How do I stop my dog running off on walks?

Luna's file is where the radius is built, tested, scored, argued with and given away, and it is the case that should define the concept. She is a nervous case typed on day one as environmental rather than genetic, and the radius is both her treatment and her thermometer: a nervous dog's radius is wide, so tightening it is the physical proof that the nerves came down - "it was affecting her radius". The procedure is stated to the owner in full: set the leash to a specific length, give her the e-collar every time she hits the end of it for a session or two, then loosen the leash and let the occasional excursion go. Free room first, then close the circle. What is left is the "mental field" - a sense of distance that survives any leash length because she is answering the collar rather than the line. It is scored on a long line a week later as "locked in fairly decently", given an explicit upper bound - not glued to me - and then handed over as a dial the family may turn: he tunes it to a medium and tells them how to keep it where they want it. And the same file contains the strongest argument against the whole method, twelve days earlier, in what reads as the same voice.

Sage gets the mildest version, because she is nervous and because the handlers working her are junior. Her number is the small-dog number - about ten feet, eight to ten - and it is confirmed out loud by the senior handler when her handler asks. The enforcement is barely there: an unlocked line, one almost imperceptible tug when she turns toward a distraction, a correction reserved for actually pushing past. It also produces the clearest illustration in the corpus of a problem nobody warns owners about: a dog who never leaves the radius cannot be taught the radius. In the store her handler says he would correct her for going out of it, but she has not given him the chance - and in an earlier session the senior handler has to trick her into going far enough away to show where the boundary is. Sage's radius is the one that is talked about rather than shown, and it is honest about that.

Deuce is ninety-six pounds of ten-month-old dog with no idea how strong he is, and his file supplies the two things the rest of the corpus lacks: the mechanism and the owner-facing test. The mechanism is that the pressure teaches, not the length. The e-collar is fired at the exact instant the flat collar comes tight, so the lesson is about the sensation of the end of the line rather than about a number of feet - which is why the rule holds on a leash locked short and on one fully extended. The test is the shortest sentence in this concept: he doesn't hit the end of the leash, we're good. And his lesson catches the failure mode of success on camera. He has stopped hitting the end and started riding it - "he's not hitting the end but he's hanging out right at the end with you" - which looks like a pass and is not, and the answer is to lock the line shorter for a while until the radius resets.

Louie's file is where the method is taught rather than used, and where it is called house standard rather than preference. The number is the big-dog number, fifteen to twenty feet at most, and the installation is one sentence: recall him every time he gets to the edge, and he will start checking in and circling back without being told. It also carries the argument for why the radius is not just loose-leash walking under another name. Louie does not pull; he lives at the end of the line, and the handler treats that as a fault to fix rather than a standard met - he watches the leash and the collar continuously and tightens the radius up rather than letting him ride the end. In a later session radius and recall are described as one job with two names.

Bullet arrived at zero - no sit, no down, no heel, no recall - and finishes the program with ten to fifteen feet of self-imposed range and a line he wears for other people's benefit. His session is also the one place in the corpus where a trainer says out loud that this behaviour is trained for the client rather than for the dog: he would not bother installing a radius on his own dog, and he does it because clients want it and adapting to what is in the client's interest is the job. The method is the same as everywhere else - cue him every time he reaches the edge, either a recall or a let's-go - and the result is described as inbuilt, meaning the dog's own range rather than the equipment's.

Dolce is a protection dog whose obedience ran months ahead of her confidence, and her file treats the radius as a named exercise in its own right - worked before the leash comes off, separately from both heel and recall, and enjoyed: "radius is a thing it's i i enjoy doing radius quite a bit". Three different handlers describe it independently across her sessions, which is unusual in this corpus. Two things are distinctive here. Its real product is named as self-control rather than obedience - the perimeter is what makes dropping the leash thinkable. And by mid-July it is what a session is scored on: not the commands, but the fact that she keeps herself in range and checks in unprompted.

Dutchess supplies the criterion, and the way to test it. Her last recording before the drive home is a short off-leash check in which off-leash readiness is defined as an uncued behaviour rather than as obedience: she checks in with him without being called, and that is what an owner should be looking for. What makes it method rather than observation is how he tests it - he deliberately keeps quiet and stays off her, so that the answer cannot be something he produced. It is the only definition of off-leash readiness anywhere in her file, and it is checkable by a family with no equipment and no trainer present.

Lola's file argues the negative case, which is the half most owners never hear. Good-looking commands are worthless without a check-in, because a dog who does not care where you are cannot be reinforced at any distance - and the counter-example is a shepherd with immaculate obedience who ran halfway through a Home Depot before noticing his handler had gone. For Lola, a nervous pound dog, the check-in on a long line is the thing praised at the end of her last training session, above any command. The households' own version of a successful walk, recorded in her second owner lesson, is the same measure in lay words: the dogs stay with the group and look back, rather than being held close.

Zeus is the refusal, and it is the most interesting entry in this file. His handler will not install a radius at all. He teaches the words and the communication system; the number is the household's to choose, because anything he built would be generic and would then have to be undone - if he set it further out, the family would have to bring it in. So what is handed over is a parameter and the tools to set it, not a finished behaviour. He is also the one who says what actually makes it real, and it is not the number: it is never changing the number. An inconsistent radius produces a confused, frustrated dog rather than a boundary. His estimate for how long a consistent family needs is five or six walks. This is the clearest instance in the corpus of a program handing over a design decision rather than a result.

Finn is the case where it is named as an assignment and then explicitly withheld. He is a major flight risk who bolts, and on his first trip out the radius is set as his next piece of work alongside the recall. Weeks later, with a handler new to him, it is deferred again in plain terms - not tonight, he is not ready, he needs to ease up around her first. His file also supplies the cleanest owner-facing framing of why the radius is not the same as not pulling: your dog may not pull and still run to almost the end of the leash, and that is chaotic. By the go-home lesson he is reported keeping his own radius while other dogs have to be called back onto one.

Roxy's file gives the order of operations and one unusually honest correction. The order: improve the recall first at whatever distance the dog currently works, pay it with release to freedom, and only then tighten the radius - then more distance, then a long line, then off leash. The radius is treated as the unit of freedom rather than leash-on and leash-off. The correction comes on her housemate Ramsey in the same set: the handler holds the leash at the intended length and corrects each time Ramsey hits it, Ramsey panics at the collar, and the handler says so on camera, apologises to the dog, halves the level twice and takes the e-collar out of the session entirely for a while. The radius work is what produced the mistake, and the mistake is left in.

Is the thing being taught a distance, or a response to pressure?

Both positions are in Luna's file, in what reads as the same voice, twelve days apart. The Tier 2 arc flags it as a live tension with itself and declines to average them. It matters practically, because the two accounts imply different drills: one asks you to fix the length and mark the edge, the other asks you to make leash pressure meaningful at any length and never let the dog become dependent on it.

Deuce's file reads as a third answer that dissolves the problem - fire the collar at the moment the line comes tight, so the dog learns the pressure and the length is incidental - but nobody on film connects it to Luna's objection, and it is not offered as a reconciliation.

Should the program install a radius, or hand the family the dial?

Both handlers are talking about the same handover moment and reach opposite conclusions about what the client should receive.

Nothing in the record reconciles these. They are two handlers' answers to the same question about the same deliverable, and the decay argument cuts both ways: if the radius goes when it is not reinforced, a trainer-built one is the one most likely to go.

Whose benefit is the radius for?

Not a disagreement about method but about justification, and it sits oddly against the rest of the corpus, which argues that training enlarges a dog's life.

Both are true statements about the same behaviour and the corpus never puts them side by side. It is one of very few places where a trainer says a behaviour exists for the owner's comfort rather than the dog's benefit.

Where to start the e-collar level when marking the edge of the radius.

This one runs against a concept already on the site. `start-high-not-low` is in the bible; the radius work in Luna's and Deuce's files does the opposite, and one of the two handlers notes that starting low is unusual for him.

Ramsey's session shows the cost of getting it wrong in the other direction: a level that read fine a minute earlier produced a panic at the end of the line, and the handler halved it twice on camera. Nothing in the record settles which order is house standard for radius work specifically.

How many feet should the radius be?

The numbers the trainers actually say are eight to ten feet for a small or nervous dog, ten to fifteen for one dog, and fifteen to twenty feet at most for a bigger one, described in that last case as house standard rather than personal taste. But the most careful answer in the corpus is that the number is yours. One handler refuses to set one at all, because anything generic he built would have to be undone to fit your household - he would happily let his own dog go thirty feet, and he knows some people do not want theirs past ten. What is not negotiable, on his account, is that you pick one and stop changing it. The number is arbitrary; its constancy is the whole mechanism.

How do I actually build it? What do I do differently on tomorrow's walk?

One thing, done without exception. Put him on a line long enough that the distance you have chosen is inside it, and do not lock it. Let him range and sniff. The moment he reaches the edge of your distance, cue him - a recall if he is ahead, a 'let's go' if he is behind you - and then let him go back to what he was doing. Every time, not when you remember. What the trainers report is that the dog stops needing the cue: he begins circling back, or stopping to wait for you to catch up, on his own. One handler puts the timescale at five or six consistent walks before it looks like a habit. The other half of the job, if you are using an e-collar, is that the marker or the tap lands at the moment the line comes tight rather than at some length you are counting, because the dog is learning about the pressure and not about your tape measure.

How do I know whether it is working, without a trainer watching?

Two tests, and both are free. The first is the leash: it never comes tight. Not that he returns quickly when it does, not that he yields nicely - that it does not happen. A trainer states the pass mark in six words on a street walk with a Rottweiler who had pulled hard enough at intake to worry him. The second test is for off leash, and it is a test you have to be quiet for. Stop talking. Do not call him, do not raise your voice, and watch whether he comes back to check on you by himself. A handler filming a final check before a dog went home does exactly this and says so out loud: he stays deliberately quiet so that the check-in cannot be something he produced. A dog who checks in when called has told you nothing.

Isn't this just teaching him to be clingy?

The trainers say no, and they set the ceiling explicitly rather than leaving it implied. "I don't want them to be glued to me. That's not what you want anyways" is said while judging a radius acceptable, not while correcting one. In one owner lesson the over-checking dog is named as its own fault - a dog walking two feet ahead of you and looking back every few seconds is not being a dog, and the handler says he dislikes it. The point of the radius is that the dog has a boundary and does his own thing inside it: sniffing, ranging, exploring. If your dog has stopped exploring, you have gone too far, and the corpus has a separate warning about the version of this that happens with the heel command.

He was great at this when he came home and now he isn't. What happened?

Nothing unusual, on the trainers' own account. The radius is described as the skill that decays fastest of anything they teach: "the radius goes and it makes awfully significantly harder to to do". It is the reason the last session before a handover is spent maintaining it rather than showing it off, and the reason one trainer refuses to be building off-leash skills on the day a dog goes home. The thing that keeps it is the same thing that built it: the edge producing the same response every time. If you stopped cueing at the edge because he stopped needing it, the cueing is what was holding it up.

Do I need an e-collar for this?

Not to build it, on the evidence here. The installation method that gets stated most often uses no collar at all: cue him at the edge, every time, with a recall or a let's-go. Bullet's ten-to-fifteen-foot radius is described as built that way, and Sage's is enforced with an unlocked line and one barely perceptible tug. Where the collar does appear, it has a specific and limited job: it marks the edge for a session or two in Luna's case, or it lands at the instant the flat collar comes tight in Deuce's, so that the lesson is about pressure rather than about leash length - and that is what is supposed to survive the leash coming off. One handler makes the argument for it plainly, that it is the tool that lets a dog have more freedom overall. And one session shows the cost of a level chosen badly: a dog who panicked at the end of the line, an apology on camera, and the collar taken out of the session.

Treating 'he doesn't pull' as the standard. A dog can walk without pulling and still live at the end of the line, and the trainers call that chaotic rather than trained.

Changing the distance from walk to walk - tight today, the whole length of the leash tomorrow - which produces a confused and frustrated dog instead of a boundary.

Holding the dog so close that he never gets the chance to leave the radius - which means he can never be taught it, and gets no useful information when he does drift.

Accepting a recall in which the dog touches base and leaves again. It looks like a check-in and it is not, and it is the specific way owners break their own recall.

Running the walk on leash pressure, so the dog is obeying the line rather than the handler. It looks identical on a leash and it makes off leash impossible.

Over-tightening it, so the dog ends up glued to the handler. The trainers set an upper bound as well as a lower one, and a dog who over-checks is named as a fault of its own.

Letting it lapse once it looks finished. It is the fastest-decaying skill in the program in one trainer's account, and it degrades without being noticed because the dog still walks nicely.

He walks nicely and never pulls, but he spends the whole walk at the far end of whatever line he is on.

His commands look good and he still does not seem to notice where you are - you could stop walking and he would keep going.

You keep calling him back. If the way you keep your dog near you is a stream of recalls, the thing you are missing has a name.

A nervous dog who ranges further than you would expect. One trainer reads a wide radius as a symptom of the nerves, and its tightening as the evidence they came down.

He is fine on the leash and you do not trust him off it, and you cannot say what would change your mind. The trainers can: an uncued check-in.

He stopped hitting the end of the line and now rides it - always out there, never quite tight. The trainers treat that as a regression that looks like progress.