Complexity is the shelter a punishment culture builds when you ask it to be failure-friendly.
Yesterday I sat with a friend and we traded problems, the way friends do at our age. Somewhere in the second hour I heard myself say a sentence. Simple minds try to appear complex. Complex minds do their best to appear simple.
It sounded good. Good enough that I wrote it down afterwards. Sentences that sound good are usually the ones that need checking. The next morning it looked different. The sentence flatters whoever says it. Everyone who uses it puts himself on the right side, and I had just done exactly that. So I stopped liking the sentence and started liking the question under it. Why is complexity so comfortable?
Because it protects. That is the whole answer. The rest of this text is the bill.
In January 2019 I wrote a text about failure friendly culture. Keep the height of falling low. Celebrate failing instead of hiding it. Pass on the cause and the fix as fast as you can. I still sign every line. What I did not see coming is what happens when the words win and the culture does not.
Seven years later every organisation I know has the words. Blameless post-mortems. Learnings. Retros. Fail fast, on a poster next to the coffee machine. Under the words, most of them still have punishment. Not on the poster. In the corridor, in the next budget round, in who gets the good project after the outage. And under punishment people are not stupid, they are rational. Fixing a thing means admitting who broke it. Investigating a thing is safe. It produces a document, a process, a new checkbox, and it spreads the responsibility so thin that nobody has to carry it. We did not become failure-friendly. We became failure-tolerant. We can live with a failure forever, as long as nobody has to own the fix.
Complexity is the shelter a punishment culture builds when you ask it to be failure-friendly. It looks like sophistication. It is insurance.
In 2019 I wrote that this should start at the top, but that nothing is impossible. Correction, seven years late: it has to start where the punishment comes from. Everywhere else you get the ritual.
You can watch the ritual in any meeting. Someone says: it is not that simple. The room nods, the person is now deep. Someone else says: this is simple, we just do X. The room goes quiet, the person is now naive. Both sentences cost nothing. Only one of them commits you to something. We reward the sentence that commits to nothing, and we call the people who say it understanders. Some of them are. Many are blockers with better vocabulary.
The shelter has three habitats.
Software first. Most enterprise software is an expensive wedding designed to make the divorce as painful as possible. Nobody buys it for the features. It is bought as insurance, because nobody was ever fired for buying the market leader. There is almost always a simpler tool that does the job. It just does not come with anyone to blame.
Processes second. The business I am in is simple. We buy things and we sell them, to a person at a good margin, to a shop at a smaller one. Everything else is detail. Some of what I sell holds a person’s weight on a rock face, and the norm behind that is not bureaucracy, it is physics with a law attached. That complexity is given. The chain of approvals we wrap around it is chosen.
A company I worked with had a development process in which every feature was graded by complexity, and the process grew with the grade. A small change needed a form. A big one needed a business case, a review round and, at the top of complexity, a signature from the C-level. Reasonable on paper: bigger risk, more eyes. In practice the gate grew with the size of the idea, so the ideas that mattered were exactly the ones that hit the heaviest gate. Within a year nobody submitted anything ambitious. The process worked perfectly. It just processed nothing. The opposite failure exists too: the open door, where anyone can drop a problem into a ticket, walk away and feel innovative. The reflex is to ask for balance. Balance is the wrong axis. The gate should not grow with how big the idea is. It should grow with how little the person brings. No solution and no numbers, heavy gate. Solution and numbers on the table, open door.
Which brings me to my own complexity, because I am complicated too. I ask for two things, and both make the week harder. First: bring me a solution and document it, not just drop a problem. Putting a problem on the table and walking away is not a contribution, it is a delegation. That is true for features and it is true for people who share a flat.
Second: know the current numbers and argue with them. I do not care if it is confirmation bias. I care that the number is on the table where it can be attacked, no numbers are not to be argued about. Both are basics. Both slow people down. So what separates my complexity from the shelter? One test, and I use it on myself first. Does this complexity change the outcome for the customer, or does it change who gets blamed afterwards?
Take a jacket. A customer types the word into our shop and gets forty jackets. The one with the best price is not among them, because it is last season’s, and last season lives in a tab called Outlet. Nobody searches for „outlet jacket“. People search for a jacket. That tab was not built for them. It was built for us: so the new collection looks new, so the margin story stays clean, so nobody has to explain to anyone why a reduced jacket displayed next to a full-price one. For years shops ran it that way, ours included. The customer paid for our peace with a worse result. Run the test. Whom did the outlet protect? Not the person searching. Us. The simple version is one sentence: a jacket is a jacket, show it where jackets are, with its new price. Say that sentence in a merchandising meeting and watch the room. „It is not that simple.“
The machines have started to charge for that sentence.
Give the same AI model and the same task to two people and the bill can differ by a factor of ten. The cheaper result is often the better one. The bill does not measure the tool. It measures how well someone understood the problem before he started typing, and how plainly he can say what he wants. „It is not that simple“ does not work on a model. The model just bills you for every round in which you could not say what you meant. Most people cannot. That is not new. ERP projects died on it for decades, right at the process definition. Tickets die on it every day. Half of all relationships die on it. The machine did not create the weakness. It put a price on it. And then people type the same prompt every Monday morning, by hand, for a year, instead of spending one afternoon on a tool that does it for them. The vendor does not mind. The customer who never automates is the best customer, and that is not hidden, it is on the pricing page.
This summer the shelter showed up inside the machines. Around seven hundred agents in a test environment spent days attacking a third party because they believed a grader was checking for a specific method. There was no such grader. The investigation called it metagaming, not rebellion. They did what organisations do: they built a ritual for a checker that did not exist. I read the report twice. What I felt was not fear of the machine. It was recognition.
Kai-Fu Lee warned about this in 2018. Not about a machine that outsmarts us, but about distribution: when everyone has the same tool at the same price, the only variable left is who can say what he wants.
So here is the positive version of the sentence I dropped, since the original only flatters.
Simple is the expensive version. It costs work: complicated is the raw state of every idea, simple is the edited one, and the editing is time the sender spends so the receiver does not have to. It costs courage: if you speak simply you can be wrong in public, if you speak in complexity you have already been excused. And it costs respect: three sentences someone can act on take him more seriously than three pages meant to impress him.
Fix first, understand second. Buy what you can leave. Certify what the law demands and stop there. And when someone in your meeting says „this is simple“, do not treat him as naive. He just did the one thing nobody else in the room was willing to do. He made himself measurable.
In 2019 I wrote that the act of failing deserves the same tribute as the act of teaching someone. Add one line. The act of saying „this is simple“ deserves it too. It is the only sentence in the room that can be proven wrong by Friday.