspectives highlighted the tradeoff between providing a single comprehensive managed services plan, designed for streamlined staff training and high-touch customer experience, versus offering a tiered set of plans (“good, better, best”) that, according to shared data, can result in about 70% higher revenue through client segmentation and option-based sales. The choice was framed as fundamentally cultural, influencing both workforce structure and scalability, with differing risk and complexity profiles for technical delivery and sales management. Key implications for MSPs and IT leaders include the need for explicit risk assessments around reliance on AI platforms and third-party tools. Business continuity planning should contemplate not only technical redundancy but also legal and regulatory exposures to abrupt vendor or governmental action. When building service portfolios, organizations should align plan standardization or diversification with internal capacity, capability for sales-driven growth, and staff training. Establishing clear governance for evaluating the ongoing viability and risk exposure of both internally developed and vendor-supplied technology is critical for operational resilience in an environment of rapid regulatory change. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

This week, Templeton Elliott and Jason From Frozen In Carbonite are speaking with Alex Fazekas-Boone about Adjacent 2026 and skate games.

A former MLB player who coaches travel baseball says winning tournaments is quietly stunting your kid. Subscribe for the insider playbook.   Most travel baseball parents measure a weekend by the scoreboard. Danny Espinosa measures it differently. A former MLB infielder, Long Beach State Dirt Bag, and owner-coach of the OC Crush, Danny sat down with MLB agent Matt Hannaford to explain why a team can win every tournament and develop almost no one. The conversation opens with something Danny witnessed at a 9U event: coaches stealing signs and relaying pitches to nine-year-olds. When he called it out, a coach told him that he should get on board because this is the new age of travel ball. The point that follows is the one you need. Relaying signs may win a game, but it teaches your kid nothing about how to develop properly. From there, Danny and Matt separate two words parents constantly confuse: advanced and developed. The biggest, strongest 10-year-old usually succeeds early. That is not the same as the player who learns the game properly and keeps growing at 16, 17 and 18. Danny explains why he refuses to cut kids off his own roster, why he would rather a young player build strength and athleticism than obsessing over mechanical adjustments he is not physically ready to repeat, and why Freddy Freeman, whose son plays on Danny's team, preaches the importance of not over-coaching. If you have ever wondered whether your kid needs the best private hitting coach in the area, this section answers it. The most expensive mistake in youth baseball, according to this conversation, is chasing exposure. Matt makes the insider case directly: exposure does not matter until your child's junior year of high school, around 16 or 17. Before that, Danny asks the question that often stops parents in their tracks. Exposure to what? Your local high school will take the best players, regardless of how many showcases you paid for. The episode reframes the obsession on parents to spend. Put development first, and exposure becomes a byproduct of doing everything else well. Matt and Danny also work through the questions parents need to ask. Should your kid specialize in baseball or play multiple sports, and why did Bo Jackson's answer surprise a guy who believed the opposite? How many games is too many across across a season? Why are holdbacks a problem for some, especially when it's done too early and the result is a 13-year-old gets hit a line drive at 50 feet, and how might the NCAA five-and-five rule correct it? Adjacent topics include college recruiting, the transfer portal, scholarships, NIL, the MLB Draft and showcases. It ends where it should. Danny explains why he never talks to his sons in the car after a game, and what his own parents told him that he now repeats to his kids: whether you play one more day, I will always love you regardless of the outcome. If you are deciding how much to invest in your child's baseball, this conversation will change your perspective.   About Matt Hannaford is an MLB agent who gives you the insider playbook on college recruiting, the transfer portal and MLB Draft decisions. The Most Valuable Agent Podcast helps parents and players navigate the system with confidence.   Links Subscribe: https://www.youtube.com/@mostvaluableagent MVA Website: https://www.aligndsports.com/ Instagram: https://www.instagram.com/mfhannaford/    #MVAPodcast #TravelBaseball #YouthBaseballDevelopment #CollegeBaseball #MLBDraft

Podcasting 2.0 June 5th 2026 Episode 262 - "Podcleanse" Dave and Adam are joined by John Spurlock and throw a big idea into the boardroom: The Podcast Data Collective Shownotes ----------------------------------------------------------------------------------------------------------------------------------------- John Spurlock - Guest The man behind op3.dev and Livewire.io - From the Great State of New Jersey! ----------------------------------------------------------------------------------------------------------------------------------------- 01 - THE IMPRESSION HEIST — AMP TASK FORCE RATIFIES 4 EXPOSURE DEFINITIONS, NO DISSENTING VOTES Podnews press release Jun 4: AMP Task Force Introduces Cross-Platform Alternative to the Podcast "Download" — "unified impression guidance for audio and video, advancing impression-based measurement as the medium's primary transaction currency." Four exposure definitions ratified. JS Jun 4 quote: "the AMP Task Force ratified a new framework with four exposure definitions, with no dissenting votes." Podcast Play: 30 seconds of content played, audio or video, once per user per session. Podcast Audience: The number of unique users who had a Podcast Play. Ad Impression: A commercial begins playing for the user. Ad Audience: The number of users exposed to an Ad Impression. They wanted to 'hasten the demand' Backstory: AMP first emerged May 29 (Podnews) — same day PC20-261 aired — "to confront podcasting's measurement dilemma." @dave reaction Jun 4 16:12: "RE: [Podnews AMP story] More secretive, back room podcast 'industry' nonsense." PNWR Jun 5 confirms the cabal-composition critique — James and Sam open the show debating AMP. James: "they also want to define what an impression is" + "we don't have a definition of podcast." Sam: "I don't think podcasting is [defined], we can measure consumption." PNWR catches the gaps [0:09:00-0:09:30]: "Spotify yes, Acast no, Art19 missing… Apple is already doing that. Apple is already being cut [out]." Same observation @dave made — who's in the room and who isn't. @js replies @dave on AMP Jun 4: "@dave Dave there were no dissenting votes" — Mastodon-thread confirmation that JS + Dave are on the same page about the consensus-by-cabal red flag. Discussion: V4V counter-thesis — No Agenda is value-for-value (no impressions, no exposures). Open standards vs industry cabals. PNWR is independent-podcaster-aligned; AMP is platform-aligned. Podnews AMP Jun 4 press release Podnews AMP origin May 29 @dave Jun 4 reaction post JS Jun 4 quote post PNWR this week (Pod News Weekly Review) ----------------------------------------------------------------------------------------------------------------------------------------- 02 - THE OPEN COUNTERPART — PODCAST INDEX ISSUE #775 (PNWR + @DAVE BOTH ON IT) ----------------------------------------------------------------------------------------------------------------------------------------- 03 - THE WHY BEHIND IMPRESSIONS — "THE FIRST FOUR AND A HALF MINUTES" ----------------------------------------------------------------------------------------------------------------------------------------- 04 - THE PODCASTING 2.0 DATA COLLECTIVE — THE OPEN ANSWER TO AMP The Podcasting 2.0 Data Collective — the open, V4V-aligned answer to the AMP cabal. Not a consortium with ratified definitions and trade-press releases. A collective of open tools and honest sentinels: OP3 for analytics, Podverse + newpodcasts.net for corpus data, Podcast Index for the namespace, Issue #775 for client identification done right. Matthew 5:6 (KJV): "Blessed are they which do hunger and thirst after righteousness: for they shall be filled." The verse that frames the work. Open data, transparent measurement, value-for-value — righteousness in podcast governance. Those who hunger for it are the ones who'll be filled. The AMP cabal trades righteousness for an ad-tech seat at the table; the Data Collective just keeps the lights on. THE CHARTER — Adam's working document, June 5 2026 We hold more power than we give ourselves credit for. Definition of a Podcast: Syndicated delivery of media files with precise consumption data for all stakeholders. What we brought in (the Podcasting 2.0 namespace contributions): Transcripts Chapters Funding (V4V) Person Location …etc. Statistical relevance: Advertising is based on percentages. Collectively we have about 10% of all apps — statistically enough to be relevant. Godcaster app tracing proves we can measure important metrics. Data to aggregate and display: Follows Plays per episode Completion rate by time Strategy: Become the authoritative source by publishing open stats Monetize We will not be loved initially by the industry, because we will have the truth. Advertisers will love us though, as will Podcasters. Monetization: Data subscriptions Resellers (DJL) Ad Networks Podcasters themselves (consideration) Podcast Index has built the trust needed to house this data. We already have a data exchange relationship with the apps. op3.dev is critical in this equation to offset the old system for correlation. OP3 full podcast support landed this week [PNWR 1:53:00-1:54:30] — OP3.dev now has full episode-level + show-level analytics support for podcasts. Spec work also moving on private feeds (insecure feeds spec). Direct relevance to V4V infrastructure. @dave → @james Jun 5 11:50: "Do you have the daily lists that show up on newpodcasts.net available anywhere as a download? I'd love the full, historical list of feed urls that have appeared there if possible." Open-data request — corpus curation theme. @dave → @mitch May 30: "Would you be able to send me a flat list of all the feed urls in Podverse which have more than X number of subscribers/followers? Let's say more than 5?" Podverse data request — corpus quality. Anchor FM RSS restoration request — Fri 11:01 email to NA inbox (Lusso Lets). Listener can't retrieve feed data from Podcast Index. Adjacent infra beat — the unsung user-facing pain of corpus indexing. Discussion: corpus curation as a steady-state job (Dave's sentinel work) vs measurement standards (the AMP cabal) — which one keeps the ecosystem honest? The Data Collective doesn't ratify, it just shows up to maintain. Hunger and thirst. They shall be filled. OP3.dev — open podcast analytics ----------------------------------------------------------------------------------------------------------------------------------------- 05 - CAPTIVATE LAUNCHES DAX US — THE IMPRESSION ECONOMY IRL ----------------------------------------------------------------------------------------------------------------------------------------- 06 - BBC GOES ALL-IN ON CROSSED WIRES YEAR 3 — IPLAYER DEAL + "EDINBURGH OF PODCASTING" ----------------------------------------------------------------------------------------------------------------------------------------- 07 - STREAMING CONSOLIDATION — YOUTUBE MUSIC + TUBI + NETFLIX ALL WANT "PODCAST" ----------------------------------------------------------------------------------------------------------------------------------------- 08 - SUPPLY CHAIN SECURITY — VS CODE DELAYS, PHP FOUNDATION, SLSA LEVEL 3 IS NOT ENOUGH ----------------------------------------------------------------------------------------------------------------------------------------- 09 - AI BUBBLE PC20-FLAVOR — TOTO CHUCKS, MOTHER COMPUTERS, "NO 'I', ONLY MATH" ----------------------------------------------------------------------------------------------------------------------------------------- 10 - QUIPS / TRANSITIONS ----------------------------------------------------------------------------------------------------------------------------------------- Last Modified 06/05/2026 14:38:09 by Freedom Controller

John welcomes back Phil Hay (Destroyer, The Invitation) to ask, how do you get a movie made with independent financing? They look at how indie movies get made, where you get the money, deciding when to go indie, and whether streamers complicate the picture. We also follow up on testing movies with focus groups and answer listener questions on how to navigate the editing room, daily routines, and what to do when your story has too many themes. In our bonus segment for premium members, we look back on what Phil learned about D&D from his documentary The Dungeon Masters. Links: Phil Hay Scriptnotes Episode 244, Episode 377, and Episode 505 The Dungeon Masters The Answering Machine Meltdown from Swingers John's daily to-do template Pamela Ribon's One Act documentary Petey USA's The Yips KOXY College Radio Get your copy of the Scriptnotes book! Get a Scriptnotes T-shirt! Check out the Inneresting Newsletter Become a Scriptnotes Premium member, or gift a subscription Subscribe to Scriptnotes on YouTube Follow Scriptnotes on Instagram and TikTok John August on Bluesky and Instagram Outro by Craig Good (send us yours!) Scriptnotes is produced by Drew Marquardt and edited by Matthew Chilelli. Email us at ask@johnaugust.com You can download the episode here.