Dear Ashwin

I don't think that I understood your query.

You first say that want to compare the best business model and after that you want to compare de criteria

In reality, the first step of AHP does exactly what you are asking, by computing the weighs of criteria and sub criteria, ans without considering alternatives

Dear Prof Nolberto Munier

I am planning to find which supply chain business model is suitable for the glass industry by using AHP model. As the first step, I am creating criteria as Lean and Agile. I am considering 5 sub criteria affecting both the criteria. I want to find which is the best supply chain model among the Lean and Agile. I am planning to construct a AHP with only Goal (Level 1), Criteria (Level 2), Sub Criteria (Level 3) without the alternatives (Level 4). Is it feasible to create one?

Looking forward to hear from you :)

Thanks,

Ashwin

If you're using AHP, you shouldn't expect any approving feedback from dear Nolberto :)

Dear Mahmut

Thank you for your referral, because I like criticism, but unfortunately yours is misleading.

I have been in RG for almost a decade, and many times I have said that for certain type of problems, like trivial and personal, the best method to solve them is AHP, for the simple reason that they are lineally hierarchized, and because benefits and consequences produced by its result will fall on the same person. Is my assertion reasonable? If not, I will be glad to receive your comments.

Regarding complex scenarios I gave many times the reasons by which AHP can't be used in complex scenarios. Simply, they are unable to model that scenario. Am I wrong? I look forward for your answer.

I even wrote a book denouncing 29 drawbacks of the method and the verbatim opinion of 102 researchers, including of course Saaty, Harker and Vargas. I did not receive a single negative comment about the book, quite the opposite, it has, at present, more that 5700 consultations (official data from Springer, my publisher for many years).

If what I said is not true or debatable, why people pay Springer to access that ebook?

In addition, I have said many times that a problem with related criteria, can't be solved by AHP. This opinion is not mine, but from Saaty who established that in writing, and he was right. Is it negative feedback?

Don't expect positive feedbacks from me when I read papers that use AHP only because it is a well-known method, with no need to think or investigate, and where a formula COMMANDS the DM to correct its own estimates, and still worse, assuming that the DM estimates and transitivity can be transferred to the real-world. Have you thought about this?

I also have had negative comments for other MCDM methods like SWARA, B&W, COMET, MOORA, etc. and even on ELECTRE, PROMETHEE and TOPSIS, for me the best methods in the MCDM inventory, because they are rational, demand analysis from the DM and are mathematically supported, all qualities absent in AHP.

In all honesty, I believe that AHP was a very good method in the 50s., because when Saaty invented AHP, problems were lineal, but about the 60s. their structure changed, and the lineal hierarchy no longer applied. Why do you think that Saaty created ANP? Because the matrix structure represents more faithfully actual problems.

If according to you, I am against AHP because my negative comments on the method, please explain me, how, along 8 or 9 years in RG, NOBODY, you included, did not refute what I said. Criticism is healthy, but it needs to be documented; words are not enough.

I would very much like to receive that feedback and discuss it openly.

As you probably know, I am reviewing some papers submitted by RG practitioners, not all, but many of them, applying AHP. If you think that my feedback is negative, I suggest reading some of them.

Looking forward to your answer

Nolberto Munier

Dear Nolberto, I know your sensitivity about AHP. I don't know much about AHP, but I would definitely criticize you if I did. Like many others, I've read your dozens of negative feedback about AHP over the years. I just made a witty guess based on data like ML algorithms. There is already a smiley sign at the end of my sentence. That is all!

Dear Mahmut

Thank you for your answer.

Sorry, but I don't use signs or icons, and then I don't have idea of what they mean, and for this reason I misinterpret you.

I know that you don’t like to discuss AHP, and of course, I respect that, I only reacted because your brief comment.

I am not an expert on AHP either, but I use common sense, and my only purpose is to make people aware that this method deceives students, professors and researchers.

As I said many times it makes people believe that they can solve complex problem without reasoning and research, and using only intuition

I am surprised that you a conscious and prolific researcher in MCDM, is not interested in analyzing the most used MCDM method.

Dear Ashwin

Lean manufacturing aims at improving production processes and efficiency.

Agile is a collaborative effort and cross functional teams with customers and users

Perhaps I am mistaken, because I don't kmow your problem in depth, but it appears that there exists in close rellationship among criteria, and if it is so, you can't use AHP

And if you don't have the alternatives, how can you desing criteria, since these are a function of the first

Dear Nolberto,

First of all thank you for your nice words and insightful observations. I can say that I am confused about AHP. Since I don't have a very deep insight into this method, I prefer to stand back a bit. But I wanted to share my personal comments in order not to be silent on your call.

First of all, AHP is a subjective method. AHP is a ranking/prioritization method for both criteria and alternatives, which is not a function we come across often in other MCDM methods. I would like to talk about the criterion weighting dimension of AHP. As you know, AHP is basically based on expert opinions. It starts with the linguistic terms we see in fuzzy and survey applications and the scoring of linguistic terms. With the help of expert opinions it assigns points to criteria with a scale. As in BWM and SWARA, the point that determines the coefficient of criterion importance is the experts. Here, in my opinion, there is both a problem and a solution. That is, if the experts are real experts with insights, there is no problem. But it seems difficult to predict the usefulness of AHP computation through experts whose expertise and know-how are disputed. This is the controversial aspect of the business. The other dimension is group decision making dimension. If real experts come together to discuss and reach a solution, it will be a good example of collective wisdom. And this increases the performance of AHP.

Natural language processing models are already a fruit of the collective mind, in my opinion. You may underestimate individual human intuitions, but I think you should not underestimate the accumulations of the collective mind. Today, AI and ML technology is a synthesis of objective big data and subjective big data. I recommend using and testing AI applications related to natural language processing models. Try to understand how he does objective syntheses from subjective data. You will see that collective human intuitions can actually come to an objective conclusion when combined. By discovering patterns, insights, patterns, relationships, and clues in seemingly subjective data, we learn tremendously useful information. The AHP methodology will break its shell and give more confidence if it can leverage AI technology in the future. The robustness of the AHP in the current situation will depend on the competence of the experts. For AHP, I think we should discuss the mandate of expertise rather than Methodology. I think methods like AHP should be more concerned with selecting experts rather than criteria and choosing alternatives. In fact, when you choose the expert correctly, your hit rate increases, but if you choose the wrong one, you will encounter inaccuracy. Alternatively, my expert advice for AHP is ultimately AI applications.

MB- First of all thank you for your nice words and insightful observations. I can say that I am confused about AHP. Since I don't have a very deep insight into this method, I prefer to stand back a bit. But I wanted to share my personal comments in order not to be silent on your call.

NM- You are very honest in saying that you are confused about AHP, and you are confused because you are conscious, analyze and research, and think, before using a method, whatever it might be. Many people use AHP because it is known, does not demand any effort and even correct the user, and due to the advertisement produced by the so called ISAHPs.

MB- First of all, AHP is a subjective method. AHP is a ranking/prioritization method for both criteria and alternatives, which is not a function we come across often in other MCDM methods.

NM – True, but that is a task for all MCDM method, it is not an advantage of AHP.

MB- I would like to talk about the criterion weighting dimension of AHP. As you know, AHP is basically based on expert opinions.

NM- Do you consider that using pair-wise comparisons and assigning arbitrarily values of preferences is an expert opinion?

MB- It starts with the linguistic terms we see in fuzzy and survey applications and the scoring of linguistic terms. With the help of expert opinions, it assigns points to criteria with a scale.

NM - May I remind you that AHP is already fuzzy, as Saaty clearly stated?

MB- As in BWM and SWARA, the point that determines the coefficient of criterion importance is the experts. Here, in my opinion, there is both a problem and a solution.

NM- BMW and SWARA are also based on opinions. So, an expert decides considering a list of criteria that normally involves economics, financial, environment, engineering, which is the most important and assigns a coefficient? Based on what, for both, the selection of the criteria and the coefficient?

True, experts may be a problem and a solution. Trouble is that personal opinions are a problem and not a solution even if it becomes from experts.

Who guarantees that another expert thinks the same?

However, experts are welcome if they think, analyse, investigate, ask questions, etc., which allow them to have a clear panorama, which is very effective, mandatory and extremely useful when they work with documented data. Of course, there are many subjective questions, and to answer them are surveys, statistics, experience, etc. It is here when the expert is unvaluable, not inventing data or give solutions based on them.

MB- That is, if the experts are real experts with insights, there is no problem. But it seems difficult to predict the usefulness of AHP computation through experts whose expertise and know-how are disputed.

NM -Agreed in a 100%. However, the paradox is that you can have very qualified experts but it does not mean that they can reach a solution working together. Why?

Because an expert in engineering for instance, can’t judge a matter related with finances and vice versa. As an example, when two experts must decide if criterion C5 is more or less important that criterion C11, when the first relates to geology and the second to health for instance, how two experts, a geologist and a doctor can opine when they are respectively addressing aspects that is completely unknown for them?

That is, how can the geologist judge about medical equipment or a doctor discuss about the exploitations of lithium mines? In a complex problem that has say ten different fields we would need ten different specialists, and even then, they will be talking different languages, like a Babel tower.

I have expressed this many times and nobody came to refute it or at least give an explanation.

MB- This is the controversial aspect of the business. The other dimension is group decision making dimension. If real experts come together to discuss and reach a solution, it will be a good example of collective wisdom. And this increases the performance of AHP.

NM -Well, this is what I answered in the above point. Please, tell me where is the failure in my reasoning. Twenty years ago, I was the moderator of a group of experts on defining for instance environmental indicators, at national level in Canada, but the three of them were experts in that field, and each one gave a different valuation, that needed to be consolidated, and I did it scientifically, not using my opinion. If you are interested, I can give you full details of this assignment, which was also published.

MB- Natural language processing models are already a fruit of the collective mind, in my opinion. You may underestimate individual human intuitions, but I think you should not underestimate the accumulations of the collective mind.

NM- Sincerely, I don’t see any relation between Natural language processing models (NPL), based in AI, and the fact we are discussing.

I don’t underestimate individual human intuitions when they refer to one person, but I don’t consider that they are applicable to a bunch of persons.

Imagine that you are to board a plane and you have a bad intuition about an accident. Why do you think that all pax must share it? Have you heard that that intuition is collective and a flight was cancelled for that?

MB- Today, AI and ML technology is a synthesis of objective big data and subjective big data. I recommend using and testing AI applications related to natural language processing models.

Try to understand how he does objective syntheses from subjective data. You will see that collective human intuitions can actually come to an objective conclusion when combined.

NM- I am not a psychologist, then, perhaps you can explain this point to me.

Please, don’t use a bunch of words that say nothing. Put examples and reasoning and I will believe you.

MB- By discovering patterns, insights, patterns, relationships, and clues in seemingly subjective data, we learn tremendously useful information.

NM- Yes, I know that, it is the basis for AI and ChatCPT

MB- The AHP methodology will break its shell and give more confidence if it can leverage AI technology in the future.

NM- Maybe, but I doubt it. AHP is a non-sense method, and you know that I can easily put many examples justifying what I say, something that I always do. Many people, based on my many comments against AHP can charge me with ignorance, misinformation, being erroneous, etc., why don’t they do that?

Because even when I am not the owner of the truth and can be mistaken as anyone else, they don’t have valid arguments, and recognize that what I say is easily verifiable.

MB- The robustness of the AHP in the current situation will depend on the competence of the experts.

For AHP, I think we should discuss the mandate of expertise rather than Methodology.

NM- Mandate of expertise? And what is that?

MB- I think methods like AHP should be more concerned with selecting experts rather than criteria and choosing alternatives.

NM -In my opinion, you are mistaken. Selecting criteria is one of the most important tasks of the DM, and alternatives are given, not chosen by the DM

MB- In fact, when you choose the expert correctly, your hit rate increases, but if you choose the wrong one, you will encounter inaccuracy.

NM- And how do you now that an expert (in AHP?) is correct? Where is the yardstick to measure it? Maybe, it can work in your field, because you can compare with actual results.

As a bottom line, what is your proposal? We need reasoning people, that investigate, that think, not people filling a matrix with data coming from intuitions, and waste a lot of time arguing about consistency, or how fuzzy can solved problems, with invented data, and using method that was great at its beginnings, following a military structure, but became completely obsolete with new organizational structures.

You don’t need to have a very deep knowledge of AHP. You don’t need formulas, analyze the method using common sense, reasoning, analyzing each step, and then, reach a conclusion. Look for rationality.

MB- Alternatively, my expert advice for AHP is ultimately AI applications.

NM -Fine, based on what grounds?