Pragma, my AI change advisor, has fielded the same sentiment four times this month — dressed in different words, identical shape every time: I am accountable for this change and I control none of the levers that would let me deliver it.
Sound familiar? Nobody puts it on a slide. (Imagine the workshop where they did..."career limiting"...)
I have lived the industrial-scale version. Around the NBN rollout I was a GM of Business Services in one of the Tier 2 telecommunications and engineering firms. It was a title that sounded far more in control than the job ever was.
The multi-technology mix had been set politically. NBN Co was an execution organisation, not a design organisation. Risk was pushed down through the Tier 1 partners to firms like ours, and we carried the outcomes of decisions made several floors and one commercial relationship above our heads.
The 2013 Strategic Review costed copper remediation at $2,685 per node. Leaked documents later put the real figure at $26,115. The cost changed by a factor of ten…but the plan didn't. Same targets, same timelines, same triumphant milestones — now resting on numbers everyone doing the work knew were fiction.
I watched good people carry the blame for outcomes that had been decided far above them, and wear it as personal failure. Not one of them was incompetent.
It took me a long time to see the real problem. It wasn't the people, and it wasn't even the blowout. It was that what we could see from the ground had no way of reaching anyone who could act on it. Our concerns and suggestions didn't get rejected — rejection would at least be an answer. They got absorbed.
Accountability sinks.
Dan Davies coined the term in The Unaccountability Machine: structures that absorb the consequences of a decision in a way that stops feedback ever reaching the people who made it.
His everyday example is the airline gate agent. They speak in "the voice of an amorphous algorithm", and you have to speak to them as a person. You've had this exact conversation, haven't you? The agent agrees with you completely — and can change absolutely nothing. Your complaint has nowhere to land, so it gets absorbed.
At enterprise scale it is the middle manager who can see the gripes coming and has three practical ideas for smoothing the changeover. But those ideas have no real forum for exploration...so the manager follows the 'top-down' instructions anyway, and hopes for the best.
And honestly...who could blame them?
The 2005 playbook did not produce accountability sinks by accident. The playbook is the accountability sink. Appoint a sponsor, hire a change manager, run an assessment, do comms, run training, go live, declare success. Read that sequence again and look for the point at which the decision itself can still change…there isn't one. The change role exists to broker compliance with a decision already made.
Which means responsibility without authority is the working output of a design. It didn't simply happen to you.
'Options to consider' quietly becomes 'future state'.
There is a moment in most programs when the most senior person stops asking questions. The slide that read "options to consider" quietly reads "future state". Nothing is announced. The leader is no longer trying to understand the problem; they are winning an argument.
I call it Destination Change. From that moment, every part of the organisation downstream starts operating in service of the destination rather than the truth.
Robodebt is the version with a Royal Commission attached. The Department of Social Services' own legal advice in 2014 was a flat no: the Social Security Act was not designed for income averaging. The advice was filed, not escalated. Ministers pointed at the department; the department pointed at the algorithm. Public servants who knew did not say, and those who spoke were not heard. Commissioner Holmes found the scheme "neither fair nor legal".
(Sitting in one right now and want to think it through before you take it to anyone? Pragma exists for precisely that conversation.)
Get out of the sink with what you've actually got
If you're in the middle layer — executing a vision you had no say in shaping, with no time to question it and no authority to fix it — it's time to face the reality and run in the lane you have.
Drop the self-blame first: Barry Staw's research on escalation of commitment shows managers keep investing in a failing course of action despite mounting evidence. What looks irrational is the entirely rational response of a person operating inside a system that punishes changing course. The structure is producing this outcome; you wearing it as personal failure is part of how the structure works.
Shrink to the minimum pathway: The destination is fixed above you, but you can manage your exposure: what is the smallest experiment, inside the work you already control, that would tell you — and eventually "them" — something true?
Work the map at your own altitude: Quiet Influencer Mapping needs no room booking and no mandate; change travels on relationships, and your relationships come with you regardless of title. One conversation at a time is the method built for people without authority. Find the one person with standing who already half-suspects what you can see, and hand them the evidence in their language.
Leave a trace: In Robodebt, the legal advice that could have stopped everything was (to great detriment) filed, not escalated. But thankfully it existed, and in the long run a Royal Commission found it. Feedback that lives only in meetings gets absorbed. Feedback in writing survives the sink. Write what you can see in a kind, factual tone to someone who could act.
And if the honest answer is that nothing you can see will ever reach anyone who can act...that's data too. Some structures you influence. Some you outlast. Some, pragmatically, you leave...but with a clean conscience and "lessons learned" for that next interview.
"Uh oh...I'm a senior leader who made one of these sinks..."
At Campbell Page I had 85 locations and no central change office. I made it in person to 56 of them, Noarlunga to Moe, Warwick to Caboolture, and called the rest. You cannot parachute in from a Sydney corporate office and mandate your way to change in Batemans Bay.
By the nature of the industry and the government contract, accountability sinks were a daily occupational hazard, especially given I had a key role in the definition of organisational performance.
Two moves I used to keep from going down the plughole:
Draw the map you are actually standing on: The official org chart and the real influence map are two different documents, and the second one is what the change rides on. A handful of people per site move everything; find them. Change travels on relationships, not reporting lines.
Interestingly enough, my senior colleagues also intuitively knew these relationships existed...so they greatly valued someone 'doing the work' and illuminating the reality.
Make the decision rights visible: Not to win them. To make them seen: who decides, on what cadence, and which war stories take precedence over recent data. Drawn, in the room, in front of everyone. I have written before about making the invisible cage visible, and the reason it works is unglamorous — an invisible cage is a cage; a visible one is a conversation.
Middle management is in the middle for a reason
None of this hands you the levers to magically create accountability or drastically reshape the dynamics of the organisation. If you are genuinely 'in the middle', you can create proxies for what you can see, reaching the people who need to see it.
So if your feedback has nowhere to land, the problem sits upstream of your performance. You are standing inside a structure built to absorb it. Structures can be named, drawn and tested. (And unlike the gate agent's algorithm, structures occasionally listen.)
What is the smallest experiment you could run in the next fortnight?
Accountability sinks: Dan Davies, The Unaccountability Machine (2024). Escalation of commitment: Barry Staw (1976). Robodebt Royal Commission, Commissioner Catherine Holmes, 2023. NBN copper remediation figures: 2013 NBN Co Strategic Review and subsequent leaked costings.
The full argument is Chapter 10 of Pragmatic Change, out now.