The False God of the Almighty algorithm

Evolving Excellence has a great article (that gave me some pause when I read it) about the False God of the Almighty algorithm. The reason that I read the article at least twice and then all the following comments as well is because I do have a deep interest in algorithms as well as lean principles and the article made it seem as the twain shall never meet. Perhaps, they should and perhaps they shouldn't but a larger point about invoking, adopting and implementing ERP systems was made which I find quite relevant. I am of the opinion that ERP systems are bound to face radical competition but I didn't think about the source of that radical competition. The reasons for holding this opinion are as follows:
1. They're too cumbersome to implement and consequently the failure rate is high for firms that go that route
2. The firm has to adapt its processes to what the ERP is mostly configured to do (No, you won't hear that from any vendor. In fact, you'd hear exactly the opposite). This is quite akin to Ford's famous quip a while ago, "You can have your car in any color as long as its black."
But that is not what the article is really about. In a nutshell, the following quotes illustrate what the story is really about:

"These are two radically opposite worlds ... One is the tech-savvy and IT-powered optimization world and the other is the pencil-and-paper problem-solving world. Which world should we live in?"

and,

"What do we know about SAP, and how well it integrates with lean principles (or lean implementations)."

The first thing to note is that optimization problems are notorious to handle all by themselves (but elegant in the formulations that have been drawn up) but when they're applied to real world problems, the modeler first creates a system (an artificial world), then models the problem within the system and finally attempts to solve (hopefully) the problem to optimality (hopefully) in a brief flurry of command line code instructions. And voila, you have an optimal answer. However, the model, the answer and the system are first and foremost matters of interpretation. Why there is a lot of WIP sitting between two furious machines is not a matter of interpretation as much as it is a matter of fact. The other (secret and don't you ever say that I said it) shady part of optimization applied to real world problems is that not all of what is called 'optimal' is truly optimal (or mathematically optimal) but no salesman will ever tell you that either out of sheer ignorance or cognizance that its the best that can be done. If I were to peer behind any ERP vendors so called optimization algorithms, I'd be sure to find (hopefully) a hybrid of optimization and heuristic algorithms embedded within that are spitting out answers. So that's what really happens on the optimization side of things.
On the lean side of things (and as lean goes, I shall be very brief in my answer) - you work in a philosophical framework that is bent on eliminating waste in a real system.
So what happens when the optimization world of IT clashes against the problem solving world of lean (or at least that's the way the clash is cast):

wasteful processes being proceduralized in algorithmic stone, monstrous amounts of extra inventory generated to accommodate the cascading "schedule risk" of individual operations, and of course implementation costs that can exceed $100 million buckaroos. And interestingly enough, several people chimed in with how they have gone back to using simple visual controls and Excel spreadsheets to schedule complex operations.

If one wants to clarify the clash, it would be instructive to consider some real evidence (as opposed to the neat tick-tock world of a an optimization/heuristic modeler) and that evidence is Toyota:

About the most complex type of factory is one that makes almost a thousand cars with several hundred permutations every day. And Toyota does it with no MRP-type shopfloor control. MRP is used to handle financials, inventory costing, and the like... but shop floor control is pure manual pull with a small number of e-kanban type applications.

This kind of evidence is particularly damning because in contrast to the system that the optimization resides in, the application of lean on a shop floor deals with reality - a tangible profit creating and wealth generating reality. You can't explain that away. And,

Excellence through simplicity." To me that quote from Lao Tzu has always epitomized one of the fundamental tenets of real lean. Don't proceduralize complexity, and don't make something more complex than it needs to be. Manufacturing really isn't all that hard... you make something, preferably one of it, and you get it out of the operation as quick as possible. Once you remove the loads of WIP from the floor by focusing on the velocity of the single unit, you begin to realize how so much of that perceived complexity is due to not having an unwavering desire to get a product through the flow as quickly as possible.

The above discussion (or polemic, depending on how you see it) is why I go hmmmm... because ERP is asking for competition. So what am I going to do? That is the only question.

Tags: ERP, MRP, Job shop, Planning, Lean, Flow, Optimization, Heuristics, Algorithm, Supply Chain, Supply Chain Complexity

posted by AJ @ 2:56 PM