The Bike Shed

Aji and Joël explore technical debt through the sharper lens of risk, weighing likelihood, impact, cost, and the consequences of waiting. They discuss onboarding through codebase investigations, the perils of refactoring for aesthetics, performance planning for unpredictable traffic, and why risk may communicate technical tradeoffs better across teams than financial metaphors.

Joël takes Bike Shed out onto the conference floor at RailsWorld as he interviews programer and streamer Rachael Wright-Munn, aka ChaelCodes. Rachael talks about her efforts to index all ruby events around the world with RubyEvents.org, how she augments her workflows with years old generators rather than LLMs, the rising cost of tokens and why AI may not always be your best solution to a problem. — Keep up to date with all things Ruby conferences, past and future, with Rachael’s index at RubyEvents.org Read more about the SPIDR approach Rachael mentioned. Our guest for this episode was Rachael Wright-Munn, connect with her on LinkedIn or check out her website to follow along with her work. Your host for this episode has been thoughtbot’s own Joël Quenneville. Don't forget you can now also watch the show over on our YouTube channel! If you would like to support the show, head over to our GitHub page, or check out our website. Got a question or comment about the show? Write to our hosts: [email protected] This has been a thoughtbot podcast. Stay up to date by following us on social media - YouTube - LinkedIn - Mastodon - BlueSky © 2026 thoughtbot, inc.Support The Bike Shed

A deep dive into why willpower and team conventions are unreliable safeguards in software development. The discussion explores temporary API adapters, architectural boundaries, static analysis versus tests, types and invariants in Ruby, complexity metrics, lint overrides, security scanners, and dependency automation. It also examines turning recurring code review feedback into rules and using deterministic analysis to improve code generated by LLMs.

A buggy React form sparks a discussion about incrementally taming complexity and structuring multi-step workflows. A heraldry side project reveals ChatGPT’s anchoring bias. The conversation then turns to GitHub’s stacked pull requests, storytelling commits, reviewable dependencies, partial delivery, and why vertical slices and thoughtful decomposition matter even more when working with LLMs.

Playful imagining of which Muppets fit modern service team roles. Discussion of gov.uk-style role definitions and how those map to familiar characters. Rapid-fire casting debates for designers, researchers, delivery leads, developers, QA, and stakeholders. Light travel and conference anecdotes bookend the playful roleplay.

A lively look at what 'principal' or 'staff' engineers actually do and how career paths split between IC and management. They compare roles to academia and architecture, map common archetypes, and talk about balancing visible work with protected focus time. The discussion also covers when principals should shape AI strategy and how people drift into broader-leverage work.

Jimmy Thigpen, a developer who applies AI in workflows, and Tess Griffin, a software developer shaping company AI policy, join to discuss thoughtbot’s AI Ethics Task Force. They talk about why the task force formed and its goals. They cover client expectations vs. practical AI workflows. They explore preserving development best practices and where AI reasonably fits in product work.

Discussion about graphs and trees and why their structure matters in software. Tales of a contracting system that required true graph handling and the pitfalls of serializing graphs to databases. Explanations of adjacency lists, closure tables, Postgres LTree, and when a graph database is appropriate. Thoughts on preferring trees when constraints let you simplify problems.

Aji and Sally are back together again, this time to discuss the different apps they use to make their workflows and To Do lists easier and quicker to achieve. Sally dives into the Notion calendar system which she uses to coordinate her many Google calendars, Aji looks back on using Jira to co-ordinate their international move, before they both reminisce about the benefits of using Alfred as people with ADHD. — There’s still time to secure your place at thoughtbot’s upcoming UK meet ups over the next month. London Tech Leader Meetup - Tuesday June 23rd Brighton Tech Leader Meetup - Wednesday June 24th Brighton Ruby - Thursday June 25th Evolve - Friday June 26th Your hosts for this episode have been thoughtbot’s own Sally Hall and Aji Slater. If you would like to support the show, head over to our GitHub page, or check out our website. Got a question or comment about the show? Write to our hosts: [email protected] This has been a thoughtbot podcast.  Stay up to date by following us on social media - YouTube - LinkedIn - Mastodon - BlueSky © 2026 thoughtbot, inc.Support The Bike Shed

Discussions about what makes a strong technical blog post and how to keep writing focused and scannable. Clear advice on defining your audience and scoping content to avoid tangents. Practical talk about consistent terminology, framing motivation, and choosing the right post archetype. Views on confidence, reducing hedging, and using automated tools and AI as editing aids.