Prof. Forrester stated that it is not possible to omit soft variables that lack quantitatively measured values. To omit them, their values are being set to zero, the only value known to be wrong. I like this proposition very much.
But I see that such omission of soft variables is possible. Fundamental to the problem with soft variables is their level
of abstraction. Soft variables are concepts of which the level of abstraction is too high to be measurable. To omit them,
I simply make them less abstract, and thus more concrete, by choosing concepts that have lower level of abstraction.
For example, Jay's model in World Dynamics contains a soft variable called pollution, which cannot be physically
measured by any physical instruments. A hard variable for pollution is pollutants. We can omit pollution by conceptualizing it as pollutants instead. Consequently, the concept pollutants has the same level of abstraction as the concept resources and both are objectively measurable variables.
This issue is related to the SD method of judging model validity. It has long been accepted that a model component may be valid if it can be easily recognized or commonly mentioned in published papers. Here, it is important to stress that such a component, if it is soft/even passing that test, is absent in the physical (real) world. Rather, it merely exists in the mental world that can never been objectively measured. Mental world in human head is merely a model of the physical world. An SD model that is consistent with mental models may not precisely represent the considered aspect of the physical world. We must conceptualize it so that it can be measured.
For another example, l tried to build a model of writing quality to see the dynamics pattern of my writing quality. Writing quality WQ is a soft variable. The model consists of five components:
WQ = 5C = Clear + Concise + Correct + Creative + Coherence
All the concepts on the right hand side are also soft variables. We can omit them by using concrete concepts such as number of jargons, number of needless words, and many other concrete concepts that indicate the quality of my writing.
It is possible to omit soft variables. The question is: is it good to do so?
Soft variable omission
soft variable
Hi
The answer is; it depends.
I try as you do relating non measurable soft variable to other measurable data. This is very common not only in SD.
But unfortunately the example you mentionned, does not work. Pollution is not equal to pollutants, because pollution is a stock and pollutants inflence the input rate of pollution.
Even if you know the pollution at a certain time, it will be difficult to calculate later on, because pollutants may be more or less well used, and pollution can decrease with the time and other factors too.
So trying to find equivalent measurable variable is a good idea as long as on can evaluate the relation between the two.
As to Forester's statement, I think it depends on the importance of the soft factor. If the factor is important it is better to consider it. But did Forrester said that one should automatically quantify it?
I do not need anymore to refer to what has been said, because one must always consider the context which is not always clearly identified.
I do not incorporate in my models all soft variables, or I do not incorporate them all together at the same time, because my models would be too complicate and simplicity in models is critical if you want them to be useful.
For instance, I work on investment and pricing.
The effect of pricing depends on a lot of parameters.
I do not try to build a model that delivers the definive pricing policy. It would necessitate incorporating all the factors that influence the policy.
So I build smaller models, and I test the influence of different parameters relative to a reduced purpose (not the definitive pricing policy). It can be for instance, what could be the influence of accepting customers that make long travels
and I try to understand how the model is working and why it givies specific answers hence the necessity to have relatively simple models.
After having made different models, I have a better idea of the real world, and can then build a policy manually, but with the help of the understanding delivered by the models.
Models are for me a way to look at reality that goes hand by hand with other methods, like direct experience for example.
So for me, using or not using soft variables?
No definitive answer.
Regards.
JJ
The answer is; it depends.
I try as you do relating non measurable soft variable to other measurable data. This is very common not only in SD.
But unfortunately the example you mentionned, does not work. Pollution is not equal to pollutants, because pollution is a stock and pollutants inflence the input rate of pollution.
Even if you know the pollution at a certain time, it will be difficult to calculate later on, because pollutants may be more or less well used, and pollution can decrease with the time and other factors too.
So trying to find equivalent measurable variable is a good idea as long as on can evaluate the relation between the two.
As to Forester's statement, I think it depends on the importance of the soft factor. If the factor is important it is better to consider it. But did Forrester said that one should automatically quantify it?
I do not need anymore to refer to what has been said, because one must always consider the context which is not always clearly identified.
I do not incorporate in my models all soft variables, or I do not incorporate them all together at the same time, because my models would be too complicate and simplicity in models is critical if you want them to be useful.
For instance, I work on investment and pricing.
The effect of pricing depends on a lot of parameters.
I do not try to build a model that delivers the definive pricing policy. It would necessitate incorporating all the factors that influence the policy.
So I build smaller models, and I test the influence of different parameters relative to a reduced purpose (not the definitive pricing policy). It can be for instance, what could be the influence of accepting customers that make long travels
and I try to understand how the model is working and why it givies specific answers hence the necessity to have relatively simple models.
After having made different models, I have a better idea of the real world, and can then build a policy manually, but with the help of the understanding delivered by the models.
Models are for me a way to look at reality that goes hand by hand with other methods, like direct experience for example.
So for me, using or not using soft variables?
No definitive answer.
Regards.
JJ
As to pollution and pollutants, I see that pollutants is a stock affecting the pollution generation rate (should be made implicit in the model). However, in reality, no pollution has a tolerance limit. Rather, there is the excessive amount of pollutants that affects the quality of life. I mean we can omit the term pollution by using the term pollutant. Pollutants, not pollution, are compared against environmental quality standards. Furthermore, to be valid, the 'pollution ratio' in the World model should be called 'pollutant ratio' (hazard quotient) instead. Also, pollution generation should be called pollutant generation. People cannot immediately generate pollution, but they generate pollutants. It takes time before pollution occurs.
Hi Monte,
I would be looking the omission examples you mentioned from a different perspective. For example in the writing quality example you mention. Using quantifiable variables instead of the five soft variables may mean two things;
Case 1: Level of clarity is a soft variable. You replace it with a quantifiable variable that you believe can be used as a proxy, as an indicator of clarity. I would not call this behavior "omission of the soft variable". You are still trying to capture the influence of clarity via another measurable factor.
Case 2: You delete the 5 soft variables on right hand side, and plug in some quantifiable/measurable variables that has nothing to do with the first five soft ones. Then, this stands as rejecting your original model and constructing a completely new one for me. And this is clearly the omission case in my opinion; just pushing out anything that you cannot measure directly or indirectly via a hypothesized indicator.
I had the impression that the examples you gave fits more to my case 1. Am I right?
When I listened Jay and read the same statement in various publications of him, my feeling was he was referring to the case 2 by saying "... values being set to zero."
regards,
gonenc
I would be looking the omission examples you mentioned from a different perspective. For example in the writing quality example you mention. Using quantifiable variables instead of the five soft variables may mean two things;
Case 1: Level of clarity is a soft variable. You replace it with a quantifiable variable that you believe can be used as a proxy, as an indicator of clarity. I would not call this behavior "omission of the soft variable". You are still trying to capture the influence of clarity via another measurable factor.
Case 2: You delete the 5 soft variables on right hand side, and plug in some quantifiable/measurable variables that has nothing to do with the first five soft ones. Then, this stands as rejecting your original model and constructing a completely new one for me. And this is clearly the omission case in my opinion; just pushing out anything that you cannot measure directly or indirectly via a hypothesized indicator.
I had the impression that the examples you gave fits more to my case 1. Am I right?
When I listened Jay and read the same statement in various publications of him, my feeling was he was referring to the case 2 by saying "... values being set to zero."
regards,
gonenc
Hi Gonenc,
Yes, you are right. Thank you very much.
I read your note carefully and then consulted BD (p. 854). It seems that I define the term 'soft variables' differently from that did by Jay. To him, soft variables mean measurable variables that still lack measured values.
So, clarity is not a soft variable in Jay's context. To him, a soft variable refers to something like the number of jargons in a paragraph such that its value (say, n jargons) has not yet been counted by anybody.
It appears that what I call 'hard variable' is what Jay call 'soft variable'. So, it is not possible to omit the variable 'number of jargon', or we will assume that the paragraph system contains no jargons--probably the only state that is known to be wrong.
Yes, you are right. Thank you very much.
I read your note carefully and then consulted BD (p. 854). It seems that I define the term 'soft variables' differently from that did by Jay. To him, soft variables mean measurable variables that still lack measured values.
So, clarity is not a soft variable in Jay's context. To him, a soft variable refers to something like the number of jargons in a paragraph such that its value (say, n jargons) has not yet been counted by anybody.
It appears that what I call 'hard variable' is what Jay call 'soft variable'. So, it is not possible to omit the variable 'number of jargon', or we will assume that the paragraph system contains no jargons--probably the only state that is known to be wrong.
I reconsider a point in the first post: "Prof. Forrester stated that it is not possible to omit soft variables that lack quantitatively measured values. To omit them, their values are being set to zero, the only value known to be wrong. I like this proposition very much."
In fact, Prof. Forrester omits one important hypothesis. The value of an omitted variable is not always being set to ZERO, but sometime to ONE. Forrester's proposition needs an auxilairy hypothesis: "the formulation is additive". For example,
a = b + c + d
If d is omitted, then
a = b + c
and
d = 0
His proposition can be fasified when the formulation is multiplicative. For instance,
a = b*c*d
If d is omitted, then
a = b * c
and
d = 1
In fact, Prof. Forrester omits one important hypothesis. The value of an omitted variable is not always being set to ZERO, but sometime to ONE. Forrester's proposition needs an auxilairy hypothesis: "the formulation is additive". For example,
a = b + c + d
If d is omitted, then
a = b + c
and
d = 0
His proposition can be fasified when the formulation is multiplicative. For instance,
a = b*c*d
If d is omitted, then
a = b * c
and
d = 1
I'm still learning. My thoughts below.
Number of useless words etc. does make sense, even though it goes against a recommendation that I read about using positive names. Some positives can be discribed by the absense of a negative. To keep the logic clear, you could say "concise" and then link into it with metrics for "useless jargon" etc.
Level of detail and what is relavent seems to be the question. The statement above seems correct. An assumption of 1 can be misleading, as in assuming that a person's energy level will remain at full force. The above posts provide some clarity..if soft metrics would normally be included in that level of detail, then they should be included. I read something once that said "if a change in a metric will half or double an important factor, then include it". Depends on if it's a factor, how much of a factor, and if it makes sense to include given how much time you have from the group and other factors.
It seems that the introductory questions of "how much detail", "do we include soft variables", "causal loops of stocks and flow" etc. each depend upon the context of the group viewing it, and the problem to which it is applied. Causal loops have some advantages just as brainstorming as advantages. I appears that "what to include" can't be known without having some guess as to the viewing-and-discussing context of the model. A systemic analysis of the model viewing context seems like it would be worthwhile. "Quantity/quality of audience attention" (seniors may be distracted), how much time you have, "risks of open communication" etc. A model aimed at dealing with actual pollutants or chemicals would include different detail than a model with the single purpose of making a case for recycling our paper plates. Each level of detail can take up time or attention from the group, and in corporate environments, a senior interest in some levels of detail can be silly or intrusive. I used to think that soft metrics were great because people respect things that they know exists....but real life is less idealistic. I can imagine scenerios like "headquarters needing to get more info on the staff" for the soft metrics.
Number of useless words etc. does make sense, even though it goes against a recommendation that I read about using positive names. Some positives can be discribed by the absense of a negative. To keep the logic clear, you could say "concise" and then link into it with metrics for "useless jargon" etc.
Level of detail and what is relavent seems to be the question. The statement above seems correct. An assumption of 1 can be misleading, as in assuming that a person's energy level will remain at full force. The above posts provide some clarity..if soft metrics would normally be included in that level of detail, then they should be included. I read something once that said "if a change in a metric will half or double an important factor, then include it". Depends on if it's a factor, how much of a factor, and if it makes sense to include given how much time you have from the group and other factors.
It seems that the introductory questions of "how much detail", "do we include soft variables", "causal loops of stocks and flow" etc. each depend upon the context of the group viewing it, and the problem to which it is applied. Causal loops have some advantages just as brainstorming as advantages. I appears that "what to include" can't be known without having some guess as to the viewing-and-discussing context of the model. A systemic analysis of the model viewing context seems like it would be worthwhile. "Quantity/quality of audience attention" (seniors may be distracted), how much time you have, "risks of open communication" etc. A model aimed at dealing with actual pollutants or chemicals would include different detail than a model with the single purpose of making a case for recycling our paper plates. Each level of detail can take up time or attention from the group, and in corporate environments, a senior interest in some levels of detail can be silly or intrusive. I used to think that soft metrics were great because people respect things that they know exists....but real life is less idealistic. I can imagine scenerios like "headquarters needing to get more info on the staff" for the soft metrics.
soft variable omission
Hi
The problem of taking into account soft variable is not specific to SD.
One should first define what a soft variable is.
For me a soft variable is a variable that is not directly measurable.
Not difficult to measure, but by essence impossible to measure.
For instance “headquarters needing to get more info on the staff”
The problem is then the mixing in a quantitative model of measurable and not measurable data.
One simplistic solution is to ignore all soft variables.
The problem is that in many problems, soft variable are more important than hard ones.
The second solution is to keep the model qualitative.
One can already get a lot of insight with a qualitative diagram and it is often a good
option to stay at that level, at least for a while.
For instance, the headquarters may think about the information is needed about how the staff
work, about the objectives assigned to them and the feed backs they get from the headquarters about their work.
One question to be considered is too the importance of this info from the staff and correlatively the importance of the staff itself. Is the staff operational or not etc.?
It is evident that the question is more qualitative than quantitative at first.
The danger with SD, is the temptation to use powerful mathematical tools that may not
add a real value in some circumstances. It is then best to study first the problem qualitatively and then quantify it with care.
If one wants to have quantitative results from the overall model, the solution is to work
with sensibility analysis, trying to quantify all data, even the non measurable.
This solution will require a lot of work and reflection, but can be too valuable.
If I have a problem evaluating the opportunity to incorporate or not soft data, I first try to
define precisely the problem, what I would like to improve, and the present policies used.
It is important to know where one starts from.
In some cases, when the situation is completely knew, or when there are big changes expected, past policies may be of no interest or course.
I just try to improve policies, or build knew robust ones, without trying to look for ideal solutions, and I try to evaluate the cost of these improvements and the benefit expected.
I think that the objective of SD, is not to find marginal ameliorations that due to the stochastic character of the future, will be useless, but to find important ameliorations, generating substantial benefits without any doubt.
If it becomes necessary to get into too much details or working too hard to find some expected results, to justify SD analysis, there is a great probability that SD is not the right method to be used in that case.
All books dealing with SD, completely avoid the delicate subject of the utility of SD.
Logically when one starts an SD study, one should have done a pre study, making sure that the time and effort needed to accomplish it will generate with a sufficient probability a sufficient result. This characteristic is due to my opinion to the fact that SD is mostly (99% of the public experiences but hopefully less if one considers all the practical experiences generally not made public) used in an academic and research environment where there is generally no identified paying customer and if there is one, it is directly or indirectly a kind of sponsorship.
Regards.
JJ
[Edited on 30-1-2008 by LAUJJL]
The problem of taking into account soft variable is not specific to SD.
One should first define what a soft variable is.
For me a soft variable is a variable that is not directly measurable.
Not difficult to measure, but by essence impossible to measure.
For instance “headquarters needing to get more info on the staff”
The problem is then the mixing in a quantitative model of measurable and not measurable data.
One simplistic solution is to ignore all soft variables.
The problem is that in many problems, soft variable are more important than hard ones.
The second solution is to keep the model qualitative.
One can already get a lot of insight with a qualitative diagram and it is often a good
option to stay at that level, at least for a while.
For instance, the headquarters may think about the information is needed about how the staff
work, about the objectives assigned to them and the feed backs they get from the headquarters about their work.
One question to be considered is too the importance of this info from the staff and correlatively the importance of the staff itself. Is the staff operational or not etc.?
It is evident that the question is more qualitative than quantitative at first.
The danger with SD, is the temptation to use powerful mathematical tools that may not
add a real value in some circumstances. It is then best to study first the problem qualitatively and then quantify it with care.
If one wants to have quantitative results from the overall model, the solution is to work
with sensibility analysis, trying to quantify all data, even the non measurable.
This solution will require a lot of work and reflection, but can be too valuable.
If I have a problem evaluating the opportunity to incorporate or not soft data, I first try to
define precisely the problem, what I would like to improve, and the present policies used.
It is important to know where one starts from.
In some cases, when the situation is completely knew, or when there are big changes expected, past policies may be of no interest or course.
I just try to improve policies, or build knew robust ones, without trying to look for ideal solutions, and I try to evaluate the cost of these improvements and the benefit expected.
I think that the objective of SD, is not to find marginal ameliorations that due to the stochastic character of the future, will be useless, but to find important ameliorations, generating substantial benefits without any doubt.
If it becomes necessary to get into too much details or working too hard to find some expected results, to justify SD analysis, there is a great probability that SD is not the right method to be used in that case.
All books dealing with SD, completely avoid the delicate subject of the utility of SD.
Logically when one starts an SD study, one should have done a pre study, making sure that the time and effort needed to accomplish it will generate with a sufficient probability a sufficient result. This characteristic is due to my opinion to the fact that SD is mostly (99% of the public experiences but hopefully less if one considers all the practical experiences generally not made public) used in an academic and research environment where there is generally no identified paying customer and if there is one, it is directly or indirectly a kind of sponsorship.
Regards.
JJ
[Edited on 30-1-2008 by LAUJJL]
To me, using only measured, hard, variables in a model is equivalent to this sufi tale-
A man saw Mulla Nasrudin searching for something on the ground 'What have you lost Mulla?' he asked.
'My key' said the Mulla
So the man went down on his knees too, and they both looked for it. After a time, the other man asked, 'Where exactly did you drop it?'
'In my own house.'
'Then why are you looking here?'
'Because there is more light here than inside my house'
Regards
Malli
A man saw Mulla Nasrudin searching for something on the ground 'What have you lost Mulla?' he asked.
'My key' said the Mulla
So the man went down on his knees too, and they both looked for it. After a time, the other man asked, 'Where exactly did you drop it?'
'In my own house.'
'Then why are you looking here?'
'Because there is more light here than inside my house'
Regards
Malli
soft variable
Hi
Of course one must take into account soft variables.
But the more there are soft variables in a problem, the more chances that
the interpretation of the problem may vary depending on the point of view of the modeller or the client. These variable interpretations of the problem will generate different structures of models. Logically to do a correct job, one should build different models corresponding to these different point of views, and try to understand why they generate different policies and
how these different policies are related to the different structures and point of views.
I wonder whether modellers will make the additional effort to do it.
There is a difference between what should be done and what is really done.
Hence the necessity to find a balance between practical work and ideal work.
Regards.
JJ
[Edited on 1-2-2008 by LAUJJL]
Of course one must take into account soft variables.
But the more there are soft variables in a problem, the more chances that
the interpretation of the problem may vary depending on the point of view of the modeller or the client. These variable interpretations of the problem will generate different structures of models. Logically to do a correct job, one should build different models corresponding to these different point of views, and try to understand why they generate different policies and
how these different policies are related to the different structures and point of views.
I wonder whether modellers will make the additional effort to do it.
There is a difference between what should be done and what is really done.
Hence the necessity to find a balance between practical work and ideal work.
Regards.
JJ
[Edited on 1-2-2008 by LAUJJL]
Hi,
I agree that we need to find a balance between practical and ideal work. But some experts, as seen from the Review, tend to have a demand for "best system dynamics practice"--ideal work. I prefer a wise (or good) practice to the best one. No one can clearly define what "best practice" is, nor can one tell me what the worst system dynamics practice is.
After buiding our models, I am sure that my modeling practice is neither the best or the worst practice, just like football superstars: Among T. Henry, C. Ronaldo, and so on who is the world best football player? Here we might observe how many goals they have made in this season and how beatiful their playing is. In SD, we might observe how many firms they have been employed as a consultant and how elegant their models are. Maybe you have some better criterea.
MK
I agree that we need to find a balance between practical and ideal work. But some experts, as seen from the Review, tend to have a demand for "best system dynamics practice"--ideal work. I prefer a wise (or good) practice to the best one. No one can clearly define what "best practice" is, nor can one tell me what the worst system dynamics practice is.
After buiding our models, I am sure that my modeling practice is neither the best or the worst practice, just like football superstars: Among T. Henry, C. Ronaldo, and so on who is the world best football player? Here we might observe how many goals they have made in this season and how beatiful their playing is. In SD, we might observe how many firms they have been employed as a consultant and how elegant their models are. Maybe you have some better criterea.
MK
soft variable
Hi
The notion of best system dynamics practice can be differently interpreted.
For me a good dynamic model has nothing to do directly with elegancy, but with utility, even if an elegant model is more susceptible to be useful than an obscure and tricky one.
You can have an elegant model that is totally useless, I have made plenty of them and a much less elegant one useful.
There is no general method accepted by all the SD community to build and analyse models.
The point of views may be very different. But I think it is always useful to know them as to be able to see things from a distance and better choose the method adapted to your present preoccupations and experience of SD.
I have read books, taken courses to find some practical methods and have only found one that treats the whole problem and trains the reader to use it. If you know one I am very glad to know it. But I can still change my mind as the time passes and my experiences accumulate.
I have the general impression that the whole community is not very interested by the practical use of SD. You can see that with the SD mailing list, where there are nearly never any model
exchanged and where the original intention of the SD mailing list is not really respected.
You can read from the SD association web site the objective of the SD mailing list.
The purpose of the System Dynamics Mailing List is to promote discussion around issues in building and using System Dynamics models. Examples of topics that are of interest include:
• Issues in conceptualization of a problem into a feedback structure including identification of reference modes and the creation of dynamics hypotheses.
• Ideas and questions about how to formulate a particular type of structure (hiring policy, price adjustment, fertility determination and so on).
• Interesting models you have built and the results arrived at. Questions to others about examples of models for certain types of problems.
• Validation of System Dynamics models.
• Implementation issues.
And other topics directly related to the development and use of System Dynamics models.
Discussions in the mailing list are generally very philosophical and not practical at all, excepted with questions about examples of models for certain types of problems.
This situation leaves the beginning modeller with no real sound method to rely on and from where he can build experiences. This is to my opinion very detrimental to the field.
It has taken me more than 5 years to find at last a book that exposes a complete and coherent method that satisfies me. It is Geoff Coyle’s book ‘System dynamics modelling a practical approach’.
But nobody can guaranty that five years later I will still find that book valuable.
I found Sterman’s book ‘Business dynamics’ very interesting 6 years ago, and I find now this book uninteresting from a practical point of view!
I think that the way one sees and understands SD, changes with the time.
I see now that field and how one can use it completely differently than I saw it even two years ago.
This is too one of the reason of the lack of practical books, that may be considered interesting by some and not by others.
Maybe if I had started Geoff Coyle book’s six years ago, I might have found it totally uninteresting having not experimented enough with the field.
It is the same with ordinary fiction books.
I remember having read Kafka’s book ‘the transformation’ that writes about somebody being transformed into a cockroach, totally obscure and boring, being 18 years old, and having found
It fantastic 20 years later.
Regards.
JJ.
The notion of best system dynamics practice can be differently interpreted.
For me a good dynamic model has nothing to do directly with elegancy, but with utility, even if an elegant model is more susceptible to be useful than an obscure and tricky one.
You can have an elegant model that is totally useless, I have made plenty of them and a much less elegant one useful.
There is no general method accepted by all the SD community to build and analyse models.
The point of views may be very different. But I think it is always useful to know them as to be able to see things from a distance and better choose the method adapted to your present preoccupations and experience of SD.
I have read books, taken courses to find some practical methods and have only found one that treats the whole problem and trains the reader to use it. If you know one I am very glad to know it. But I can still change my mind as the time passes and my experiences accumulate.
I have the general impression that the whole community is not very interested by the practical use of SD. You can see that with the SD mailing list, where there are nearly never any model
exchanged and where the original intention of the SD mailing list is not really respected.
You can read from the SD association web site the objective of the SD mailing list.
The purpose of the System Dynamics Mailing List is to promote discussion around issues in building and using System Dynamics models. Examples of topics that are of interest include:
• Issues in conceptualization of a problem into a feedback structure including identification of reference modes and the creation of dynamics hypotheses.
• Ideas and questions about how to formulate a particular type of structure (hiring policy, price adjustment, fertility determination and so on).
• Interesting models you have built and the results arrived at. Questions to others about examples of models for certain types of problems.
• Validation of System Dynamics models.
• Implementation issues.
And other topics directly related to the development and use of System Dynamics models.
Discussions in the mailing list are generally very philosophical and not practical at all, excepted with questions about examples of models for certain types of problems.
This situation leaves the beginning modeller with no real sound method to rely on and from where he can build experiences. This is to my opinion very detrimental to the field.
It has taken me more than 5 years to find at last a book that exposes a complete and coherent method that satisfies me. It is Geoff Coyle’s book ‘System dynamics modelling a practical approach’.
But nobody can guaranty that five years later I will still find that book valuable.
I found Sterman’s book ‘Business dynamics’ very interesting 6 years ago, and I find now this book uninteresting from a practical point of view!
I think that the way one sees and understands SD, changes with the time.
I see now that field and how one can use it completely differently than I saw it even two years ago.
This is too one of the reason of the lack of practical books, that may be considered interesting by some and not by others.
Maybe if I had started Geoff Coyle book’s six years ago, I might have found it totally uninteresting having not experimented enough with the field.
It is the same with ordinary fiction books.
I remember having read Kafka’s book ‘the transformation’ that writes about somebody being transformed into a cockroach, totally obscure and boring, being 18 years old, and having found
It fantastic 20 years later.
Regards.
JJ.
Hi JJ,
I know two good SD books: Ford's Modeling the Environment and Forrester's Road Maps. Ford's book is readable and concise, and is the only SD book that gives many examples of environmental application. Road Maps is the best introductory book for beginners, with ten chapters, teaching from basic to advanced topics. Moreover, there are models written in Vensim language, and the model structures are presented equation by equation.
Above all else, at a more advanced level, I think human resources (e.g. you and Bob) are more helpful than books, becuase SD knowledge in books is just a written database, while the knowledge people have is a mental database.
I agree that Bussiness Dynamics is no longer interesting, but this book remains a useful methodological reference, isn't it?
Experience is important in understanding deeply the concepts described in SD books. I usually turn back to read Industrial Dynamics after reading other SD books, to make understanding of Jay's concepts, many of which are too subtle or too brief for the beginner to understand.
MK
I know two good SD books: Ford's Modeling the Environment and Forrester's Road Maps. Ford's book is readable and concise, and is the only SD book that gives many examples of environmental application. Road Maps is the best introductory book for beginners, with ten chapters, teaching from basic to advanced topics. Moreover, there are models written in Vensim language, and the model structures are presented equation by equation.
Above all else, at a more advanced level, I think human resources (e.g. you and Bob) are more helpful than books, becuase SD knowledge in books is just a written database, while the knowledge people have is a mental database.
I agree that Bussiness Dynamics is no longer interesting, but this book remains a useful methodological reference, isn't it?
Experience is important in understanding deeply the concepts described in SD books. I usually turn back to read Industrial Dynamics after reading other SD books, to make understanding of Jay's concepts, many of which are too subtle or too brief for the beginner to understand.
MK
Hi again,
There is a new book by Prof. Coyle, "Practical Strategy: Structured Tools and Techniques," reviewed in SDR 23(4), 475-8. Details of the book is available at http://www.pearsoned.co.uk/Bookshop/det ... escription
MK
There is a new book by Prof. Coyle, "Practical Strategy: Structured Tools and Techniques," reviewed in SDR 23(4), 475-8. Details of the book is available at http://www.pearsoned.co.uk/Bookshop/det ... escription
MK
Sorry, the above link is broken. The correct one is:
http://www.pearsoned.co.uk/Bookshop/det ... 0000031651
http://www.pearsoned.co.uk/Bookshop/det ... 0000031651
LAUJJL said:
>For me a soft variable is a variable that is not directly measurable.
>Not difficult to measure, but by essence impossible to measure.
That's insightful. Thanks.
I found an excellent book regarding "who is the work for" and what type of environment it will be conducted in. The name of the book is "Creative Holism for Managers". It describes several systems theories and the best application for each, the pros and cons, and even had a chart for various types of customers and how each effects the outcome. There are systems theories for simple, complex, coercive, groups etc, theories for creativity, theories for prediction, theories for analysis, theories for mechanics, emancipative system theory" I think was one, etc, and various theories for each situation. It describes the limitations of theories and communication issues. It discusses various metaphors. I don't know how else I would have sorted this out.
>For me a soft variable is a variable that is not directly measurable.
>Not difficult to measure, but by essence impossible to measure.
That's insightful. Thanks.
I found an excellent book regarding "who is the work for" and what type of environment it will be conducted in. The name of the book is "Creative Holism for Managers". It describes several systems theories and the best application for each, the pros and cons, and even had a chart for various types of customers and how each effects the outcome. There are systems theories for simple, complex, coercive, groups etc, theories for creativity, theories for prediction, theories for analysis, theories for mechanics, emancipative system theory" I think was one, etc, and various theories for each situation. It describes the limitations of theories and communication issues. It discusses various metaphors. I don't know how else I would have sorted this out.