Greg here—Bob has emerged from his igloo in Minnesota with this column.  Enjoy!

Something to look forward to …

Members of the KJR community know that businesses can optimize their operational processes in no more than three at a time out of these six possible dimensions: Fixed cost, incremental cost, cycle time, throughput, quality, and excellence. For definitions, look here: Do six dimensions make a business multiverse? – IS Survivor Publishing .

If you’ve ever been involved in a process optimization effort you’ll know that relatively few process consultants even recognize the six dimensions’ importance. You’ll also know that simplification is the process optimizer’s go-to solution to all process improvement challenges.

What simplification means in practice is that business management wants fewer defects and lower incremental costs, and wants them enough to give up on excellence – the ability to customize, tailor, and adapt – to get them.

As a practical matter this quality vs excellence and cost trade-off explains why, when men buy pants, they can get them with even-numbered inseams only. It’s personal – I need pants with a 31” inseam, but the best fitting pants I can buy come in at either 30 or 32 inches.

A pants manufacturing and inventory stocking process that delivers only even-numbered inseams, it appears, must be simpler than one that delivers inseams to the nearest inch.

Why does simplification work?

It works because we human beings are limited critters. Show a business manager a swim lane diagram with seven or fewer boxes and branches and they will understand how the process is supposed to work at a glance.

Show the same business manager a swim lane diagram with twenty boxes and they’ll have to study it to make sense of what’s supposed to happen.

So in our current state of process engineering sophistication, if we want excellence – the ability to create highly customized and tailored process outputs – we have to be willing to pay for it with higher fixed costs (say, buying more precise pant pattern cutting machines), reduced quality (31” inseams that actually vary from 29.5” to 31.5”) or increased waste leading to higher incremental costs by increasing inspections and discarding pants with inseams more than a half inch different from 31”.

Imagine some bright staffer suggests a way to get the excellence you want to bring to the pants marketplace, but their solution entails a radically more complex process – one with 23 swim lane boxes and branches.

I’m so sorry.

Except that when you look at the situation more closely you’ll recognize a hidden assumption – that the seven-box process limitation is intrinsic.

But there’s no reason to make that assumption. Enter a speculative possibility: Use artificial intelligence to overcome the seven-box limit.

There is, after all, no reason to expect a competent process-design AI to succumb to it. Quite the opposite: we can predict with some confidence that an AI process designer, equipped with genetic design algorithms, could cope with far more process complexity – more boxes and branches, that is – than we lowly humans could manage if left to our own limitations.

Bob’s last word: Ten increasingly short (“decreasingly short”?) years ago I predicted a shift in business emphasis, from quality as primary driver to excellence (“More storm warnings,” 3/4/2014). I’ve seen nothing that suggests the process optimization industry has kept up. Lean, six sigma, and lean-six-sigma still monomaniacally focus on quality improvement. Theory of constraints still says quality improvement is nice, but increased throughput is the name of the winning game.

And business process re-engineering’s devotees still don’t seem to have figured out that BPR as usually practiced is just waterfall software development without the software, but with the same structural flaws.

Interestingly, a bit of googling (and CoPiloting) reveals that there has been work done to apply genetic algorithms to complex process optimization challenges (for example, “An improved genetic algorithm for multidimensional optimization of precedence-constrained production planning and scheduling | Journal of Industrial Engineering International (springer.com) , Son Duy Dao, Kazem Abhary & Romeo Marian, 2017).

And this was published before the recent wave of AI-everywhere enthusiasm. Which leads me to suggest a name for this new partnership of artificial intelligence and genetic algorithms: AIGA (pronounced “EYE gah”).

Whether the current crop of process optimization consultancies will adopt AIGA or not is anyone’s guess.

My guess? Nope.

# # #

If you’re following AI trends and want something that isn’t just the same old same old, you’ll want to read Bob’s last two CIO Survival Guide entries: SharePoint Premium highlights the hard road CIOs face with generative AI | CIO and The last thing most CIOs need is an AI plan | CIO .