There’s No Such Thing as a Bad Exercise. There’s Such a Thing as the Wrong Person.

A boxing jab and a loaded shoulder press are, mechanically, close cousins. Both drive the arm forward and up. Both involve pronation, supination, flexion and extension. Both finish with the shoulder somewhere near the chin.

If a client performed a shoulder press through the range a jab travels, I’d stop them. The load is wrong for that position, the joint is exposed, and the moment they lose control of the weight there’s nowhere for it to go but down onto them.

If a boxer didn’t travel through that range, they’d have no punch. That range is exactly where force is delivered — into a bag, into pads, into another person. Shorten it and you’ve removed the thing that makes it work.

Same movement. Opposite verdict.

This is the part of exercise prescription that almost never makes it into the conversation, because it doesn’t reduce to a list. And everybody wants a list.

The Problem With Lists of Dangerous Exercises

You’ve seen them. Behind-the-neck press: dangerous. Upright row: dangerous. Deep squat: dangerous, or essential, depending which decade you’re reading. Sit-ups: dangerous. Leg extensions: dangerous. Kipping pull-ups: dangerous.

Every one of those has been on a banned list somewhere, and every one of them is performed daily by people it’s doing no harm to at all.

The lists fail because they treat contraindication as a property of the movement. It isn’t. It’s a relationship between the movement and the person doing it — their trained state, their sport, their phase of training, their injury history, their body awareness, and what they’re actually trying to achieve.

A lunge in racket sport routinely goes further than anything you’d practise in a gym. Rugby involves loading and positions I wouldn’t prescribe to a general population client in a hundred years. Martial arts require volumes of repetition and impact that would be reckless in someone deconditioned — and are the entire point for someone who isn’t, because the brain learns through repetition and the skill demands it.

None of that makes rugby or boxing wrong. It makes the question wrong. “Is this exercise safe?” has no answer. “Is this exercise appropriate for this person, right now, for this purpose?” has one.

Most of It Takes Care of Itself

Here’s something that gets lost in the hand-wringing.

For a sedentary or deconditioned person — and honestly for most people with a year or two of consistent training who aren’t carrying an injury — the high-risk exercises largely rule themselves out. They can’t do a depth jump with enough height to generate a damaging force. They can’t get into a full pistol squat at all. The high kick doesn’t get high.

The exotic movement is rarely what injures people. What injures people is the ordinary movement, done with the wrong mechanics, repeatedly, for years.

That’s the unglamorous truth. Nobody’s shoulder went in the first week of doing lateral raises with poor scapular control. It went in year four.

What I’m Actually Watching

When I assess someone, I’m not running through a mental list of forbidden exercises. I’m watching for a handful of things that tell me what the body is doing under load.

Does the knee track over the toe, fall inside it, or push outside it. Does the heel lift. Does the foot collapse in or roll out. Is there pronation or supination that doesn’t belong in that movement. Is one limb doing the job noticeably better than the other.

That’s what I’m looking for. Not whether the exercise appears on a list.

Are You Doing the Exercise You Think You’re Doing?

This is the one I’d put above almost everything else, and it’s rarely discussed.

Watch someone do a bent-over row. A lot of the time, what you’re watching is a biceps curl in disguise. The arms are doing the work. The lats are barely contributing.

That’s the wrong action for the intended outcome — they’re training the muscle they didn’t set out to train, and wondering why their back isn’t developing. But it’s also the injury risk, and this is the part people miss. Latissimus dorsi contributes to supporting the lower back and holding spinal alignment. When the lats aren’t contributing, the load has to be managed somewhere else. So you get the worst of both: less stimulus where it was wanted, and more stress on the elbow, the shoulder and the lumbar spine.

Same with squats. A fair number of people who believe they’re squatting are, mechanically, performing a bent-leg deadlift. The hips shoot back, the torso folds forward, the knees barely travel. It isn’t necessarily dangerous. But if the goal was quadriceps development, it isn’t going to happen.

The choice between a semi-stiff-leg deadlift, a bent-leg deadlift and a squat should follow from what someone actually does with their body — not from what the movement is called in the programme.

Why So Many People Do the Same Thing Wrong

Here’s the question that follows, and it’s the one that actually matters. If the arm-dominant row is a mistake, why is it everywhere? Thousands of people, in thousands of gyms, making the identical error.

Because it isn’t thousands of separate mistakes. It’s one chain, running in parallel.

It starts with standing posture. Rounded thoracic spine, shoulders forward, head translated. Not pathological — just what happens to a body that spends its days at a desk and its evenings on a sofa.

That posture then becomes the set-up posture for the exercise. Someone hinges into a bent-over row carrying the same rounded upper back they were standing with. And now correct execution isn’t merely difficult, it’s mechanically unavailable. From that position the lats can’t do their job. The scapulae can’t retract properly because they’re already protracted before the movement starts. So the arms take over — not through laziness, but because the arms are the only thing left that can move the weight.

Then prior learning locks it in. Whoever showed them first — a mate, a magazine, a video, a gym induction fifteen years ago — set the pattern, and every rep since has reinforced it.

By the time they arrive in front of me, the movement is automatic. And automatic is the hardest state to change, precisely because it’s stopped requiring any conscious attention. They aren’t thinking about it any more, which is exactly the problem.

This is why I’m careful about how movements are learned in the first place. Unlearning an established pattern is substantially harder than learning a clean one, and it isn’t close. You’ll see specific numbers quoted for this — a few hundred repetitions to build a new pattern against several thousand to overwrite a faulty one. I’d treat those figures as folklore rather than measurement; they’re attributed to a textbook that nobody seems able to quote a page from, and they inflate every time they’re retold. But the direction is right, and any coach who’s tried to fix an ingrained squat will tell you the same.

So when I see the arm-dominant row, I’m not really looking at the row. I’m looking at the last visible link in a chain that started with how someone stands.

Which is also why correcting it usually doesn’t start with the row.

Different People, Different Permissions

I’ll give two people the same exercise and coach it completely differently.

Someone with low body awareness gets props. A dowel along the spine, a box to squat to, a wall to keep their shins off. The constraint does the teaching, because telling them isn’t working. This is not fussiness — for someone who genuinely cannot feel where their pelvis is, external constraint is the fastest route to the pattern.

Someone who’s skilled, robust, recovers well, eats and sleeps properly, and has demonstrated tissue tolerance over time? They can be taken into high fatigue, occasionally past strict form, and that has its place. Adaptation lives at the edge.

For everyone else, poor form under fatigue is too high a risk to permit. My default is keep some in the tank — roughly two reps short of failure. It’s a rough guide, not a law, but it’s the right rough guide for most people most of the time.

The same exercise, the same load, two entirely different prescriptions. Because the exercise was never the variable.

Look Around the Gym

Something I’ve watched for thirty-seven years, and it explains more than it should.

Walk into any gym. The free weights area is mostly people who put muscle on easily. The yoga studio is mostly people who bend easily — including the instructor.

You don’t often see the naturally slight person in the heavy weights section, and you don’t often see the stiff, heavy-set man in the yoga class failing to reach his toes.

People gravitate to what they’re already good at. It feels better. You get feedback faster. You look competent.

The consequence is that most people spend years training the attribute they already have and neglecting the one they need. The mobile person gets more mobile and stays weak. The strong person gets stronger and stays rigid.

So when someone tells me what they enjoy, I take note. But the more useful question is what they’re avoiding.

The Loop

Here’s how the cycle usually runs.

A conditioned, injury-free young athlete demonstrates a movement online. It looks impressive, because they’re good at it. Someone deconditioned watches, attempts it, fails or gets hurt, concludes they need to try harder, and passes the video on. My friend does pilates. My friend does kettlebells. My friend does MMA.

Take one concrete example. There’s an entire cottage industry built on deep single-leg squats with the knee travelling well past the toes, heels lifting to allow more knee flexion, dropping all the way down to the heel.

And here’s the thing — the underlying principle is legitimate. Where ankle mobility is limited, allowing some knee travel over the toe is often better than the alternative, which is the lumbar spine picking up the slack. I coach that.

But taken to the extreme, you’ve placed nearly the whole body’s load on a single hinge joint through its most compromised range. For a sedentary client, a deconditioned client, or an older client, that’s genuinely contraindicated. The principle was sound. The demonstration was performed by someone it suited. The person copying it is not that person.

What to Ask Instead

I don’t think exercise selection is complicated, but it isn’t a lookup either. Before deciding whether a movement is appropriate, I want to know five things.

Answer those and the movement usually chooses itself. Skip them and you’re guessing, however good the exercise looks.

One person’s ordinary daily movement is another person’s accident waiting to happen. One person’s essential loading is another person’s pointless workout — one that won’t achieve what they came for and carries more risk than benefit while failing to.

Which, if I’m honest, describes most of what I see in gyms, and nearly all of what I see online.