Forum Message

Topic: Re:snowballing effect in modeling?
Posted by: Robert Ballentine
Date/Time: 12/12/2003 12:10:59


One of the purposes of modelling is not only to reduce the trial and error phase, but to avoid at all costs, if possible. What would be the point of it, the purposes of modelling is both in the replication as much as the efficacy.

Imagine, when a newbie (excuse me please, have to use cricket as an example) batsman or batsperson stands in posture…

(feet together positioned adjacent and about 1/2 meter away from the wickets, cricket bat held in both hands, with the tip of the bat resting on the ground in front of the toes forcing the bending of the back and the batspersons head twisted through 90 degrees to face the attention of the bowlers oncoming piece de resistance) 

…the learner is to face a ball made from leather which is hard enough to cause serious injury, which is then bowled in a manner depicted of correct cricket; that is, it is swung with the arm behind the bowler in a forward arc and released at around the 10.00o'clock position towards the batsperson with the speed ranging anything from 20 -120 miles / hour. Hopefully the former for the leaner.

At this point, if the ball is travelling too fast the learner will do the honourably wise thing and dive out of the way of the oncoming ball, or if more tenacious and spirited attempt; in that great English tradition of stiff upper lippedness, to hit the ball.

Now depending upon the trajectory of the ball, the batspersons innate skill and abilities of hand eye co-ordination to oncoming solid objects at speed, a number of outcomes could happen, not least of all, head butting the ball, batting the ball or the ball meeting it's target and knocking off the wooden bails from the stumps.

So, take this through to the learner; not having been scared off by solid objects flying through the air at them, having spent weeks, months and perhaps years of 'trial and error' in learning the different non-verbal cues the a bowler will signal to the batsperson in what type of bowling action the bowler will use either to spin the ball top or bottom, aim the ball off wicket, or on wicket, straight at the wicket, bounce off the grass etc etc.

And then the unconscious signals of the firmness of the ground to the feet of the batsperson, indicating the likelihood of how the ball will respond to bouncing off the grass, with the unconscious signals of the wind and how that effects the trajectory of the ball etc… starting to get the picture.

So that after years of this, the batsperson has created intuitive patterns of behaviours set within the boundaries of the scope and range of his or her behaviours, reflective to that from the diverse bowling styles, cues and environmental conditions, that result in; accordingly to our imaginary model of genius, a set of commensurate skills of excellence.

Now, supposing we model this person, and after 'however long a period' we arrive at the conclusive evidence that our ability to replicate the defined behaviour in our said 'context' (this being a definitive variable worthy of note when modelling) meets the criterion set out in whispering. That is we are able to perform near exacting behaviours to the model and achive near exacting responses, for in cricket, hit the ball and achieve either the ball out of the boundary, for 6 points or rolling out of the boundary for 4 etc etc, depending upon the criteria set for modelling this person of excellence.

So now we have the ability to replicate the models behaviours and achieve our desired result. Set within the confines of a) our own physical abilities and b) the constraints of the context of the modelling project.

Context: just a note here. Context is an important variable within the modelling project. This defines the scope and boundaries of the project, that is, what portion of the models behaviour are we going to model. For instance - All of it, or partial, or specific, - All of it, we could say, includes, different weather conditions, different pitches (grass, grounds, stadiums), different bowlers, top spinners, bottom spinners etc. different cricket bats… and the list goes on. Partial - any subset or division of All. Specific - a specific subset of behaviours that come from Partial.

So the division of these sets would be driven by the modellers own way in partitioning the sets of experience called cricket.

Example of partition = modelling batting against top spinner bowlers. Which could include, the positioning of batsperson on the pitch to when the bowler makes their run and then bowls, to the extradition of the ball from the batsperson's cricket bat to the other side of the pitch.

Example of Specific = modelling the positioning of the batsperson on the batting line.
Well, lets now take it to the point in question, that is, what about all the errors and learning the person has taken throughout their journey to arrive at excellence and all the 'great lessons' they have learnt along the way.

Firstly, 'Great lessons' is a personal criteria judgement based upon their own interpretation of their experiences. 'One persons seat is another's mile stone' or 'one person's food is another's poison'

Well, yes, the model would have undoubtedly learnt many things, including what to do when a ball is flying at 120mph towards their head, and how to avoid it, (hopefully) but again this is part of the context of modelling. What do you want to model?

So when there are, the examples of - "what to do when…" or "what if x happens, then what…" these are all the consequential parts of different modelling partitions within the larger frame of the modelling project, if it has been included in.

However as modeller, you have the choice to arrange and set the contextual partition as you see fit for your own purposes. However one of the criteria that maybe outside of your control, is the level of access you have to the model in question. If you have a limited amount, again, this may narrow and filter / streamline the area to be modelled.

So these 'great lessons' are a) subjective b) contextual c) relegated to classes of behaviour that do not perform and achieve the required outcome

Or to offer another example - I raise my arm from my leg to my midriff. I have spent weeks trying to get it to move in a straight line, after many many attempts at moving it in a straight line I finally achieve it, I now have a set of strategies that I go through to achieve this behaviour. Throughout this process I have relegated all of the trial and error process and kept all those meet my outcome, until at such a point the differences between a certain number of moves are ironed out to achieve the one move that meets my criteria.

So what happens to all of the other moves…in one way we could say they are all excellent moves for achieving some other outcome but not the one in question.  So if someone were to model the 'moving of the arm' what is the point in learning other sets of behaviour that are great at achieving another outcome. Surly my outcome as a modeller is to model the outcome I want, that is the movement of the arm to be level to the midriff and not all of the other movements that will lead my arm to a, b, c, and not X.

So in answer to the second part of your statement, the models behaviours 'keen skill' is the successful move, and is not based upon unsuccessful attempts, but successful achievements. You are confusing logical sets between one set and another. The set of achieved outcomes become the refined behaviour, the set of unsuccessful attempts are not inclusive within the set of 'keen skill'. To explicate, set Y = keen skill, set X = unsuccessful attempts. Set X does not = set Y. Set X = all other behaviours that are not part of set Y.

Therefore any modelling should be done within the set of Y and not X.

Unless, a member of the set X falls into the set of Z which could be the larger context of the modelling project, that is the 'what ifs…' or another behaviour. Where both set X and Set Y are a subset of Z.

Whew.. a lot longer than I anticipated. But I hope this gives some insight.

Why model - non-excellence (or another set of behaviours that don't achieve the outcome), if what you want is excellence.

As I was taught in musical college - learn it our way through years of modelling other musicians we know that our way works, when you can do it our way you have the choice to go back to how you learnt it initially or if our way is better, then you can continue to use it. Context: In 'how to hold a plectrum'.

The final part is, once the model has been patterned, the behaviours implemented, the code explicated, the model refined to the least amount of points to achieve the same outcome.. . where else will it work, in what other contexts… can it be cross fertilised into other contexts, areas and disciplines? Now begins the process of generalisation - is the pattern robust to be cross-disciplinary? is it a prime pattern or model?. Where doesn't it work? Etc etc.

For instance, can the pattern (in the context of cricket) be used against a top spin bowler as well as a bottom spin, what adjustments have to be made if any, to accommodate both types of bowlers. And on and on and on…

In answer to question 2) don't ask, do. (you will get much more value out from doing it yourself than in asking)



Entire Thread

TopicDate PostedPosted By
snowballing effect in modeling?12/12/2003 00:33:24scott
     Re:snowballing effect in modeling?12/12/2003 12:10:59Robert Ballentine
          Re:Re:snowballing effect in modeling?13/12/2003 03:52:12scott
               Re:Re:Re:snowballing effect in modeling?13/12/2003 22:35:14nj
          Re:Re:snowballing effect in modeling?30/12/2003 12:57:01ny
               Re:Re:Re:snowballing effect in modeling?30/12/2003 16:23:08Robert Ballentine
                    ReReReRe:Re:Re:Re:snowballing effect in modeling?30/12/2003 18:41:53ny
                         Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?31/12/2003 12:10:06Robert Ballentine
                              Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?31/12/2003 23:38:22ny
                                   Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?01/01/2004 01:03:59John Grinder
                                        Re:Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?01/01/2004 04:29:29ny
                                        Re:Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?02/01/2004 09:27:25ny
                                             Re:Re:Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?02/01/2004 14:53:29Spike
                                             Re:Re:Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?02/01/2004 16:58:15John Grinder
                                                  Re:Re:Re:Re:Re:Re:ReReReRe:Re:Re:Re:snowballing effect in modeling?02/01/2004 17:27:04ny

Forum Home