Interesting thread. Especially seeing the difference between training rally and agility. Clicker training has been great and yet, drives me crazy. Learning to do it correctly is great. Me doing it, well, I have my moments of greatness and my moments of sheer stupidity. Someone needs to click me when I click at the right time. LOL. I find it to be a great tool.
For those dogs that are noise sensitive. It has been suggested to use a quieter item that makes a single click noise. Like a pen. If it's too much, like my friend's dog, then a verbal is good, although not as consistent (our tones change). For the deaf, it has been suggested to use a flash of a light. All you need is something short (light, sound, etc) that is consistent. Our voices do work, but again, we change our tone so much, it just takes longer. And for some reason, the clicking noises goes straight to the brain (I forget where/why, but there's a book explaining it all). C/T can be used for humans too.
I was testing it on my nephews. It was working, if I was around them long enough, but I had to keep finding new "treats".
I have to say, Mom2Melli, I do understand what you're saying and why. I thought the exact same thing. So when I was working with distance in agility with Koji, I'd click, have him do something else, click, and treat at the end. I later found out, that yes, I was doing it incorrectly. The whole click/treat thing is really fascinating because it is pure science. I forget the book, but it talks about how clicker training first came about (back in the days of training wild dolphins at Marineworld - or Sea World, I dunno which one).
After doing a lot more research into clicker training, because hey, I felt I was right, I learned so much. Of course, I was wrong. The clicker should not be the reward. With Koji, he sees me with a clicker and he's at attention. He knows it means food.
The one thing I'd like to say is....remember, a reward can be almost anything. In agility, when Koji stops at the end of the dogwalk, while on the course, me saying yes is his marker. Me releasing him to go play (aka run the rest of the course) is his treat. And unless I walk off the course with him, he always gets a treat for running. In class, it's when you use the NRM and stop and redo if necessary. No treats.
In regards to this comment. I'll tell you how I learned it. And how I was taught to do the distance work, while C/T from far away.
So that follows with distance work. Let's say have the dog on her 20 foot lead and I am with her 19 feet from her rug. I send her to the rug. I click. She turns to face me. I say/sign down. She downs. I click. I say/sign stay. I walk the 20 feet, circle her, and return to the heel position. I click and treat while she is still down IN position. I release her and we heel off. How could I treat for each step occurring 20 feet away from me? The click is the confirmation the action was right and a treat is coming at the end of the sequence. It is a given that each of go to rug - down - stay were individually taught prior to mastery level with click/treat. Chained, the reward comes at the end of the treat but the clicks in between are a YES confirmation of correct action. I don't think everyone is making the distinction of one single basic task (such as sit or down) being taught and a chain of mastered tasks to form a more complex maneuver (such as send to rug/down/stay or call forward return right).
First you break everything down. And train until it's proofed. Actually, the 3 things I listed are overly simplified and would need to be broken down even further. When every item is proofed, you don't need individual clicks between.
-Send to rug
-Down
-Circle while in down
We taught "go to your mat". First you teach the dog, while on the mat, it's a good thing, C/T. Then you teach your dog down on the mat, C/T.Then you start moving away, 1 foot, 2 feet, etc. C/T by tossing it. When you get too far to toss or you can't toss it accurately, after you click, you WALK BACK, treat and walk away. THEN release dog from position. If the dog breaks the position, say as you're walking to them, you wait until they get back into it. WITHOUT giving another command of any sort, then treat. If the dog breaks position after you given the treat, same thing. Wait for dog to get into position, then release, WITHOUT an extra command. They have to learn that the ENTIRE sequence is the thing you want, not just parts. Once that's proofed, then you move onto say, the walking around the dog.
So, in your case, by the time you start walking around the dog, the dog should already know the commands to be sent out and down and STAY down until you release. You'd start out circling close by. C/T. Then further out, C/T, further out, C/T, just like the send to the mat. By the time you are circling 20 feet out, dog should know the send (no need for C/T), should know the down (no need for C/T), and know to stay until you release (no need for C/T). Obviously, the circling is the last thing, so once that is proofed, then you C/T for the ENTIRE sequence.
If you break things down and the items are proofed, you shouldn't have to C/T anymore. C/T is for training and reinforcing at times, but it shouldn't be needed all the time. That's like if I C/T for every obstacle my dog did in agility, when in truth, by the time he can do 20 obstacles, he shouldn't need the C/T anymore.
Here's a
list of how I trained his dogwalk contact. Do you see how detailed it is? How much I broke it down? Truth be told, I probably should have broken it down even further. By the time he runs across a dogwalk and gets into position, everything leading up to that should be proofed. And therefore, not needing a click. Versus me clicking him running across, looking for his target, slowing down, stopping in position, waiting for me to give the next command, etc.
BTW, Koji knows he'll get a treat after his run. He pulls hard to get back to our area for just that. He has been trained that if he completes the course, he gets a treat. I don't need to C/T. If he doesn't do something I want, like the dog walk, I give him an NRM on the course, stop what I'm doing, pick him up and walk off. He gets no treats because he did not complete the course and didn't do what I asked. He knows what was not completed (this is akin to you stopping in a yes-no-yes sequence - as you said, go back and redo, no treats). But the difference is, I'm not clicking him for completing every obstacle. He should know what to do already.
I was told, when you click, the dog gets a reward. EVEN if you screwed up the click. You canNOT click and NOT give a reward (of any kind). You are teaching the dog the clicker has no meaning (or is making it confusing). Now, maybe that's not true for your 2 dogs, but I can guarantee if you did that with 100 dogs, there are going to be a lot of very confused dogs.
I hope you realize I'm not saying what you're doing isn't working, obviously it is. But there is a science behind clicker training and why it works, when done properly. Remember, a stopped clock is right twice a day. I learned that I was wrong. I mean, it was working, so what's wrong with it, right? But I decided that I wanted to learn correctly, for the next dog, and the next, and the next. Not because "it works" for Koji.
I hope you really do some of your own research into the science behind clicker training. It'll give you a new found respect to why it works and why it can work consistently with any animal (and even humans), if done correctly.
)
while writing it!!
). I certainly remember the rewards being rapidly dispensed. 
The 'jackpot' idea was one I struggled with initially. I must be a stingy person as till then a tub of treats would last several days rather than one training session! 
