Fixes That Fail

There's a story economists love to tell about colonial Delhi. The British administration, worried about the number of venomous cobras in the city, offered a bounty for every dead one. It worked, at first - people brought in snakes by the sackful. Then enterprising locals started breeding cobras to claim the reward. When officials twigged and scrapped the scheme, the breeders released their now-worthless snakes, and the city ended up with more cobras than it started with. The intervention produced precisely the opposite of what it intended.

It's a wonderful story. It may also be largely untrue - when people have gone looking for the original records, the hard evidence for the cobra-breeding part turns out to be thin, and the tale seems to have hardened into received wisdom over a century of retelling. But it has lasted for a reason. It endures because it captures something we recognise instantly, something that really does happen, again and again, in organisations of every kind: a sensible fix that quietly produces the opposite of what it intended.

So let's take one we know happened.

In 2014, it emerged that hospitals across the United States' veterans' health system had been falsifying how long veterans were waiting for appointments. The system had a target - veterans should be seen within fourteen days - and bonuses were attached to hitting it. So staff at the Phoenix facility, and eventually at more than a hundred others, kept two sets of records: an official one that showed appointments happening on time, and an unofficial one where veterans waited months. The real average wait in Phoenix turned out to be around 115 days. Some veterans died waiting. The target meant to reduce waiting times had, instead, produced an elaborate machinery for hiding them.

Here's the part worth sitting with. Almost nobody in that story set out to do harm. There was a real problem - veterans were waiting too long. There was a reasonable-sounding fix - set a clear target and reward people for meeting it. And the fix, applied to understaffed hospitals where the target was close to impossible, didn't change how long veterans waited. It changed what got written down. Defenders pointed out, fairly, that fourteen days was unrealistic for the staffing those hospitals had - which is exactly the point. An impossible target doesn't summon the missing capacity. It just teaches people to game the number.

This is one of the most reliable patterns in organisational life, and it has a quietly unsettling quality: the better the fix looks in the moment, the more likely it is to be the thing you're wrestling with a year later.

Two ways a good fix turns on you

When you look closely, fixes go wrong by two different routes - and it's worth being able to tell them apart, because they feel different from the inside.

The first route is the one the cobra story and the VA story share: a perverse incentive. You attach a reward to an outcome, and people give you the outcome you measured rather than the one you meant. Pay for dead cobras and you get cobra farms. Reward short reported wait times and you get short reported wait times - which is not at all the same thing as short waits. The fix fails because people are clever, and they optimise for whatever you actually rewarded, not whatever you were hoping for.

The second route is slower and quieter, and it has no villain at all. You apply a fix; it works; and the very same fix sets off a chain of effects that takes months to surface and worsens the thing you were trying to solve. A service team drowning in complaints adds a triage step to catch problems earlier. The complaints ease - and six months on, the team is slower than ever, because the triage step became a bottleneck nobody connects back to the original fix. A leadership team frustrated by slow decisions adds a weekly sign-off meeting. Decisions speed up - and a year later, nothing moves without a meeting, and the organisation has quietly taught itself that authority lives in the room rather than in the role.

Peter Senge, in his 1990 book The Fifth Discipline, called this second pattern fixes that fail - one of a handful of recurring shapes he found showing up across organisations of every kind. But the two routes are really faces of the same coin. In both, a sensible intervention aimed at a visible problem creates a less visible one. In both, the fix looks like it worked, right up until it doesn't. And in both, the trouble was set in motion at the moment of the fix - it just took time, or human ingenuity, to arrive. The cure for both turns out to be the same, too, which we'll come to.

Why the cost arrives late and unsigned

The mechanism underneath is an asymmetry of speed and visibility, and once you see it you can't unsee it.

The benefit of a fix is fast, obvious, and easy to attribute to your action. The triage step catches the complaints this quarter. The target lifts the reported numbers this month. You can see the improvement, point to it, and connect it to the thing you did. So you trust it.

The cost is slow, diffuse, and almost impossible to attribute to anything in particular. By the time the new bottleneck or the gamed metric or the lost capability shows up, it's wearing a different costume, in a different part of the organisation, often months later. It doesn't announce itself as a consequence of the earlier decision. It announces itself as a fresh problem. Of course the fix felt like it worked. By the time the bill arrives, you've long since stopped looking at the thing that ran it up.

This is why fixes that fail aren't a sign of poor judgement. They're a sign of normal judgement working with the information it had. You judged the fix by what you could see, and what you could see was genuinely good.

There's a name for the psychology underneath, and it's surprisingly old. In 1936, the sociologist Robert Merton wrote what is still a foundational study of why deliberate action so often produces effects nobody planned for. One of his causes has a wonderful name: the imperious immediacy of interest. It describes the moment when you want the intended result so much that you simply don't look at what else your action might do. Not because you can't see it - because wanting the fix to work quietly switches off the part of you that would go looking for the catch. The urgency isn't a character flaw. It's the most natural thing in the world. And it's exactly the condition in which fixes that fail get made.

The cure is almost never where the symptom is

There's a deeper reason these fixes keep coming back, and it's the most useful part of the whole pattern to hold on to.

A symptom shows up in one place. The cause usually lives somewhere else entirely. The complaints land on the service team, but the cause might be a product decision made two floors away. The gamed waiting list shows up in the scheduling office, but the cause is that the hospital was asked to hit a number it didn't have the staff to hit. A fix aimed at the symptom treats the place where the pain is felt - which is almost never the place where the problem is made.

This is what separates a fix from a cure. A fix makes the symptom go quiet. A cure changes the thing generating it. And the catch is that the fix, by quieting the symptom, removes the very pressure that might have led you to the cause. The pain was information. The fix turns the information off while leaving the cause running.

You see this with painful clarity in healthcare, where the incentives are unusually visible. In China and a number of other systems, hospitals were funded partly by reselling drugs to patients, and doctors were given prescription targets to match. The fix solved a real problem - it helped fund hospitals. It also, predictably, produced large-scale overprescription, because the system had quietly made "prescribe more" the rational thing to do. The symptom (underfunded hospitals) was real. The fix (let them sell drugs) worked. And the cause - how you fund care without tying a doctor's income to a patient's pill count - was left untouched, generating a new and bigger problem one layer down.

Senge's point was never that quick fixes are bad. Sometimes the building really is on fire and you put it out before doing anything else. His point was subtler: the quick fix and the real cure are different things, and trouble starts when we mistake one for the other. We reach for the fix, feel the relief, and never come back for the cure - because the fix worked, and working is exactly what makes it dangerous.

Why the cure feels like it isn't working

Here's the part that makes the cure genuinely hard to choose, rather than just easy to skip.

In the early 2000s, two MIT researchers, Nelson Repenning and John Sterman, studied why so many organisational improvement efforts quietly fail. What they found has an uncomfortable shape to it. When an organisation finally stops firefighting and invests in fixing the underlying problem, the first thing that happens is that performance gets worse. You're spending time and attention on the deeper work, which means less of both for the daily fires. Things dip before they lift.

They called the pattern a capability trap, and the trap is this. Managers are used to effort paying off quickly - that's what firefighting trains you to expect. So when the real cure produces an early dip instead of an early win, it reads as failure. The improvement gets abandoned right at the bottom of the curve, just before it would have started to pay off. The organisation lurches back to firefighting, reaches for the next quick fix, and concludes that "we tried that and it didn't work" - about the one thing that might actually have worked, given a little more patience.

This is the quiet tragedy underneath fixes that fail. It's not only that the quick fix is tempting. It's that the real cure asks you to accept a period of looking worse in exchange for being better later - and most organisations, under pressure, can't hold their nerve through the dip.

What makes an organisation prone to it

Not every organisation falls into this at the same rate, and the conditions that feed it are worth knowing - because they're conditions you can change.

The first is pressure. When everything is urgent, the fast visible loop wins every time. There's no slack to ask "and then what?" - there's only the next fire. Organisations under sustained strain, whether from a merger, a regulator, or rapid growth, tend to accumulate fixes faster than they can absorb their consequences. Each one made sense. Together they form a kind of sediment that nobody laid down on purpose.

The second is weak feedback across boundaries. The fix happens here; the consequence lands there. If "here" and "there" don't talk - if the service team that inherited the bottleneck never connects with the team that added the triage step - then the loop stays invisible, and the same kind of fix gets reached for again and again. This is one of the quieter costs of organisational silos: not just that teams don't coordinate, but that the side effects of one team's reasonable decisions disappear into another team's in-tray, untraceable.

The third is a culture that rewards the fix and never asks about the cure. Repenning and Sterman quote an engineer who put it perfectly: nobody ever gets credit for fixing problems that never happened. The dramatic save is visible, attributable, and celebrated. The quiet prevention that meant there was no crisis to begin with is invisible by definition - so it goes unrewarded. W. Edwards Deming made the same observation more sharply: do the job right the first time and you're invisible; let it go wrong and rescue it, and you're a hero. An organisation that runs on that logic slowly fills its senior ranks with brilliant firefighters and quietly teaches everyone else that the way to be valued is to be seen putting out a fire - which is a powerful incentive to let one start. The pattern stops being an event and becomes a habit. The organisation gets steadily better at solving the wrong layer of the problem.

Seeing one move further

Here's the shift that changes things, and it's smaller than you might expect.

The fix isn't a thinking failure. It's first-order thinking - the answer to the question "what happens if I do this?" That question is necessary. It's just not the whole question. The trouble is that first-order thinking feels complete. You ask what happens, you get a good answer, and the relief of a working solution closes the loop before the second question has a chance to form.

Second-order thinking is just the next question: and then what? What does this set in motion that I'm not currently looking at? Where does the pressure go when I push it down here? And, for anything with a reward attached: what behaviour does this actually pay people to do - not what I want it to pay for, but what it really rewards? None of these questions is hard. The difficulty isn't intellectual - it's that pressure and relief conspire to stop you asking them. The fix works, so you move on, and the second-order effects build quietly in the space where your attention used to be.

This is the cure for both kinds of failure, which is why it's worth dwelling on. The slow hidden loop and the perverse incentive both get caught by the same habit: looking one move past the immediate result, before you act, and going back to check the layer you couldn't see at the time. You can watch this thinking at work in the people who seem to make unusually durable decisions. They're not smarter about the first move. They're simply slower to call it finished. It's not caution. It's a habit of looking one move further than the problem demands.

That habit is learnable, and it doesn't require you to predict the future or model the whole system in your head. It requires one thing: a pause, before you act, long enough to ask what the fix might set in motion - and a note in the calendar to come back, after it's worked, and look for what it quietly started.

The question underneath

There's something freeing in this pattern once you see it clearly. If the problems that keep returning aren't caused by bad decisions, you can stop hunting for the bad decision-maker. The recurring problem isn't evidence that someone keeps getting it wrong. It's evidence that the organisation keeps solving the symptom and leaving the cause to run.

That reframe matters, because the two readings lead to completely different work. If the problem is people, the response is to find better ones, or hold the current ones more tightly to account - which, if the real issue is the pattern, just adds another fix on top of the pile. If the problem is the pattern, the response is to build the conditions that let you see one move further: the feedback across boundaries, the slack to ask "and then what?", the nerve to hold through the dip when the real cure is working but doesn't look like it yet.

So the next time a problem returns wearing slightly different clothes, it's worth resisting the obvious question - who let this slip again? - in favour of a quieter and more useful one: what did we fix last time, and what did that fix set in motion?

The answer is often sitting in plain sight, one move further on from where everyone stopped looking.