Staring at the terminal, the cursor blinks at a steady, rhythmic rate that feels like a taunt. It is 3:09 AM. Sarah is currently the center of the universe, or at least the center of a very frantic Slack channel where 29 engineers are watching her every move through the digital veil. A database migration has gone sideways-not just a little bit, but the kind of catastrophic failure that suggests the laws of physics might be optional. Sarah is typing. She is digging through logs. She is, quite literally, saving the company from a loss of approximately $899 per minute. And when she finally fixes it, when the green lights return at 4:29 AM, she will be greeted with a shower of praise-hands emojis and ‘hero’ tags. She might even get a $99 gift card or a special shout-out in the all-hands meeting.

The Cognitive Bias of the Kitchen Sink

I found myself thinking about this today while I was cleaning out my refrigerator. I threw away a bottle of honey mustard that expired in 2019. There was a jar of capers that had turned into a biohazard, and a half-empty container of yogurt that was essentially a science experiment. Why did I wait until the smell became unbearable to act? It’s the same cognitive bias. We ignore the slow rot because it’s not an emergency yet. We wait for the crisis to feel like we’re doing something meaningful. I realized that my fridge was a metaphor for our tech stack: full of legacy debt and expired ideas that we only deal with when the stench finally forces our hand.

[We reward the splash, not the bridge.]

The Invisible Work of System Architecture

Perceived Value vs. Actual Effort (Hypothetical Data)

Daniel P., a crowd behavior researcher I’ve followed for years, once explained to me that human groups have a visceral reaction to visible effort. He conducted a study involving 89 different scenarios where emergency interventions were required. He found that bystanders and observers consistently rated the ‘hero’-the person who ran into the fray at the last second-as more valuable than the person who had quietly spent 59 minutes ensuring the crowd didn’t bottleneck in the first place. In Daniel P.’s world, the ‘preventer’ is invisible. If you do your job perfectly as a system architect, nothing happens. No sirens, no Slack celebrations, no adrenaline-fueled midnight war rooms. And in a corporate environment that measures value through visible output, ‘nothing happening’ looks a lot like ‘doing nothing.’

I remember a specific mistake I made 9 years ago. I was leading a small team of 9 developers, and we had this one guy, let’s call him Mike. Mike was a ‘wizard.’ He was always the one staying late, always the one on the 2:59 AM calls. I praised him constantly. I gave him a bonus. I didn’t realize until much later that Mike was actually the source of most of our instability. He wrote code that was so complex and brittle that he was the only one who could fix it. He wasn’t a hero; he was a bottleneck with a cape. By rewarding his ‘heroism,’ I was actually rewarding the technical debt he was creating. It took me 19 months to untangle that mess after he left for a higher-paying gig at a firm that probably also thinks he’s a wizard.

The Lie of Inevitability

This systemic addiction to adrenaline is what we try to dissect regularly in

Ship It Weekly, looking at the operational guts of how teams actually survive their own success without burning out their best people. Because the reality is that Sarah is tired. She didn’t want to be awake at 3:09 AM. She didn’t want the praise-hands emojis. She wanted her refactor ticket to be approved 9 months ago so she could sleep through the night. When we frame these moments as heroism, we are gaslighting our employees into believing that systemic failure is actually a personal opportunity for glory. It’s a way for management to dodge the responsibility of poor planning by shifting the burden onto the individual’s resilience.

Management often argues that they can’t afford the ‘luxury’ of preventative maintenance. They say the market is moving too fast. But if you look at the numbers, that’s a lie. The cost of Sarah’s 49 minutes of downtime, plus the subsequent ‘recovery’ time where she and the other 29 engineers are too exhausted to be productive, far outweighs the cost of the original refactor. We are just bad at accounting for invisible costs. We see the $899 per minute loss during the outage, but we don’t see the $19,999 in lost productivity and attrition risk that comes from a team that is constantly in ‘fight or flight’ mode.

[Silence is the sound of a system working.]

The New Reward Structure

To fix this, we have to change what we celebrate. We need to start giving shout-outs to the engineer who successfully migrated a legacy system with zero downtime. We need to celebrate the PR that removes 999 lines of unnecessary code. We need to make the ‘fire inspector’ the highest-paid person on the team. If an engineer comes to you and says, ‘I spent the last 39 hours making sure this thing we built 9 years ago doesn’t explode,’ that person deserves the gift card more than the person who put out the fire.

It’s uncomfortable because it requires a level of technical trust that many managers lack. It’s easy to see a fire; it’s hard to see the absence of one. It requires managers to actually understand the work enough to know when someone is preventing a disaster. It requires us to look at our backlog not as a list of ‘stuff we’ll do if we have time,’ but as a roadmap for our future stability. If a ticket has been sitting there for 199 days, it’s not just ‘low priority’-it’s a ticking time bomb.

The Path to Resilient Systems

I still feel a bit guilty about those expired condiments I tossed. They were a reminder of all the times I chose the easy path of ignoring a small problem until it became a gross, smelly mess. In my kitchen, the stakes are low. In a production environment, those expired ‘condiments’-the outdated libraries, the unpatched servers, the brittle deployment scripts-are what eventually lead to the 3 AM calls. We have to stop acting like these calls are inevitable acts of god. They are choices. Every time we prioritize a flashy new feature over a boring security patch, we are choosing to have a crisis later.

Defining System Health

Daniel P. noted in his research that the most resilient systems aren’t the ones with the best emergency responders, but the ones with the lowest ‘surprise threshold.’ They are built by people who are obsessed with the mundane. They are built by engineers who hate drama. If your team has a ‘hero,’ you don’t have a star employee; you have a process gap. You have a single point of failure that is currently being masked by someone’s willingness to sacrifice their sleep. That is not a sustainable business model. It is a slow-motion car wreck.

Next time you see a ‘Sarah’ saving the day at 3:39 AM, don’t just send an emoji. Ask her for the link to the ticket she wrote 9 months ago that would have prevented this. Then, go find the person who deprioritized it and ask them why they thought a 49-minute outage was a better use of company resources than a week of preventative work. We have to stop worshiping the fire and start respecting the water. Heroism is a symptom of a broken system, and it’s time we started treating it like the red flag it actually is. Are we building a cathedral of stability, or are we just really good at digging ourselves out of the rubble? The answer is usually found in the tickets we choose to ignore.