Pierre Simon LaPlace and the BPM Demon

BPM Warning Sign

I suggest that BPM ought to come with a large, bright-yellow warning sticker. Let me lead you through my thinking that started with a number of posts on the subject of Adaptive Case Management versus Business Process Management.

First there is Keith Swensons interesting exchange with Jean-Jacques Dubray on ‘Do Unpredictable Processes Exist?’ Be warned, as it is a long post, but quite worthwhile. Then there is Clay Shirky’s ‘The Collapse of Complex Business Models’, also referenced in a post of similar name by Jacob Ukelson.  Please read them if you have the time, but I will try to present and discuss the viewpoints here.

What can we know about complex adaptive systems?

To avoid long discussions like the one between Keith and Dubray, we could start with some epistemology — the theory of knowledge — right here, to discuss what the terms ‘complex‘ or ‘unpredictable‘ represent in the following. It is not just simple term definition, because the subject is: What CAN we know how? ‘Complex’ is considered a synonym for ‘complicated‘ and that already creates misinterpretations. I have used — and so have others — the term ‘complex’ for a working system that cannot be decomposed into parts, in difference to a complicated one. We can understand causality in a complicated or deterministic system, but we can’t analyse and predict it in complex or social systems. Complex social systems can be adaptive, meaning that they exhibit for example self-organization in hierarchies.

That is in fact the essence of the communication between Keith Swenson and Jean-Jacques Dubray. I read the whole conversation several times, because I do agree with Dubray largely about entities and their states and still end up disagreeing with him. The problem is about getting actionable knowledge from data interpretation (as per information theory). Dubray seems to believe that we can know enough to see the world as decomposable — ‘The entire universe has well defined states.’ he says. — into bits and pieces. Therefore, he feels, it must be possible to clearly define all future combination of such entities and their states and create a working map that will cover all possibilities. That is a very common, pseudo-scientific delusion. The great mathematician Pierre-Simon Laplace fell prey to the same in 1814 when he pondered about a demon, who would be able to calculate the future of the universe if he could access all the information about its present state. Laplace’s demon was based on the premise of classical, time-reversable mechanics, whereas the thermodynamics involved in all real processes are currently thought to be irreversible. The similarity to human processes is not farfetched, because not all state-changes can be perfectly undone. Not all actions have definite consequences. We are unable to identify ALL relationships and dependencies. And there can be no certain analysis of WHY! Taking a British car to pieces does not tell you WHY it’s steering wheel seems on the wrong side. Therefore there can be no prediction either. Model thinking is always inaccurate.

The entire universe has well-defined states?

To my amazement, the subject of events never came up in the Swenson-Dubray conversation. Unpredictability is however strongly connected to events. Dubray only describes processes in manufacturing terms, but not as customer service activities where the outcome is a very broad ‘happy’ customer perception, while Keith likes to use the ‘healthy’ patient. While these are understandable human states they are not clearly definable and NOT measurable. There aren’t any entity state changes that will ensure a ‘happy’ state in a customer. Healthy can mean very different states too. Goals are in fact not states but rules that set a state. In a complex system these rules map to unknown patterns of states in the state space. There can be many patterns related to a certain state of one entity and even when observed the information is correlation but not causation. (This is the area of my transductive training agent patent so trust me on this.)

While I am in full agreement with Dubray on the problems of BPM modeling if it ignores entities and their states, his main fallacy is the belief that the impact and effect of relationships in complex systems can be causally determined. The most important element of business processes are people and not things. If the people are removed and their actions fully automated then yes, processes can be fully predicted. They just do not serve a human purpose and will not fulfil human goals. Dubray even suggests that we can predict the weather, when there is no such thing. At best we can calculate some rough probability of a certain weather pattern being more prominent than another at a very short time in the future. If Dubray says that processes can only be predicted as far into the future as the weather then I agree with him. If however, BPM outcomes are as accurate as weather forecasting then don’t expect it to improve your Key Performance Indicators for the yearend!

A different view: Clay Shirky travels quite some distance to make his case for the potential catastrophic consequences of complex business models for our society. Like me, Shirky points to bureaucracy as the culprit.  Jared Diamond has also suggested similar issues in his book ‘Collapse‘. I see this as the normal progression of the world and believe that not too much can be done about it, without being a fatalistic person. I see this as the mechanisms of evolution. We can however understand the mechanisms of evolution in areas where we have more influence than on the scale of whole societies. While the wish for control is a human trait we don’t gain when we control too much. The strongest oak will eventually be brought down by some storm, while the tiny reed will simply bend. Buildings that are too brittle tumble in the slightest earthquake. Organizations that are rigidly controlled lack resilience to outside change. I see bureucracy and methodology needed to get BPM to work as such brittle, rigid structures themselves and they create brittle, rigid processes and if fully executed a brittle, rigid business. Jacob Ukelson thus tries to defend Adaptive Case Management as an independent domain of functionality from BPM. I agree but don’t see much of a chance, because BPM proponents will continue to claim that it is just a sideline. For them BPM bureaucracy creates the safe and predictable routes on the business maps that will ensure that business activities will reach their goal. Makes it all perfectly manageable and predictable, right? No warning stickers needed.

Navigating the oceans of business on a virtual map.

It reminds me about the guy who ran his brand new 80ft yacht into a rock just outside the marina. I talked to him while the boat was being lifted onto land in the shipyard with a gaping hole in the bow. He was lucky that it had floated. He was furious, threatening to sue the chartplotter and electronic map manufacturers. He claimed that he was at least 50ft from the reef on his navigation display. ‘How could I know that these guys are too stupid to build an accurate system?’ he said. I mentioned the disclaimer that is on each paper and electronic map and GPS about their lack of accuracy. ‘There is a disclaimer on everything today,’ he said. ‘They are just for idiots.’ This idiot had 8 guests on board when this happened so I left without another word.

Yes, I see the process maps produced by BPM just like a planned route on a naval map. They are very useful and I would not go to sea without a chartplotter today. Actually the one on my iPhone is pretty good. Would I want to captain a boat that could only follow predetermined electronic routes that I could not change anytime I want? Certainly not. Without constant lookout I would have nearly hit a school of whales once. Would you want to be a guest on a yacht where the captain can only follow the predefined routes on an inaccurate map, by means of inaccurate GPS, believing that the weather can be perfectly predicted? I don’t think so, but despite that obvious reality that the economy and business is as dynamic and unpredictable as the high seas, thats what everyone wants to do with BPM.

A business (unit, process team) should be run like an offshore yacht with electronics that tell me where I am accurately enough that I can use my skills to navigate safely. I should be able to navigate ‘the old way’ to know what to do when the electronics end up being fried by lightning or give really weird advice. I also need a means to get frequent weather reports because every mariner can tell you that predictions of more than 24 hours are insufficient. In some locations you need 3-6 hours to be safe. I also need a means to communicate and a radar to get a real-time view in bad visibility.

Therefore, that’s what Adaptive Case Management gets the process owner for his team: Real-time, approximate location, information about invisible stuff, communications, estimates on upcoming conditions, the ability to plot new routes and waypoints (and save them for the future) and calculate ETAs. He depends on the electronic stuff (ACM) despite all the warnings but uses his brain — as well as a solid boat, a good crew and lots of experience — to get to the agreed upon target.

Why is there no disclaimer sticker about inaccuracy on each monitor that runs BPM? Ahh, I nearly forgot. NO NEED! BPM can only drive the predetermined route and neither captain nor crew know what they are doing.

Lifejackets anyone?

17 Comments on “Pierre Simon LaPlace and the BPM Demon

  1. Max

    Very good analogy with the ship. Next to traveller and general IT nerd I have been a sailor in a profesional way. It is so very true what you say. With all the best tools most of the time you get where you need to be, but if something goes wrong you need people and teamwork to solve it.
    “Wrong” is actually the wrong word, our ever changing universe brings us ever unexpected events.
    All is relative and therefor can theoretically be calculated even the weather. But then again, theoretically, Achilles can never overtake the tortoise, Zeno’s paradox.

    Cheers,

    Erwin

    Like

    • Erwin, thanks. Right and wrong is only related to human expectations but has no natural absolute.

      Let me add that BECAUSE everything is relative it cannot be predicted, not even theoretically. We cannot discover all possible relationships and their causal influences as everything is connected to everything. If all things were absolute then theoretically it would be possible would the necessary data-bandwidth and computing power be available in this universe. As the number of bits would be larger than the estimated number of atoms in the universe that seems unlikely too.

      I too was fairly strongly influenced by Hofstedter, but Zeno’s Paradox is not really a paradox. It shows two fallacies: 1) If you do stop after each step you have to consider a cutoff point of either control (ability to take a small enough step) or measurement (not being able to halve the next step, latest at a Planck length). It seems a perfect logical model but it is not connected to the real world — actually as ALL logic. 2) As we take steps our body moves continuously and therefore Achilles is much faster than the tortoise in reality.

      BTW, been doing mostly powerboating but lately I went sailing a lot (Pang-Na Bay, Belize Barrier Reef, Los Roques).

      All the best, Max

      Like

  2. Max:

    thank you for your detailed response. I don’t necessarily disagree with you. I am not delusional (you may read some complements here: http://www.ebpml.org/blog2/index.php/2010/03/30/bpm-discussion-with-keith-swenson about the relationship between irreversibility and time).

    In the end we have to find a process-centric programming model or decide once and for all that there is no process-centric programming model and we can all happily CRUD and occasionally use MVC until the end of times (or IT, whichever comes first).

    As you pointed out there are many more dimensions to BPM than just the programming model and it is true that I focus mostly on that question, simply because I find it foundational.

    Like

    • Jean-Jacques, thank you for the comment. As I said, there are so many things that I agree with in your perspective. I obviously did not say that you are delusional in general, but that functional decomposition is a common view that takes deterministic system thinking into complex systems. Your ‘bold’ claim that ‘all states and transitions can potentially be known’ definitely conflicts with current scientific understanding. Even if we decompose it to the lowest level, Heisenberg’s Uncertainty principle prohibits such endevaour. You do however amend a point later that the complete path of states and transition cannot be known, and I apologize that I have to point out the lack of logic in combining those statements. Enough of that.
      We apparently do agree that business processes do NOT exist for different reasons. They are simply a part of our timeline-sequence-memory illusion and do not exist in reality. You do point to the underlying system of interdependent states that interact in complex patterns and I need to point out that the entity relationship mechanims are event driven and MUST be considered as part of the system. These are not predictable and yes, therefore I propose to focus on people and outcomes and not on the process flows. I do not see BPMN as the holy grail, but of all the process management definitions is has the most flexibility and thus allows to do both rigid and adaptive processes.
      About the state of the IT industry I wholeheartedly agree with you! Most IT concepts being sold lack the slightest plausbile proof, but are all built on anectotal evidence being sold with billions of marketing dollars. After 20 years of BPM there is not a single long-term, vendor independent study about its benefits for a business. The same is true about predictive analytics and some other snake-oil being sold.
      I developed the Papyrus Platform with its object-relational, state/event driven entity model over ten years ago and I feel that it is completely in-line with your perspective. I see the flowchart and sequential programming paradigm as the greatest drawback for modeling and therefore acting in the real world. That is the TRUE DELUSION! Thanks again.

      Like

  3. Max:

    I think you are misunderstanding my proposition. I grew up staying Prigogine so I think I understand a thing or two about irreversibility and states. Now we, as human, keep understanding the states of the universe. Recent advanced with the LHC or string theory are exemplary with that regard. The advanced in biology are even more specetacular. Again it is not because you know the states that you will be able to prdict the path the u iverse will take. I understand thAt very well.

    The fundamental question is can activity be perform that violate or does not rely on a given set of set and transitions. IMHO, this is know because you don’t know some state types and state instances that this principle is violated. If it is indeed a principle that governs BPM, how come it is not reflected anywhere?

    I don’t think there is any evidence today that would suggest this principle is wrong nor should it be ignored. I’d be happy to be proven wrong, but in this particular instance I believe I have the entire universe supporting my argument. For some reasons humans love to ignore the universe they live in, let alone trash it.

    JJ

    Like

    • JJ, thanks again. Yes, I must be misunderstanding what you mean by knowing about states. I also fail to understand which advances in terms of String Theory (that is heavily disputed), biology or what the — only recently working — LHC provides, you are referring to. All our knowledge about states and all measurements in physics (as in the LHC) are purely probability based. The wavefunction in quantum mechanics is a function of space that maps the possible states of the system into probability amplitudes — elements in a complex Hilbert space — as the squares of the absolute values that give the probability distribution that the system will be in any of the possible states; either a complex vector with finite or infinite number of components or a complex function of one or more real variables. For systems with multiple particles, the underlying space represents the possible configurations of all the particles and the wave function describes the PROBABILITIES of those configurations. Which states are you referring that we can definitely identify?

      I am a believer in David Bohm’s approach that we look at space topology the wrong way because of General Relativity. I believe there are real states, but they are a complex interdependent with all other states so we can’t identify them individually. Because we try to look at single particles, that’s why we only get probabilities. The relationship creates the BE-ABLE as Bohm called it.

      You can’t claim that you have the whole universe as proof without saying what elements and models you use as such proof. We can however leave it at that, because clearly, if you go for String Theory and I go for Quantum Field Theory we will have a hard time to discuss BPM.

      I already said, that we do agree on BPM in terms of a flow-paradigm being too limited to model the real world. So BPM does currently neither represent yours or my thinking. No matter how much time we spend to infer models from analysis we are limited by the statistical significance of our observations (as per Shannon) and thus we can reduce our uncertainty but we cannot be certain.

      If you now create some model and force people to only work inside the model theory then yes, they may not be able to violate state transitions you did not foresee but clearly that will not perfectly represent the real world. The idea of adaptive processes is now that we can add/amend/change those models as we learn about them without having to go back to the model drawing board. We can add/improve goal rules and react to customer peceived states we did not think of. So do not get hung up on BPMN, as it is exactly that what we ACM apostels are trying to change.

      I am not sure why you throw ‘humanity trashing the universe’ at us, when we are not that far from your perspective. If I understand, the disagreement was about whether the world does exhibit unpredictable behavior or not. You said yourself, that even knowing all possible states does not allow us to know the path some process will exhibit. That is exactly the ACM concept we are proposing, and yes, it can’t be mapped with a predefined BPMN flowchart.

      Are we in agreement now, allowing for a different view on ‘knowable states’?

      Like

  4. Max, can you please clarify – what BPM concept do you address in the title of this post – a discipline, a tool, a practice (a process-based system implemented within an organisation) or all together?

    Thanks,
    AS

    Like

    • Thanks for the comment. I am referring to the general practice of BPM to create rigid, interlinked processes, because they are considered more efficient and produce a more consistent result. I propose that this does not allow an organization to truly focus on needs and goals, which can in my mind not be expressed in simple KPIs. No BPM model can accurately represent what is needed. Obviously, the worst is when those processes are hardcoded into a BPMS and then people of lesser knowledge at lower cost can be used and the BPM effort is justified with that. It is my experience that organizations suffer with orthodox BPM due to a lack of intrinsic motivation and that the external improvement and innovation cycles are lacking to non-functioning due to bureaucracy that mostly focuses on creating the expected numbers. Agility is a human capability, not software and not methodology. If software is employed it has to enable human, in-process agility. Process maturity is defined in how much the actors are enabled to control the process and not how well the process controls them. The longterm business benefit comes from transparency and customer perceived quality and not from quick-fix, cost cutting. If an organization uses BPM to define all the process owners and give them the authority to execute then this would not cause the mentioned problems. Hope this clarifies my position.

      Like

  5. Thanks for speedy and clear reply. I fully agree with you that BPM (as a discipline) is often misused. I saw BPM-based implementations which are flexible, empower the process owners and help companies to move forward faster (of course, this is relative to the traditional application development). As the result, I can say that problems you mentioned are not with BPM (as a discipline), but with _architecting_ BPM-based solutions. Would this be a correct conclusion?

    Thanks,
    AS

    Like

    • Thanks again. There is no definition of good and bad BPM, even though I have tried and so do you. It depends a lot on the qualification of executives. The common approach of BPM is to put a lot of ‘methodology’ — which is another name for bureaucracy — in place. Process centers of excellence are such a fallacy. Also those could be very effective if they work with training and coaching but in most cases they represent the process police. Especially where BPM is combined with SixSigma and numbers become the dogma. If one focuses on numbers, thats what one gets. No more.

      Having read around 20 standard books on BPM, I have to say that my problem is with the generally suggested BPM approach and not just with some implementations. Process orientation makes sense, process management does not. Yes, there is the rare implementation that does what you and I suggest. The amount of money that is claimed as being saved by BPM is achieved by a hidden reduction in service quality, skilled workforce and in people agility. There are no numbers being put on these intangibles and therefore BPM is the big moneysaver in the eyes of the quartely focused executive.

      Like

  6. Max,

    I understand and support your point fully, there is no right or wrong, in computer terms right and wrong could be described as the state of a bit I or O. In QF we could say the opposites matter and anti-matter.
    Again it only lseems there is logic in this and therefor causality, but then we indeed only talk about the states of a entity not about their relations. Indeed because everything is relative it cannot be predicted.

    If we look into the universe it could be that actually NOT everything IS related, which makes it even more unpredictable. I am refering to the fact that there is not enough anti-matter to support al matter, as if there is not enough Yin to support Yang.

    If CERN is able to prove that there is not necesarilly a balance, and there is a certain chaos factor then we are stuck with the fact that some entities might or might not have a relation making predictability even harder.

    Proving the point more over that processes to a certain degree can be predictable enough for us to work with as a human but are in essence not predictable and therefor are hard to put in a model.

    So, we cannot predict the weather, just more or less have a general idea of what it will do the comming few days, still I need to offer our customers practical solutions on how to manage those unxpected events, so they ask me to build applications for them that can give them a good weather forecast. The only awnser I can give them really is that I will go with them wherever they feel the need to go and carry their bag, in it? A good warm pull-over, an umbrella, and the knowledge of the fact that it not smart to shelter under oak trees in thunderstorms.

    All the best,

    Erwin

    Like

    • Erwin, great comments, thanks! If there is absolutely no relationship between two entities then they in each others world they don’t exist. The question is: By what means would they suddenly enter each others world? Divine intervention? Things that we say to each other influence a lot of people we don’t know through the other as an intermediary. So there are maybe indirect relationships that from the perspective of the actor are completely unknown and can’t be predicted. Matter and antimatter are not opposites, they bot represent energy. But yes, as all energy in this universe needs to cancel out to ZERO, as otherwise where would it have come from at the start of it, we need to try and find the opposite. I propose that we find it hard to identify the sign of energy because typically appear at the same time. I propose that the energy of mass is balanced by the energy of gravity and the energy of time and distance by the zero-point energy of space, and so on.

      Models of complex adaptive systems are for understanding but not for control. The control that we have is real-time only. Predictive analytics are complete nonsense, because we look at a statistical number and then try to use it to predict a single event. That is truly ignorant. If you use it define how much stock you ought to buy for the Christmas season then it can be useful if the final decision remains a human intuitive one.

      I agree that we need to give people better models and more real-time information and more real-time control. BPM is not doing that and ACM is just a first step into the right direction. Regards, Max

      Like

  7. Max,

    I like the analogy with sailing. What came to my mind reading it, was the need to train your skills for the event of failure of any of these devices. If you never learned how to sail without these devices, how can you managed to return safely if they break? Or even if you learned it but didn’t had to use your knowledge for 20 years, would you be able to managed such a situation?

    Even worse, if one of the devices does not fully fail, but show some slightly inaccurate data, it would need a lot of experience to recognize.

    My point is, transfering the analogy back to the business world, how can one expect a person to react in an accurate way when an unexpected event arises, if this person just followed strict and predefined paths through business processes?

    Don’t we need daily training to get experience on how to handle the unexpected? If we would have to handle the unexpected daily, wouldn’t this train our “brain-muscle”, like a hiker whose ankle muscles are well trained and don’t get hurt on any stone he is stepping on by mistake?

    I think when we try to put people in a role where they just react on predefined ways unsing predefined behaviour, this would decrease the overall performance of an organisation, because organisation will be confronted with the unexpected on a regular basis.

    I believe humans are willing to do their best, we need only to give them possiblity to do so and select the ones for our organisations, that have the same goals, values, and interests. Restricting them with the tight ropes of BPM will frustrate them. Frustrated persons only work for their income, not for reaching goals.

    Cheers,
    René

    Like

    • René, thanks. Good points. I have also stated before that BPM people-control and KPI based payments do not increase people motivation. There are a number of scientific studies that show that extrinsic motivation with money is not very effective. Additionally your KPIs may not be producing the effect you were hoping for. Intrinsic motivation, meaning that people enjoy what they do and do it for the joy or reward of being a well-accpeted member of a team, is a lot more effective in terms of outcomes. The skills and interests of a large enough group of people will always distribute according to the Gauss bell curve. That is the reason for the adage that 20% of your people produce 80% of revenue. But we need the rest also. We can only increase the steepness of the curve but we can not cutoff the lesser qualified people (quality) or the higher qualified (cost) ones. But with the right tools, both can work together towards a common goal and each do their share. Regards, Max.

      Like

  8. Max:

    I am not quite sure why you oppose a probabilistic view of the world with the notion of states. The two are fully compatible. This is my definition of unpredictability. We can identify the possible states with a high precision, yet, we can only come up with the probability of a system reaching a particular state. The path to a given (complex) state is what is unpredictable, not the states themselves. The same thing happens in BPM. Most companies define or identify states with a very high precision, yet the path taken by a particular “process” is often “unpredictable”. A “process definition” therefore, does not exist. It is only the representation of the most common path, the cheapest path,… only the simplest resource lifecycles could lead to a “complete” process definition which is hardly ever the case, most often in school books.

    Like

    • JJ, we have a misunderstanding. I do not oppose a probabilistic view, rather the opposite. I keep saying that we ONLY have a probabilistic view and therefore it is impossible to identify A PRIORI all the possibly states with a high precision. Therefore the system could reach states that we have not modeled. Therefore both possibly states and summary states are unpredictable. We not only lack certainty about possible states but also certainty about the causes for state changes.

      Yes, people do model states (not sure how you model with high-precision) but that does not mean that they are an accurate model of reality. For a desirable final ‘outcome’ summary state you also have to define what your goals are. Because efficiency and effectiveness (reaching a desired goal) are always influencing each other.

      As I said before we are not really in disagreement, but you assume states can be accurately defined and I don’t. As we agree that the process of state changes can not be predicted, the consequence is the same. Or at least thats how it seems to me right now. Agreed?

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: