We use cookies and other similar technologies (Cookies) to enhance your experience and to provide you with relevant content and ads. By using our website, you are agreeing to the use of Cookies. You can change your settings at any time. Cookie Policy.
Start a Discussion

Do we need a new modeling language for more adaptive processes?

Vote 0 Votes
A question brought up in this blog: Do we need a new modeling language for more adaptive processes?

15 Replies

| Add a Reply
  • No. Last thing we need is yet another language to learn. BPMN hasn't even reached acceptance yet, why bother with another from scratch, or a variation.

    The nature of adaptive processes means that tightly prescriptive modeling is of little to no use.

    Don't waste the effort unless you're just looking to sell another book.

  • I have never been a fan of modeling languages. When you finally have a process modeled and you put it into practice, adding the human element changes everything.

    My experience has been to work with the people in the process to come up with the new process and then use it. You will find ways to improve it as you use it.

  • This question is making my synapses jumping form left to right. Do you need a map for a road you don't know yet?

    Every process has a result, but if it must be adaptive, what is the use of making a model?

    Adaptive to me means that the process must be able to cope with changing circumstances and still be able to deliver what it promises. So we must still reach the destination, but the road we take might change depending on circumstance. So good live information supply is worth more than a process model for these kind of processes.

    I think you must differentiate between operational adaptiveness (changes on the playing field, during the game) and, what most like to call, process improvement (changing the process in the locker room and try to play better in the next game).

    Actually I have a hard time answering this question, because I believe execution is the only thing that matters for a process. And adapting during execution....do you need models for that?

    Maybe, only when the process is executed by a BPMS which is model driven. Many vendors promise us that the model is execution. SO when you change the model, execution will change immediately. In that case you don't need a special model for that only a fast way of changing it.

    I am not sure if all these stories are true. (but I will not do vendor bashing in this post ;-)

    But when a process is not supported by a bpms, what value does a model add? Working instruction, checklist, overview of process rules?

    I think then adaptation comes to flexible people with guts and empowerment with process focus. And as insight in a process is not the same as having a process model, you probably don't need a model for that. You need insight in all the cases in your process, if they are on track.

    So the combination between useful Process information, and empowered process aware employees is worth more than any modelling language for adaptiveness.

  • Do we need a new "modeling language"? No.

    Do we need a new model? Oh yes. The traditional flowchart-style model is still useful, but as I discuss at somewhat more length here, it has limitations. For example, it has resisted some valiant attempts to make the dimension of time a more integral component of the model

    So, we came up with another one, an MS Project-style model that supports time as a key component. Most customers will use both, taking advantage of the strengths of each to create complex but easy-to-modify processes.

    Others will have different approaches, but the bottom line is that the flowchart is simply not powerful enough to be the alpha and omega of process models.

  • I agree with everyone here. But let me take a deep breath before I type out this detailed response.....and forgive me if this post turns out too long.


  • In a word no; not required. Have a read of this blog it gives a vision which is now reality http://infullbloom.us/?p=3222 “….how those models can become applications without any code being written or even generated”.
    “If I’m right, you’ll want to be on the agile, models-driven, definitional development side of the moat thus created…..”
    Naomi went on to say later “vendors are in trouble if not there (best) or moving quickly to object-based, metadata-driven, definitional development.”
    Emiel, they are not stories it is true the model where both build and changes take place is the application! The fact is that neither BPEL nor BPMN are required, indeed there is no “language” other than the model which in itself is the application built from business language working direct with users. It uses a declarative technique from the model to set up the database and is then ready to run. So the resultant application becomes as robust and as scalable as the database with no interpretation language required. I agree with Scott M time is essential as are rules with both incorporated. This site contains explanatory documents – A brief guide on how, a Q&A, key issues design philosophy and more. http://bit.ly/qlSUvM
    Every one of you could use and build any enterprise level “BPM” driven application without being a coder. I understand cynicism but it was just a question of time for such a step change in Business Software Technology. After all business logic never changes - think about it?

    • David,

      When I am not blogging and bashing I work for one of those vendors ;-) So I know it is some kind of true.

      But (to me) it is still locker room adaption. It will take some time to bring changes to the playing field.

      I am a greater fan of processes that are agile on the field, during execution. Some cases are getting too late? Change priorities or give employees more roles.

      Too many orders coming in because of an accident? Change process rules to not overload some process paths.

      That kind of things. And that cannot be in the model.

      That is live, supported by good people, right information, supporting facilities and maybe a flexible bpms.

  • No, I don’t think we need a new modeling language. We will create more confusion. But I do think it requires a different mindset when you look at trying to map vs. discover adaptive or unstructured processes.

    We used a simple modeling approach where we just create independent events on a canvas and then let them be sequenced at run-time with an event-based rather than a flow-based BPMS. This means that there are no pre-defined flows or “arrows” on the model. Just a bucket of possible events. The real value comes when you analyze the historical data with process mining and look at the actual paths of these adaptive processes.

    Here is an extract from a reply to another blog post on some research that we did on adaptive/unstructured processes.
    “We’ve spent the past few weeks analyzing actual customer transaction logs to look at the real variations in processes that are mostly unstructured. We have the advantage of using an event-based process engine rather than flow-based, so our process mining exercises on various different industries, customers and processes showed the real variations that exist.

    We found, for example, that a customer on-boarding process for a specific services type organization currently have 24 identifiable “events” or possible steps that can happen. Much of the flow and process variation is dictated by the customer, governance requirements and process context. We analyzed 100 transactions in our process mining module and found 87 different process variants. Only when you look at it graphically you get sense for the real complexity. Most of these process paths are acceptable from the customer’s point of view as each had an explanation of “why” it happened that way. Some paths they would like to avoid and built appropriate process controls but the reality is that there are many unstructured processes where variation is the rule.

    A process with only 6 possible steps or events at a banking customer showed 75 path variations over 100 transactions.”

    Our research shows that these graphical process-mining models can be very complex and trying to create modeling language will have to account for the real complexity of visualizing or mapping even the simplest unstructured or adaptive processes.

  • Absolutely *Yes* a new language/model/approach is needed if you want an adaptive system. Emiel cam the closest to explaining this: You need something that is designed for change. You need something that any business user can modify, without having to know anything else about other parts of the process, and that change is always valid.

    Scott also makes excellent points suggesting that the change is not just the language but the complete way that you model. I think that is true. It might not be a new language, but an old one that fit the bill, but not BPMN.

    Theo is being delightfully cynical in suggesting that all new languages are only for the purpose of selling books.

    My position is that a 2-dimensional flow chart like language is unsuitable for an Adaptive System, and that is the central message in my new paper for the ACM Workshop in September. I find these three design criteria for a process definition language for ACM:

    • Ability to design a basic process quickly with very little investment by the user is far more important than the ability to define a precise process which uses more time an attention from the business user.
    • Process design must not require a skill beyond what business users possess.
    • For a process definition to be modified by business users, there must be no hidden assumptions.

    BPMN simply was never design to do this, and I don't believe any 2-dimensional flow-chart like language will do. This is not just to sell books, it is to solve real world problems.

    The program for the ACM workshop has just been formulated. This will be discussed. Here is a link:


    • Keith

      I agree, it's not BPMN but I'm not certain that we need a new modelling language. Yet.

      I think we need a new perspective and I know that you've been hard at work on it with ACM.

      Pity I can't be at the workshop but will follow it closely. It's great that non-workflow BPM is getting a voice.

    • To me this is still process management (it focuses on the process), but in a more adaptive way of executing it.

      So I always wondered why it is not called Adaptive Process Management.

      Case management should focus (at least that is what I think) more on the case and finaly will be a combination between BPM (APM), ECM and CRM (and maybe some other 3 letter words).

      But I'll see if I can collect some money to buy diesel to drive to Tallinn ;-)

  • This is to respond to some issues raised by Emiel, Keith and Pieter based upon our experience. First the ACM workshop thanks an interesting link. Just to reconfirm my earlier contribution a “declarative configurable” as to be presented in the Idea Paper already exists. Bill Gates expressed a vision in 2008 he used the phrase “declarative language” as the future but any “language” will be an inhibitor to both scale and agility. The Short Paper on “data centric BPM” again this is in existence – the trick for real agility is to remove need in build for coding, code generation and compiling.

    We have found that most business users are reluctant to do a change to a process. However in the design anticipation of change can be built in to allow easy change to say rules or operational parameters. Emiel raises good examples and such flexibility can be built in a variety of ways but remember that creation of information requires full audit trail. I think that Keith is right and wrong on user involvement. Right in the build must be in user language but to expect a user to build - it is not his day job? Plus like it or not legacy data raises its ugly head and this will require a business analyst to drive forward supported by the inbuilt tools required in any BPM solution – users will run a mile from this! Key processes may even be audited so change requires permission otherwise we will end up with another “meltdown” as the minority unscrupulous see opportunities…!

    Yes Emiel that link is depressing – SOA as linked to “BPM” just confuses the issue which of course big vendors like to do?

    I think all are agreed BPMN is not up to the job. Like wise the old “workflow” looks like is past its sell by date. I have referred to this blog in earlier responses which goes into a lot of detail (maybe too much but worth reading) http://confusedofcalcutta.com/2012/06/17/continuing-with-the-social-enterprise-and-flows/ He comes up with the concept of “collaborative flows” I think this is a good description of the way forward?

    Pieter’s analysis of events (another essential requirement) is interesting however in reality there are less that 13 generic task types including the UI that address any business requirement and from these events happen. The model is the “language” using these task types so no need for a new modelling language? Actually such simplicity allows ready build of complex process such as pricing and product configurators, means testing etc. We are now building intelligent processes that can adapt dependent upon previous action by users. The intelligent task is on the agenda!

  • I think this question is an interesting Rorschach test based on the answers. We're all bringing our biases to the table - what does "language" mean to you, what does "model" mean to you, and what does "adaptive" mean...

    • Scott good point that the exact semantics of "language" is sometimes being stretched in this discussion. However, it is not a problem because most languages, like BPMN 2.0, work only with one particular way of modeling, and so the two come as a pair. Both the language and the modeling method need to be appropriate for an adaptive process.

      However, understanding "Adaptive" is critical to the discussion, and another good point is that some may have misunderstanding about this. That would lead to the wrong conclusion.


  • Peter, you were amazingly prescient in calling for a discussion on this topic at this time, because I was just this week finishing up a workshop position paper on the same topic. I have converted this to a blog post:


Add a Reply

Recently Commented On

Monthly Archives