For most of my first two years in magic, my practice sessions had a single goal: make the routine work. Run it until it is smooth. Repeat until the method is clean. Polish until the timing is natural. Get it right, then get it right again, then get it right one more time.
This approach produced competent routines. It did not produce great ones. And it took me far too long to understand why.
The breakthrough came during a conversation with Adam Wilber. We were on a video call — one of our regular Vulpine Creations meetings where we discuss product development, performance concepts, and whatever else is on our minds. I was showing him a mentalism piece I had been developing. I performed it for him over the camera, and he gave me that look — the one where you can tell someone is thinking carefully about what to say.
“It works,” he said. “Now try to break it.”
I did not know what he meant.
“Perform it again,” he said, “but this time, try to make it fail. Try to find the thing that makes it not work. Do everything you can to expose the weakness.”
I tried. And in about forty-five seconds, I found a problem I had missed in weeks of practice.
The problem was not in the method. The method was fine. The problem was in the framing — a specific moment where, if the spectator made a particular choice that I had not anticipated, the logic of the presentation broke down. In all my practice runs, I had been making the “right” choice for the spectator, unconsciously guiding my own imaginary audience member toward the path that made the routine work. I had never considered what would happen if they went the other way.
Adam was not inventing this approach. He was applying a principle from scientific methodology — one that Gustav Kuhn describes explicitly in his research on how to evaluate magic: “To discover why a trick works, try to look at what you need to do to stop it from working.”
The Science of Falsification
The principle behind “try to break it” is falsification — the idea, rooted in the philosophy of science, that you learn more from trying to disprove something than from trying to prove it. I discussed confirmation bias in my last post — the tendency to seek evidence that confirms your beliefs. Falsification is the antidote. Instead of asking “does this work?” you ask “under what conditions does this fail?”
The logic is straightforward. When you try to make a trick succeed, and it succeeds, you have confirmed what you already believed. You have not learned anything new. But when you try to make a trick fail, you get one of two outcomes, and both are useful:
If you cannot make it fail, you have much stronger evidence of its robustness than any number of successful runs could provide. You have actively tried to find weaknesses and found none. This is genuinely informative.
If you can make it fail, you have discovered a specific vulnerability that needs to be addressed. This is even more informative. You now know exactly where the routine is weak and can fix it before a live audience finds the same weakness.
Kuhn describes a research experience that illustrates this perfectly. He had been performing and studying a particular method for nearly three decades. He understood it. He was confident in his assessment of how and why it worked. Then he designed an experiment that deliberately tried to break it — an experiment structured to find the conditions under which it would fail. In his own words, that single experiment gave him a better understanding than thirty years of performing it had.
Thirty years of successful performance, outweighed by one deliberate attempt at failure. That ratio should make every performer sit up and pay attention.
My First Deliberate Breaking Session
After the conversation with Adam, I cleared an evening in my home office and set up what I can only describe as a stress test. I took five routines from my working repertoire — pieces I performed regularly, pieces I considered strong and reliable — and I tried to break each one.
The rules were simple. For each routine, I would perform it while actively looking for problems. Not the usual practice where you check that everything looks clean. The opposite: assume something is wrong and try to find it. Play devil’s advocate with your own material. Be the hostile spectator. Be the person who does not want to be fooled.
The first routine survived. I could not find a meaningful weakness. This was genuinely reassuring, in a way that a hundred successful practice runs had never been. I had tried to break it and failed, which meant it was stronger than I thought.
The second routine broke almost immediately. There was a moment — a transition between two phases — where the internal logic depended on the audience not noticing a specific inconsistency. In all my performances, they had not noticed it. But when I deliberately looked for inconsistencies, it was glaringly obvious. The routine had survived dozens of performances not because the inconsistency was invisible but because audiences are generous and do not normally scrutinize transitions. A more attentive audience — a more skeptical spectator — would catch it.
The third routine had a physical vulnerability I had never considered. There was a specific angle — slightly to my left — from which the method was partially visible. In practice, I had always worked from the front. On stage, I had mostly worked from the front. But at a corporate event, with people seated in a semicircle, someone sitting at the left edge could potentially see what they should not see. I had never checked that angle because I had never thought to check it.
The fourth routine broke in a different way. The script contained a moment where I made a claim — something I stated as fact as part of the presentation. The claim was not false, exactly, but it was ambiguous enough that a literal-minded spectator could interpret it differently from how I intended. If they interpreted it the other way, the routine still worked as entertainment, but the apparent impossibility was diminished. The impact was weakened by an imprecise sentence.
The fifth routine survived, like the first. Two out of five held up under deliberate attack. The other three needed work.
That evening was the most productive practice session I had ever had. More productive than any ten sessions of running through routines to confirm they worked. Because instead of practicing the same smooth performance I always practiced, I had learned three specific things that needed to change. Actionable information. Concrete improvements.
The Practice Inversion
I now structure a portion of my practice time around what I call the practice inversion. Instead of practicing to make things work, I practice to make them fail.
The process looks like this:
First, I perform the routine as intended. One clean run, start to finish, to establish the baseline. This is the version I would do on stage.
Then I perform it again, but this time I change one variable. What if the spectator is standing to my left? What if they are taller than me and looking down? What if they choose a different option than the one I expect? What if they ask a question at the wrong moment? What if the venue has different lighting? What if the table is higher or lower than usual?
Each variable change is an attempt to break the routine. Most of the time, the routine survives. The variable does not matter. But occasionally — maybe once in every ten or fifteen attempts — the variable reveals a weakness. A sightline problem. A timing issue. A script ambiguity. A logical gap.
Those occasional discoveries are where all the learning happens.
I then perform the routine again with a hostile mindset. I pretend I am an audience member who wants to figure it out. What would I look for? Where would I direct my attention? What question would I ask that the performer does not want me to ask? What would a data security specialist in the second row — like the man in Klagenfurt I described in the last post — notice?
This hostile-mindset pass is uncomfortable. It requires me to work against my own interests. Every instinct says “practice until it works.” The inversion says “practice until you find where it fails.” The instinct feels productive. The inversion feels destructive. But the inversion produces better results, every time.
Why Magicians Resist This
There is a reason most performers do not practice this way: it is psychologically unpleasant. Deliberately looking for flaws in your own work requires a kind of intellectual courage that goes against the grain of how we normally relate to our creative output.
When you create something — a routine, a presentation, a performance — you invest in it emotionally. It becomes yours. Finding flaws in it feels like finding flaws in yourself. The instinct to protect the work from criticism is the same instinct that protects the self from criticism. Confirmation bias is not just a cognitive shortcut. It is an emotional defense mechanism.
Breaking through this defense requires separating your identity from your material. The routine is not you. A weakness in the routine is not a weakness in your character. A flaw in the method is not a flaw in your worth as a performer. This sounds obvious when stated directly, but in the moment of practice, when you have spent weeks building a piece and someone (even the someone that is you) starts poking holes in it, the emotional resistance is real.
I found it helpful to adopt the mindset of an engineer, not an artist. Engineers test things to failure on purpose. They build bridges and then calculate the maximum load. They design systems and then simulate catastrophic failures. This is not because engineers are pessimists. It is because they understand that the only way to know how strong something is, is to find out how much force it takes to break it.
Your routine is a bridge. Your audience walks across it. You need to know where it will fail before they find out.
The Adam Wilber Rule
Adam and I now have an informal rule for any new product or routine we develop for Vulpine Creations: before we consider anything finished, we try to break it. Not gently. Not politely. We attack it from every angle we can think of.
What if the spectator is left-handed? What if they do not follow instructions? What if they are not paying attention? What if they are extremely attentive? What if they have seen a similar routine before? What if they work in psychology or neuroscience and know the principles? What if the lighting is bad? What if the table wobbles? What if the person next to the spectator leans over and watches from a different angle?
Most of these questions produce the answer “the routine still works.” Good. But the few that produce a different answer — those are the ones that make the routine better.
Darwin Ortiz writes: “The obvious explanation is usually the last one to occur to magicians but the first one to occur to laypeople.” This is another way of stating why breaking your own tricks is so important. When you practice normally, you think like a magician. You focus on the cleverness of the method, the smoothness of the execution, the elegance of the sequence. When you try to break the trick, you are forced to think like a layperson — to consider the obvious interpretations, the straightforward explanations, the simple observations that you have been trained to overlook.
The Takeaway
Every successful performance tells you your routine might work. Every failed attempt to break it tells you your routine probably works. And every successful attempt to break it tells you exactly what you need to fix.
Stop protecting your routines from criticism. Start subjecting them to the harshest criticism you can generate. Try to break them. Invite others to try to break them. Run them under adverse conditions. Change the variables. Adopt the hostile mindset.
The routine that survives deliberate attack is a routine you can trust. The routine that has only survived friendly performances is a routine you hope will work. There is a difference between trust and hope, and the difference matters most on the night when the audience is not friendly.
That forty-five-second discovery on the video call with Adam — the vulnerability I found in weeks-old material simply by trying to make it fail — saved me from what would have been an embarrassing failure at a real event. Forty-five seconds of deliberate breaking outweighed weeks of confirming practice.
Try to break your own trick. If it breaks, fix it. If it does not break, you have earned something that practice-to-succeed never provides: actual confidence.