Sooner - And sooner still

 I published my first post on Sooner, not Faster in 2006 on Object Mentor's blog ("But Uncle Bob!" was the title), a wiki I was encouraged by Bob to use daily. Later this theme had been picked up by many people. Ryan expanded it to "Sooner, Safer, Smarter" and I appreciate that also.

 I've had years, decades in fact, to be working through what it means to deliver sooner and more often. 

 I've been introduced to many concepts that are related, and I've created a short guide to flow to help people quickly come to terms with the systems view (ToC, Reinertsen's Flow, Continuous Delivery) as a mechanical system of queues and inventories. 

 That is a very important lens that brings us to appreciate that we can do better without working harder, longer, faster, or even agentic aids. It even explains why working harder, faster, longer, and with code agents hasn't really panned out so far.

 What really does matter is story splitting, writing good code, and working together.

Yes, those are the things that all the agile methods actually recommended before some yahoos turned the agile community into a ticketing system vendor's cash cow. 

When we stopped looking at producing deliveries early and often, and instead focused on velocity, ticket closure, and properly conducting scrum events is when we lost our way and why everyone hates the things they consider Agile and Scrum.

Because those things don't work UNLESS you split stories, write excellent code, and work together. 

When you look at everything I've ever said about code craft, TDD, collaborative coding, scatter-gather, VSMs, first-time-through,  and work back-flow you will see that it all ties together. 

Great work in small end-to-end slices, fully completed, flow through the system without back-flow and without queueing and waiting. Partial stories in rapidly closed tickets and separate version control branches do not. 

I've been endeavoring this whole time, at Object Mentor, in my personal work, at Industrial Logic (where I learned so much from Joshua Kerievsky), and in todays work and blogs it's the same theme over and over...

The queues and hand-offs system is broken, and AI doesn't fix that. 

The ticket focus has failed everyone, even though teams manage to eek out software releases against the odds.

What still works is 

• small releases of e2e slices to avoid batches
• great knowledge representation in code (human or AI typing)
• working together to avoid queues and hand-offs
• managing WIP and cognitive load 
• knowing your system and watching for queues and bottlenecks
• automating away the dull tasks or the ones that OUGHT to be dull
• simplifying your system
• continuous learning

Without these, you can't make the next month any better than the last work. 

With these, you can't help but be on an upward trend.

 Sooner? Yes.

Safer? Yes

Smarter? Much

Faster? Well, sometimes it is, but that's not what really matters.