Monday, November 18, 2013

Agility is rewarding

Recently I was watching a large group of handlers 'mess around' before starting an Agility class.  I was rather gob smacked at some handlers' response to their dogs though.  Please note I am all for messing around, me and my dogs have learnt such valuable lessons while in a slightly chaotic, informal environment.  Although even a setting like that has to has its rules.  And most importantly handlers should know that it should NEVER change your feedback system. By this I mean your rewards and markers and reaction.

Controlled chaos is good for proofing and having fun too.  This doesn't mean you should have 3 of your own dogs running loose, all trying to please you at the same time.  Generally this leads to you screaming at one and affecting all.  Remember, our dogs understand body language and tone of voice, they don't ACTUALLY understand English.  If you don't believe me, ask my friend that accidentally taught her German Shepherd to lie down every time she said Damnit!  Even in craziness your dog should have your near sole attention and devotion.  I say near, because just like driving, Agility generally requires you to at least be a little bit aware.  And when I say a little bit, I mean A LOT.  While I will admit watching two people  send their dogs into the weaves from opposite sides could be amusing, it is a lot more detrimental than anything else.  Do that three or four times and instead of worrying about finding the next gate in the weave poles, Fido is going to be a lot more concerned about impending possibility of head-on collisions.  So be aware.  Of dogs, people, cars, equipment, squirrels, geese and raining meatballs.

Okay lets get to the important part though... rewarding.

Failing is training.  If you haven't failed you haven't trained.  It is important to push the boundaries and until you have not pushed it a little bit too far, you don't know where your limits are. Failing is also the difference in teaching our dogs to think and have some understanding of what they are doing and just teaching another parrot.  But at no time does this mean you should be holding your dog hostage to a reward.  Just because your dog 'failed', you shouldn't be bodily removing the reward from them, while flinging insults like a bully in the school yard.

A reward and a marker are two different things.  The reward is what the dog gets after you have marked it.  So in the dog's mind the process should be something like this: 'La-la-la, okay lets try to lie down, okay nothing, must be something different, let's try to sit,  OOOOOOH marker, yeah baby, NAILED it, now where is that damn squeaky rabbit'.  You can mark using a clicker, your voice or a French horn for all I care, but you mark immediately and then REWARD.  It is always rather amusing when I teach a class and a student comes sauntering over at leisure after doing a 15 jump sequence, just to be startled by my Boot Camp persona yelling at them to 'reward, reward, reward'.  Before listening to other people's input, before going to have a sip of coffee you really do need to reward.  Even in a class environment, well actually especially in a class environment.  Make sure you have engaged your dog in some fashion (with food, toys, cuddles), before listening to your instructor's analysis.  Honestly I would rather see handlers 'accidentally' reward (so reward to early and the dog actually made a mistake), than NOT reward immediately after the dog has done something correctly.

Pick your battles.  If you are struggling with three different sections of a 16 obstacle sequence and by some miracle, you happen to get one of these sections right.  Stop and REWARD immediately.  Don't be a hero and immediately try and run the rest.  Chances are (since you were already struggling) that you will get it wrong again at which point you are left with two options: Reward and your dog is most likely to connect it to what he last did, which was get a subsequent sequence wrong, or don't reward, leaving Fido confused as to what exactly was right and wrong.

Be in control.  People have their preferences, dogs have theirs, but sometimes people are just lazy.  You have to know the value, potential, limitations and implications of each reward you use.  For one, there are only a few instances where I choose to reward with a ball thrown ahead.  This is a very dis-connected reward, its value is higher further away from me (otherwise the fetch game wouldn't have much of a challenge now would it) and once it has left your hand, you don't have much control over it.  It is good for situations (especially with the lower confidence dog) where you want to teach dogs acceleration away from the handler.  Tug toys are in complete control of handlers, which is fantastic.  It should not be used to encourage 'velcro-ism'.  Dogs should only come into their tug toy reward once they have heard their marker.  The same goes for food.  Don't leave the reward lying on the ground if you are going to have control issues.  If your dog will shoot off in the middle of an exercise to go and find that reward.  If you were a moron and you DID do this, do not under any circumstances scream at the dog for going to get his reward.  You can't punish your dog for your stupidity and you do not want to connect any negativity to what is supposed to be your dog's reward.  Phone a Friend.  Ask someone else to help out and hold the reward and give it at the right moment.

Know your customer.  Dogs have different tolerances.  Some dogs will completely break down and phone the suicide hotline if a reward is withheld more than once.  With dogs like this, sometimes you need to offer the reward without using a marker, just to keep their confidence up.  You are basically saying 'okay you haven't gotten it right yet (no marker), but things will be okay and the world won't end today (have some food/a tug).  Other dogs don't need a confidence boost and will be happy to try and try again until kingdom come.  My Psycho Slinky for example, bless her nerdy little soul, will NOT take a reward if she feels she hasn't done something right, even when I am begging her to take it.  I also know that after a few tries of not getting it right, she will panic.  So if we fail a few times at any one thing, I will ask her to do something which she is almost guaranteed to get right (like a tunnel), mark THAT and reward.  But heaven help me if I tried a stunt like that with 'prom king' Volt, he would take advantage of me in so many ways it is not even true.  Volt was issued with 3 cc's of extra confidence when he was manufactured.  Volt KNOWS the world won't end, he doesn't need me to tell him that.

Be effective.  If you throw a ball as a reward, Fido kind of starts after it and then veers of 47 degrees North East to go and pee on a tree, your award wasn't effective.  It is fair and well to have all this theory about what toy is the best or what kind of treat, but if the reward holds no value for your dog, it is pointless.

Your feedback system is the only guidance your dog has to what is wrong or right.  So if they ARE getting something wrong, it is best you look at which part of your feedback is confusing them.

In conclusion: Reward your dogs!

Tuesday, November 5, 2013

First response - A short explanation

After some feedback from handlers that read my previous post, I would like to clarify the following (sorry international followers, again it is a South African thing).

The SA Champs regional trials and the AWC regional trials are NOT the same thing (contrary to what many handlers believe).  They follow two completely different sets of rules (Schedule 5L Appendix C applies ONLY to AWC Trials and Schedule 5L Appendix D applies ONLY to SA Champs).  Historically these two competitions have been run in conjunction (in other words the same actual rounds have been used for both competitions, but the different rules applied for the different competitions).

The regional trials would therefore still go ahead exactly as it has in the past, but ONLY the SA Champs rules/results applied to determine who qualifies for the SA Champs event.  So the top 40 large dogs, top 10 medium dogs and top 10 small dogs would still qualify for the SA Champs.

The proposal that came from us and is currently being discussed applies ONLY to the AWC regional trials.  In other words, instead of what has been happening over the years, where dogs that did NOT qualify for SA Champs and has NO intention of making Team South Africa to Europe for the AWC, but entered the AWC regional trials are invited to the finals/SA Champ, ONLY the dogs that intend to MAKE the team will be able to compete at the finals/SA Champs.  Most of which already qualified for the SA Champs.

I am hoping this clarify matters.

Monday, November 4, 2013

Here are the facts...

This is a post for my South African friends... I will try to explain the ins and outs to my international followers as well, but this post is purely about the selection of the South African Agility team for the FCI AWC.

How it currently works:

The Point System

- Points are awarded as follows:
   0 faults - 10 points
                  In addition, bonus points are awarded as follows: SCT-Dog's Time = Bonus points
   Bonus points are awarded for clear rounds ONLY

(Prior to 2013 bonus points were awarded differently: 0-1.99 seconds under SCT = 1 bonus point, 2-.3.99 seconds under SCT = 2 bonus points, 4-5.99 seconds under SCT = 3 bonus points etc)

   0.01- 5 faults - 8 points
   5.01 - 10 faults - 6 points
   10.01 - 15 faults - 4 points
   15.01-20 faults - 2 points
   20.01 - 25 faults - 1 point

- The winner in each height category is awarded 1 bonus point

The Format

- Four regional trials are run, consisting out of a Contact (Agility) and Non Contact (Jumping) round each.

- Each dog drop their worst contact score and worst non-contact score.

- Adding each dog's best three contact scores and best 3 non-contact scores, you get the trials total for each dog.  The top 10 Small dogs, top 10 Medium dogs and top 40 Medium dogs are invited to the finals.

- At the finals each dog runs two contact rounds and two non-contact rounds.

- To determine the South African Agility squad, points are calculated as follows:

Finals points + 1/3 of the regional trials points = Total points

- All dogs (irrespective of height category) are placed onto one scoring table.  The top 9 dogs (provided not more than 6 are in the same height category) will be awarded spots in the Individual competition at the AWC, the top 4 dogs in each height category will be awarded spots in the Team competition at the AWC.

The logistics

- Each year a National Course Coordinator is appointed.  This person has to source courses, historically the majority of courses were designed by local judges (in 2010/2011 it was approved that competing judges could also submit courses), in 2013 courses were sourced from the Norwegian AWC judge.

- Courses are adapted by the National Course Coordinator and coordinates for each obstacle is determined, using an A/B Baseline system.  For hurdles and the tyre jump only a centre coordinate is determined.  For the contact obstacles, tunnels, weave poles and long jump a start coordinate and end coordinate is determined.

- Each province nominates a non-competing provincial course coordinator.  The National coordinator will send the courses to the provincial coordinators once they have been selected.  Provinces can then run the regional trials in the months of January, February and March.  It is not very often that provinces can manage to run the courses on the same day.

- Should there be a mistake on the course plan or a query, the provincial course coordinators should phone the national coordinator to seek clarification.  It is also the responsibility for the provincial course coordinator to set the angles of each jump, which has to be done using a course plan (since only a centre coordinate is given for the hurdles).

- Courses are supposed to be at a grade 3 level.  Speeds used are supposed to be an average of the speeds used at the previous year's AWC.

- The courses at the final are also supposed to be of a Grade 3 level, however the judges have the discretion to set the speeds to what they want.

That is the system in a nutshell.

Now for the last 3 years, I have been researching, studying, investigating and calculating the effectiveness and fairness of the trials.  Two and a half years ago, roughly, I submitted my preliminary findings to my provincial committee.  My research included the statistics of local results, photos, video, as well as extensive research into the qualifying systems of other countries, focussing on countries with the best medal histories at the AWC.  Our provincial committee then continued the research and eventually submitted a full report and presentation to the National Agility Committee in August 2012.  Response has been few and far between, it has come to my attention that the full information with regards to our findings have not reached all handlers and/or committees.  I am therefore going to present the evidence here, as this is the most public forum I have available.

Here are the facts as to why regional trials, as we know them, are NOT a fair or practical selection system:

Before I discuss all the aspects of the trials, I would like you to read and keep the following in mind.  Remember with our point system, 1 fault = AT LEAST 2 points, and points can be directly translated into time, 1.23 seconds = 1.23 points on a clear round.  In 2013, which provided to be our most competitive year, as the AWC was local, the difference in getting a spot or not was as follows:

Small Team: Point difference between 4th and 5th dog: 4.68 points
Medium Team: Point difference between 4th and 5th dog: 2.08 points
Large Team: Point difference between 4th and 5th dog: 2.25 points
Individual Spot: Point difference between 9th and 10th dog: 2.23 points
The closest point difference for a spot, was the 0.58 points that separated the 5th medium dog (which was awarded the reserve spot) from the 6th medium dog

Looking at this model, we have to keep in mind that the regional trials only account for 1/3 of these points.  So per round in the regional trials we are looking at the following equation:

To make up the points in a regional trial:  (points missed out) x 3 (since only a third of the points will count)/6 (since only your 6 best rounds out of 8 will count) = points per round that could have gotten dog a spot in the team.

Small Team: 4.68 points x 3/6 rounds = 2.34 seconds/points per round
Medium Team: 2.08 points x 3/6 rounds = 1.02 seconds/points per round
Large Team: 2.25 points x 3/6 rounds = 1.125 seconds/points per round
Individual Spot: 2.23 points x 3/6 rounds = 1.115 seconds/points per round

This is just some silly examples to try and demonstrate the impact each small aspect of the course can have.

- Ground Conditions and Weather:  The provinces that run regional trials, vary from coastal to inland cities, this affects humidity and temperature.  Average temperatures in January (as per the South African Weather Service) vary from 25 degrees Celcius to 31 degrees Celcius. We have a province that is a winter-rainfall area and a few provinces that is summer rainfall areas.  This will obviously affect the grass growth/hardness of grounds in said provinces. According to the South African Weather Service, Port Elizabeth and Cape Town battle continually for the 'windiest city' in South Africa, with Cape Town experiencing winds of 1.6m/s 95.6% of the time and Port Elizabeth 95.7% of the time.  Meanwhile here in Johannesburg, we can hardly spell wind and certainly don't know what it feels like on a regular basis.

- Differences in equipment.  Because there are parameters for various obstacles (instead of set measurements), equipment differ very much from province to province.  Here in Johannesburg, our cross bars are 30mm pvc, some with wood-filling and others without.  Our older hurdles have angled jump cups as pictured below, while our newer hurdles have rounded, shallower cups, all metal.  From my travels, there are variations of angles cups, round rubber cups, round metal cups.

Some provinces have wood cross bars, others have pvc, varying from 30mm-50mm (all within the rules of course).  This of course will affect the statistical probably of a cross bar falling, even if hit with the same trajectory at the same force.  I will work out an average example for you of how one knock can affect a dog's final points:

The course is 150m, set at 4.1m/s (the speed used for contact agility in 2012) which puts the SCT at 36.58 seconds.  A dog running at 4.4m/s will run a time of 34.09.  If it is a clear round, he will be awarded 12.49 points, however with one knock, he will be awarded 8 points (so lose out on 4.49 points).  The dog will lose 1.49 points that it could have carried through.

The see-saw is especially important where equipment differences are concerned.  I used my own dog to compare various see-saws in the country:

For those that can't watch the videos, Gauteng's seesaw tipped in 1.11 seconds, FS in 1.51 seconds, EP in 1.79 and KZN in 1.99 seconds.

I also did a comparison of one of our lightest competitors in South Africa doing two different cloth tunnels.

Let's use the see-saw example, my dog gains 0.88 seconds on the see-saw each time.  So let's say I ran 3 clear rounds in contact agility in the regional trials that will count towards my final score, it will work out as follows:

0.88 seconds x 3 rounds = I would have gained 2.34 points in the regional trials on the see-saw ALONE.  Which will equate to 0.78 points towards my final score.  That added to all the other factors, could definitely mean getting a spot or not.  Not even taking into consideration that those 0.88 seconds I gain could very well earn me a winner's bonus point.

We also have Dog Walks lengths that differ, weave pole diameters, tunnel lengths, contact surfaces.

There is also the matter of timing equipment.  We were the last province to get electronic timing, so for years we had the advantage or disadvantage?  Now we have a timing device with multiple beams, which basically means that it is much more likely that the time will stop when the first part of the dog breaks the beams.  Other provinces have a single beam timer, which is generally set to the height where the chest of most dogs would pass over the obstacle, which means the timer would stop fractionally after the first part of the body has passed the plane of the last obstacle.

-  Judging calls.  First of all, in 2013, Johannesburg and Western Province used an up-contact judge for every single trial.  As far as my knowledge goes, no other province made use of an up-contact judge, but I could be corrected on this matter.  Once again having travelled through most of the Agility world in South Africa, I know that judges disagree very much on refusal calls.  It has recently come to light that one of the provinces in general deem it acceptable to run with a toy in the pocket, while it is a complete taboo in other provinces.

- Difficulty in setting up angles and courses.  Even with a course plan and a central coordinate, it is very hard to set up exact angles, and the smallest differences could influence a dog's time.  In the example below, you can see that (due to no fault of the people involved), a course was set up in two provinces with the tunnel two different sides of the dog walk frame.

If we once again look at it mathematically on a contact course, you needed 1.125 seconds in one round to be on or off the team, so 1 see saw and 4 spots where you could gain a mere 0.2 seconds, that would have made the difference!

Now that I have discussed some of the motivations against the regional trials, I would also like to point out certain scenarios that actually happened in the last few years, this is not aimed at anyone at all!  And sometimes things just happen, but that is the point, by sticking with a flawed system, we are opening up more opportunities for things to go wrong.

-  A video was accidentally published of a trial on social media before all provinces had completed the trial.
-  The coordinates on the first contact trial in 2013 did not work out, Gauteng and KZN was running this trial on the same day, both provinces set up the course on 'feel' from the course plan, using different coordinates.  Effectively we did not run the 'same' course
- There is a province that places second bars on all the hurdles that are jumped twice.  The reasoning is that the dog still has to 'jump' the second time if it dislodged a bar the first time.  This is not within the Agility rules, but is a regular occurrence.
- An official instructed a judge not to call faults on the break-away tyre if it broke because no other province was running with a break-away tyre.
-  A province replaced the cloth tunnel with a rigid tunnel and ran their round like this.  The national course coordinator was not informed and all other provinces ran with a cloth tunnel.
- After certain events, some provinces were instructed to not divulge any details or video of their trials at all, which to me is concerning, since surely we should be transparent in this matter.

Several solutions have been suggested for the regional trials:

I would like to start by pointing out that none of the below suggested solutions will solve the problem of equipment, grounds and environmental differences, but I will list the suggestions with my comments on them.

- Have a travelling coordinator or judge that can judge/observe and/or give input on course set up in all the provinces, in order to try and ensure that all provinces compete on the same terms.  This would be very costly first of all and is money that could rather be spent to develop Agility in South Africa.  This would also be a logistical problem, as there are 5 provinces that run the trials, so that is 5 provinces, 4 trials which would equate to 20 separate days that would have to be booked for running trials.
- Set two coordinates for each jump to determine the angle as well.  Which would mean double the work for our poor National course coordinator, but we would still sit with tons of problems.
- Standardise equipment across the country.  Well this is the dream of course, but honestly, most of the provinces don't have the money to upgrade all their equipment at once, this would take many, many years.  A person/body would have to be determined to oversee this project.

Why do we have trials?

The truth of the matter is, that even in 2013, when the amount of people to declare interest in being selected for the team almost DOUBLED, we only had 25 large dogs, 7 medium dogs and 7 small dogs paying their deposits to make Team South Africa.  In the last 5 years, ALL dogs that entered the AWC trials were invited to the finals.  Therefore the trials has NEVER been used to short list dogs or limit the numbers, as we just don't HAVE the amount of registered dogs.  In fact, MANY more dogs run in the trials than dogs that put down their deposits.  So why are we allowing more dogs to compete in the finals than are intending to go to the AWC?

The main motivation for the trials is ensuring 'consistency', which I am all for (the consistency that is).  But should dogs be penalised purely based on which province they live in?  As for finding other methods to ensure consistency is taken into consideration, please see some of my research below.

Having researched many qualifying systems and not including selection panels that some countries make use of, there are generally three qualifying methods that are followed:

- Handlers have to travel across the country and compete in different parts of the country.  This qualifying always leads to a final that is held over a weekend.

- Dogs have to meet certain requirements (a certain amount of clear rounds/wins/speeds) to be eligible and then travel to a selection weekend.

- A single weekend event where all dogs eligible (and only those dogs with intent to go) get together and have to run their rounds.

Now for many South African handlers, it would become much to expensive to travel across the country and then once they have made the team, pay a huge amount of money to travel to Europe.  So I don't think the first option is really viable for us.

The second option would be viable I think, determine a certain amount of clear rounds each dog has to run at certain speeds.  Example:  The requirement to go to the final is 6 clear rounds, 4 of which have to be contact agility, at a minimum of 4m/s in the year preceding the Finals.  The objections that was brought to this is that some provinces have fewer shows than others, as well as the difference in measuring from judge to judge.

The third option is our best option in the current environment, however I would suggest increasing the final from the current 4 rounds, to 5 or 6 rounds to increase the consistency requirement.  This option also most accurately simulates the actual AWC event, which tests the handler's ability to handle the pressures of such a big weekend event.

Please note that while the actual proposal is the work of many people, including input of team mates from various provinces.  I am open to actual, factual input.  I am open to good suggestions.  You are welcome to ask away if you may have any questions.  However if you are one of those that are just planning to say 'because' but can't counter-argue my facts, maybe you best rather not comment.

While I have the regional trials to thank for my spots on the team in both 2012 and 2013 (since I had rather terrible finals), I still feel it is not the way to go and doesn't make logical sense.  I really hope that this post serves to broaden the awareness of all South Africans that has AWC dreams in their future.