Stephen Janaway

The developer productivity paradox: Why faster coding doesn’t mean faster software delivery Developers are using Generative AI to crank out code faster than ever before, but somehow, the metrics aren’t showing an overall productivity improvement. So what’s going on?

A theme that really stood out at @gradle.com's DPE Summit conference is that while developers feel productive using GenAI for coding, this isn't necessarily translating into more features/fixes/customer value for organisations. So I wrote a blog post about it

gradle.com/blog/develop...

Thanks for having me Lean Agile Scotland @leanagilescot.bsky.social #LAScot. It was great to get the opportunity to tell folk about how our approach to production support has changed over the years at Bloom & Wild 💪💐

The weather is always this good in Edinburgh, right? 😉

A Tester's Growth-Path to Test Automation

Published: A Tester's Growth-Path to Test Automation

At one client, the topic of how a growth path for Testers in Test Automation would look in their QA Community of Practice. That triggered me to ask that question in the wild. Here's the result.

thinkinglabs.io/articles/202...

I'm very much looking forward to heading back to Edinburgh next week to Lean Agile Scotland. Come along and hear about how we've scaled our approach to production support and maybe also learn a little about flowers 😉🌹

Posts from Awesome Folks #159 A weekly round up of blog posts I've been reading

This week's Posts from Awesome folks is here hellotacit.beehiiv.com/p/posts-from...

And so the mammoth train journey to the southwest tip of England begins again. Very excited to be heading down to Agile On the Beach ⛱️

Co-signed in blood

Time to do the sponsor thing at Brighton Ruby. Which is always fun. Come and meet us at Bloom&Wild in the bar area 🌻🌻

LDX3. I'm sure this'll be fun

This lines up with Atlassian's remote work research in www.atlassian.com/blog/distrib...

> Our research shows that these intentional team gatherings lead to a 27% increase in feelings of connection, especially among new graduates and new hires, and that this boost lasts for 4-5 months.

You are not your job. And soon, you won't have one. The identity collapse hiding inside the job market crisis

carmenvankerckhove.substack.com/p/you-are-no...

A somewhat sobering read

An interesting (and balanced) thread on remote working tradeoffs

Building Successful Communities of Practice with Emily Webber Learn how to form, grow, supercharge and measure communities of practice in or across organisations. An online self-paced course at tacit.pub/learnCoPS

This labour of love is now live!
Learn to start, grow and supercharge communities of practice with:

★ Expert-led videos, discussions, tasks and downloads
★ Practical, hands-on activities to support your learning
★ Tutor support
★ A certificate of completion

Start learning at tacit.pub/learnCoPs

Estimating AI energy usage is fiendishly hard - but this report took a shot And it gets even harder when you try to estimate CO2 emissions A single person with a serious AI habit may chew through enough electricity each day to keep a microwave running for more than three hours. And the actual toll may even be worse, as so many companies keep details about their AI models secret.…

Estimating AI energy usage is fiendishly hard - but this report took a shot

🤔

If you want to encourage people to use AI, measure their productivity on LOC.
As a measure, it's proven to be destructive (more code is more maintenance work, not more value).
But hey, if your goal is AI use at any cost, that's a problem for the future.

20 Years of Domain-Driven Design: What I’ve Learned About DDD Eric Evans’ Domain-Driven Design (DDD) has shaped software development practices for over 20 years, providing a rich toolkit for tackling complex domain…

Useful summary of Domain-driven design (DDD) key points and learning from the past 20 years - from Eberhard Wolff

#DDD #TeamTopologies

speakerdeck.com/ewolff/20-ye...

Things I actually believe about AI and the future of tech employment:
1. There will still be junior, senior, and mid-career engineers. Will there be as many? Dunno. But all levels will continue to exist.
2. We will use (AI-assisted) formal methods long before engs get widely good at writing specs

Career advice in 2025. Yesterday, the tj-actions repository, a popular tool used with Github Actions was compromised (for more background read one of these two articles). Watching the infrastructure and security engineering...

Not having as much fun in 2025. This may be why lethain.com/career-advic...

The 4Ds Roles and Responsibilities Workshop People often struggle to describe who does what. Getting it wrong can mean that things fall through the cracks, or it becomes so descriptive that it hinders work. Articulating roles and responsibil...

New blog post: The 4Ds Roles and Responsibilities Workshop

emilywebber.co.uk/the-4ds-role...

How Much Should I Be Spending On Observability? In this update to her 2018 post, Charity Majors explains how much teams should spend when it comes to observability costs.

Observability engineering teams can be powerful force multipliers, while writing and owning relatively little code themselves.

Turning observability from cost --> investment has an external component (UX, revenue) and an internal component (DX, productivity).

www.honeycomb.io/blog/how-muc...

🤦

The Leadership Team Purpose Workshop

New blog post The Leadership Team Purpose Workshop

emilywebber.co.uk/the-leadersh...

Thread worth reading alert

www.seangoedecke.com/good-times-a...

Can't argue with a lot of this 😓

- Level Collapse from @matthewphilip.bsky.social
- Myths and legends about Big Tech from Kai Hansen
- AI and developer obsolescence from @sebrose.bsky.social
- You've built it, now you support it from @stephenjanaway.bsky.social

Distrust Breeds More Distrust When You Don't Trust Your People, That's THE Problem You Must Fix

mdalmijn.com/p/distrust-b...

A good read on trust (and accountability sinks which is a new concept for me)

Non-Blocking Continuous Code Reviews - a Case Study

The slide deck for “Non-Blocking Continuous Code Reviews” at FlowCon France @flowcon.bsky.social #FlowCon

How a novice team found ways to improve delivery performance.

Key takeaway:
Features arrive in production even before a review started.

thinkinglabs.io/talks/2024/0...

Thanks Flowcon, that was a lot of fun. Hopefully you learn a bit about our production support journey at Bloom & Wild and also a little bit about flowers too 🌷🪻🌹

I'm here at Flowcon France, speaking tomorrow at how we have scaled production support at Bloom & Wild. In the meantime, I really like the transparency here from the organisers on where the ticket money goes 😁