In Episode 29, Ian and Ash venture into the wild frontiers of video podcasting, with Ian's "door and chair" background sure to be nominated for design awards. The pair tackle the thorny art of story splitting, with Ash confessing his frustration at watching teams scramble around massive requirements like ants on a dropped ice cream cone. Meanwhile, Ian suggests testers might be uniquely positioned to lead such efforts, having "a clear foot in each domain" – or as Ash quips, possibly Schrödinger-like properties.

The mood darkens as Ian reveals Apple's "deeply disappointed" (not "gravely disappointed" – an important semantic distinction) announcement about disabling UK encryption features following secretive government demands. Our hosts explore the concerning implications, suggesting this puts the UK's surveillance powers closer to China than democratic peers, while wondering if police might soon arrive because "computer said nick."

Between discussions of HP's spectacular customer service own-goal (forcing callers to wait 15 minutes even when operators were available), Ian's surprisingly positive experiences with Claude Code AI, and the shocking revelation that they're podcast royalty in Côte d'Ivoire (top 10!) and Cameroon (top 100!), our hosts deliver an episode that proves that story splitting may be challenging, but splitting hairs about levels of disappointment is an art form unto itself.

Links

• The Register: HP deliberately adds 15 minutes waiting time for telephone support calls, and then ditches it.
• Live Multicam in Final Cut Pro for iPad
• What A Lot Of Things on Bluesky, and our podcast hosting service Transistor.
• The Humanizing Work Guide to Splitting User Stories
• Mike Cohn: Five Story-Splitting Mistakes and How to Stop Making Them
• Conway's Law (wikipedia) which explains why heavily delineated technical teams (front-end, back-end, database) end up splitting work along those same boundaries rather than by user-pathway centred flows.
• Youtube: Introducing Claude Code, and Claude Code: Overview
• Apple can no longer offer Advanced Data Protection in the United Kingdom to new users
• Matthew Green: Three questions about Apple, encryption, and the U.K.
• UKGov: Investigatory Powers Act 2016, and its Wikipedia entry.
• Edward Snowden (wikipedia) and Uber's God View (The Guardian)
• ...and, of course, Cynefin.

In Episode 28, Ian and Ash wade into the treacherous waters of AI-generated testing strategies, with Ian demonstrating how LLMs can now create comprehensive (but perhaps suspiciously mundane) test documentation with just a few commands. The pair debate whether testers should fear for their jobs or simply laugh at AI's confident yet risk-blind approach to testing. Meanwhile, Ash ponders the BBC's hand-wringing over the future of open source software, questioning whether the current gatekeepers might need to stop finger-wagging at younger developers and instead create more welcoming environments for volunteers.

Between discussions of 960Mbps internet connections, mind flayers in Baldur's Gate, and the correct pronunciation of "Ethernet," our intrepid hosts manage to deliver a rollicking episode that proves testing isn't dead - it's just a zombie looking for brains.

Links

• Simon Willison's blog, and his post about how "o3-mini is really good at writing internal documentation".
• Examples of o3-mini's work: Test Strategy for Ilkley Live and Ilkley Live New Developer Onboarding Guide
• Simon's `llm` and `files-to-prompt` command line tools
• We'll Give It A Go - The Spooky Men's Chorale (Youtube)
• The Cursor IDE as used by Ian.
• Ben Franklin's Famous 'Liberty, Safety' Quote Lost Its Context In 21st Century (NPR)
• Content Management Systems - Contentful, Sanity and Strapi.
• Douglas Adams quotation on age and technology
• Openreach Full Fibre Broadband
• Things Ian never wants to end: his game of Baldurs Gate 3 and Jonathan Strange and Mr Norrell
• BBC News: Will young developers take on key open source software?
• Will Young...
• Unix tools - `curl` and `wget`.
• The Register: Mixing Rust and C in Linux likened to cancer by kernel maintainer
• The naming of `git`.

In this barnstorming episode of What A Lot Of Things, Ian and Ash bravely venture into the uncanny valley of AI pair programming, where the machines are suspiciously eager to agree that you're an absolute genius. Will our intrepid hosts manage to navigate the delicate dance between genuine collaboration and what Ash describes as "an advanced rubber duck with impeccable manners"? (Spoiler: sort of!)

But wait, there's more! Just when you thought the AI world couldn't get more dramatic, enter DeepSeek R1, the plucky Chinese upstart that's got Silicon Valley clutching their very expensive pearls. Our hosts dive into this tale of hobbled chips and unexpected innovation, while simultaneously managing to reference municipal gas works, start taking over the monuments in Monument Valley, and establish the critical importance of saying "What A Lot Of Things" in hardware stores across the nation.

Plus, hear all about the wildly successful What A Lot Of Things Christmas party, where actual listeners crossed actual Pennines to join our heroes for what we can only describe as an evening of unparalleled podcast-based revelry.

Links

• Thoughtworks Tech Radar on Replacing Pair Programming with AI
• Thoughtworks Memo: Coding assistants do not replace pair programming
• Useful coding helpers in the form of Claude, OpenAI o1, and v0.dev.
• Also, OpenAI's o3 announcement (dated before recording) and o3-mini release (dated after)
• Baldur's Gate 3
• Youtube: Brian Eno – January 07003: Bell Studies for the Clock of the Long Now (2003, Full Album)
• The shadcn/ui component library
• Steam Deck and Nintendo Switch 2, with all-new Mario Kart
• OpenAI o1 System Card and Apollo Research: Frontier Models are Capable of
  In-context Scheming.
• Github: DeepSeek R1
• Simon Willison: DeepSeek-R1 and exploring DeepSeek-R1-Distill-Llama-8B
• nVidia Project DIGITS, allowing you to run models locally of only 200b parameters.