Are all backlogs ultimately flawed?

Who, in the 2020s, in digital technology hasn't tried SCRUM yet? Who hasn't seen a backlog that grows weekly regardless of how quickly things get done? Nobody? I thought so. We all have seen it. Most of us do it. There is no end in sight.?

While there are some fantastic success stories of transformative delivery, very often, what is on the other side of an agile transformation is a daily grind in mini waterfalls called sprints. Could the backlogs be the problem??

John Little proved his law 61 years ago. Since then, we have known that systems with an arrival rate higher than the throughput are unstable. What does it mean? If you have an ever-growing backlog, you will fail whatever you do. As long as you continue adding more work than what you can do, you will continuously increase the proportion of effort needed to manage that backlog and so reduce the throughput adding to the problem.?

The natural temptation is then to focus on localised optimisation. What if we could increase efficiency? What if we could make that machine work closer to 100% of the time? Could that person do something else now rather than wait for their turn in the process? Wouldn't that be better? Wrong. Little's Law explains how systems without enough spare capacity grind to a halt. Systems at 100% efficiency take, literally, forever to do anything.

To do more, we have to do less. Focus on the flow of value and not how busy the system appears. Having a backlog is bad and expensive. Having an ever-growing backlog is pure madness.?

I was just reminded about it when watching The Value of Flow - Dan North's talk from 2017 at Agile London. But there are many more good resources to find out more about it. There is Allen Holub's #NoEstimates talk on YouTube. The Peopleware book by Tom DeMarco and Timothy Lister (first edition 1987 and still very relevant!). There is also the Team Topologies book by Matthew Skelton I've been recommending to everybody for the last few years. All worth the time.?

Ideally, there would be no backlogs and simply a flow of value with just-in-time demand. But it's complicated. I know; I have an ever-growing backlog, too... or two. And so, in a twist to this story, I'd like to ask for ideas rather than offer solutions. How can we change this? How can we see the back of backlogs? What could I do in a complex organisation where I have about 40 things I'm personally waiting for at any given time? How to change a complex system organically without falling into the localised optimisation traps? Suggestions on postcards (or in comments), please.