Discover
Biweeklycast
76Β Episodes
Reverse
πΊ Watch this episode on YouTube.Dima explains why "we'll build a reliable system" makes him twitch, and he and Slava work out what it takes to turn such a promise from wishcasting into something you can believe.Simple characteristics correspond to intrinsic properties one can build directly: a square house, a red car, a four-minute songOther characteristics β reliable, comfortable, fun, effective β one can't "build" directly. They can only be derived from other, simpler properties of the thingDima believes that promising a derived characteristic without explaining how it will arise is wishcasting. One cannot mandate "from Monday the team will be effective" β they have to take specific step so that as a result the team becomes effectiveDima's approach to talking about promises for derived characteristicsSlava argues that the "what" should come before the "how": teams jump straight to solutions without a shared map of what reliability means and miss pieces they haven't run into previouslyReferences:Why the green great dragon can't exist β order of adjectives in EnglishBiweekly 52: Needs Analysis β the chain of results from actions through outputs to outcomes you can't produce directly"I'm sorry, could you explain what is a great question? Because I think it can be quite subjective." β Slava
πΊ Watch this episode on YouTube.Dima poses that every product feature falls on one side of a "hygiene bar," and he and Slava explore where that line sits, how it shifts over time, and whether the idea can become more than his own intuition.Ranking every feature on one universal scale assumes they're commensurable, which is rarely trueJim McCarthy's five feature categories(from Dynamics of Software Development): strategic, competitive, customer satisfaction, innovation, paradigmaticOther models pointing at similar asymmetries. For example, theΒ Kano model and Terry Hill's order qualifiers and order winnersDima crystalises this into the idea of a hygiene bar. Below the bar, a missing feature penalizes the product but having it earns little. Above the bar, features differentiate, but only once the product already clears the barHygiene bar is an expression of the market. Shifting hygiene bar allows to trace how market evolves. E.g. Command strips moved the market of "means of hanging things on walls"McCarthy's categories and the hygiene bar: strategic and customer-satisfaction features sit below the bar, competitive features sit above it, and investment features stay neutral until a later release. Paradigmatic features are the odd one out: they move the existing bar, pushing new requirements onto competitors."Apples are apples β you can't make a paradigm shift with an apple as a product." β Dima
πΊ Watch this episode on YouTube.Dima introduces jingle and jungle fallacies β two ways labels can go wrong β and takes Slava through examples to show harder cases that are easy to miss.Names and labels help to compress knowledge and facilitate communication, but the label is always distinct from the thing it denotesJingle fallacy β using the same label for two different things and treating them as oneJungle fallacy β using different labels for the same thing and treating them as differentIn Ukrainian, different shades of blue have distinct words β the same object would carry genuinely different meanings depending on the speaker's languageThe failure-to-differentiate and failure-to-converge cases are harder to spot and fix than miscommunication. Because of that consequences can be more damagingJingle and jungle can also be weaponised β deliberately conflating or splitting concepts in rhetoric to mislead"I would say that I probably had a jangle fallacy about jingle and jangle fallacies." β Slava
πΊ Watch this episode on YouTube.Dima poses that "trust" plays an important role in how we execute multi-step processes, and together with Slava they explore the implications.Process is a sequence of steps where each output becomes the next step's input β and before it can, trust in that output must be establishedThree ways to establish that trustSlava argues trusting yourself as executor is a stretch β you know when you cut cornersFormal vs informal verificationTrust evolves β first you check every output, then you trust the person, eventually you trust them to define the procedure themselvesFor repeated processes, optimise the trust mechanism rather than verifying every output each timeSlava suggests what matters is defining success criteria β definition of done; how it's done is secondary"This is a good coffee because I brewed it." β Dima
πΊ Watch this episode on YouTube.Dima and Slava find that scaling productivity with AI likely means going from a faster tool to running a team of synthetic labour β and that the challenges of delegation are nothing new.Productivity with a tool vs. productivity with an agent: a tool allows you to do things better/faster and with an agent you set it an objective that feeds into your bigger goalAgentic threshold: the executor must pass judgment rather than just carry out instructionsSynthetic labor β AI agents you delegate work toVerification Tax: it is more effective to delegate tasks that take long to do but are easy to verify β verifying by re-doing the task yourself defeats the purposeICs delegate to themselves, managers delegate to othersDelegating to others is a distinct muscle that needs developmentSlava is worried that most managers don't operate at the higher levels of delegation even with real peopleBuilding the pyramid down β AI lets an IC build a hierarchy of agents below them, compressing what used to require climbing the corporate ladder"Do this for me is a tool. Do it with me is borderline. Think for me would be closer to agency." β Slava




