Game Theory in Management

Alexander Tytler (1747-1813) was a professor of history at the University of Edinburgh[i], and asserted in his writings that democratic forms of government had a lifespan of roughly 200 years. During that time, he theorized, they tend to go through the following stages, in order:

• Bondage
• Faith
• Courage
• Liberty
• Abundance
• Selfishness
• Complacency
• Apathy
• Dependence,

…after which the cycle begins again.

Similarly, in Geoffrey Moore’s classic Crossing the Chasm: Marketing and Selling High-Tech Products to Mainstream Customers (Harper Business Essentials, 1991) a template-like structure is posited, with recommendations included for those instances where predictable business scenarios unfold sequentially due to failures in the management strategy or business model. I believe that Project Management Offices are also susceptible to following a predictable pattern, or structured life-cycle, but I would add in a leading indicator when such phases enter into transition mode: an accompanying Organizational Behavior and Performance marker which, when taken together with the nominal PMO life-cycle transition, provides a clear indication of the trajectory that the PMO is taking.

My list of stages that many PMOs will encounter looks like this:

1. The project-based organization will experience one or more project disasters, where the budget is overrun, key scheduled milestones are missed, and the customer(s) are disappointed.
2. Some executive will suggest – or a long-ignored PM-type will be finally heeded when they recommend – the creation of a centralized PM function, or “area of excellence,” in order to help alleviate current PM difficulties, and avoid future ones.
3. With much enthusiasm and a tentative budget, the new PM organization is started, with a combination of existing and newly-hired personnel.
4. The new PM organization will set out to write procedures and guidance documents, or import already-existing ones, and attempt to persuade all of the project teams to perform work as written. Almost everyone will comply, either because they recognize the inherent value of doing so, or because they don’t want to be seen as incalcitrant.
5. As the Earned Value and Critical Path Methodologies take root, project problems are revealed early enough to prevent overt disasters, and the overall portfolio begins to perform noticeably better.
6. Large overruns and delays become a distant memory, and smaller ones become less frequent. As a sense of complacency sets in, the budget supporting the PMO comes under scrutiny.
7. One PM will refuse to implement the PMO’s processes, citing some (invariably invalid) reason that their work falls outside the purview of the guidance. Once they get away with doing so, other PMs will also look to escape such directives.
8. With so many PMs opting out of the PMO’s view, it can no longer justify its budget. Funds go away, along with the truly talented members, until…
9. A project disaster occurs, and the cycle begins anew.

However, I have become convinced that another, more organizational and behavior-based cycle is also in play as these steps unfold. During Steps #2 and #3, talent is not necessarily the most important determiner of who gets hired or promoted. It’s not unusual for the coin of the realm for a new management team to be loyalty, not talent. The new PM executive team may feel uneasy about the prospects of their canned strategies working in the new environment, not because they feel that these strategies won’t work, but because they don’t know the level of the existing organizations’ willingness to execute the new technical approach to advancing the PM capability. Of the previous numbered steps, only #5 and #6 truly need PM talent to pull off. All of the others can be performed by the medium-level performers. This leads to something of a dilemma for the PMO Director – do you attract and retain the talent needed to find and implement potentially novel solutions to the organization’s PM difficulties, or do you hire and promote those whom you trust to execute an already-derived strategy, with minimal (or no) course corrections needed? With very rare exceptions, each time the dilemma is resolved in favor of the loyal over the talented, the PMO moves away from operating as a meritocracy, which increases the odds that the steps will continue towards their unattractive end-state. Is there a way to interrupt this cycle, with its pessimistic implications? I believe so, but it involves…

Ooops! Look at that. I’m out of pixel ink for this week. Check out this blog next week for my recommendations of how to escape the PMO’s predictable life-cycle.

[i] Wikipedia contributors. (2021, October 15). Alexander Fraser Tytler, Lord Woodhouselee. In Wikipedia, The Free Encyclopedia. Retrieved 04:50, November 1, 2021, from https://en.wikipedia.org/w/index.php?title=Alexander_Fraser_Tytler,_Lord_Woodhouselee&oldid=1050002662

As anyone who’s attended a management meeting can tell you, the capacity for disagreement and contention in such gatherings is vast. But everyone is talking about management, right? Since the first college-level business school started 135 years ago[i], shouldn’t many, if not most management issues commonly encountered have something of a pat answer to them? That a high level of conflict can erupt in what would otherwise be considered as a fairly anodyne setting points to the fact that, despite what the faculty of the myriad business schools might have you believe, there are amazingly few standard approaches to management problems. I think a major reason that this is so is due to the fact that the various management disciplines carry with them inherent conflicts of interest, and the inability (or refusal) to recognize these conflicts for what they are inevitably leads to more of them.

The main three business model structures involving Project Management that are intrinsically tied to organizational conflict are:

1. Resource versus Facility versus Project Management;
2. Asset versus Strategic versus Project Management, and
3. …the old Quality, Availability, Affordability – pick any two paradigm.

Consider a large manufacturing or experimental facility. Any such facility is relevant, but the kinds of conflict-driven management pathologies I want to point out occur more frequently (and acutely) at highly specialized ones. The casual observer may believe that the three groupings of managers – Facility, Resource, and Project – all have a common goal, but this is so only in the equivocal sense.

• The Facility Managers seek to keep their plants’ capacity as high as possible while minimizing the number of safety (or similar) incidents that would lead to a pause, or even shutdown.
• The Resource Managers’ goal is to make available the labor and machines needed for the Project work at as close to optimal levels as possible. If they have to place resources on leave, or are placed in a position of having to tell the PMs that their scope can’t be executed due to a lack of resources, they have a really bad day.
• The PMs, of course, are interested in attaining their projects’ scope within cost and schedule constraints. When they go over on either of these parameters, they have a really bad day.

To engage in a level of hyperbole, the Facilities Managers don’t care as much about missed milestones or negative project variances as they do about keeping the facility up and running. The PMs don’t care as much about having to place personnel outside of their Project Teams on temporary leave, as long as their work is on-time, on-budget. The Resource Managers tend to care about the others’ success, but more because of how it impacts their ability to level their available resources against the demand curve than for other reasons. What we have here is an inherent conflict of interest, which can’t help to manifest in those meetings where management strategy and implementation is discussed and determined.

I’ve often made the point that Asset, Project, and Strategic Management have different goals and use different information streams to accomplish them, but it bears repeating since our friends, the Asset Managers, almost always believe that any relevant management information dealing with costs simply must originate with the general ledger. For today’s example, I’ll leave PM completely out of it by pointing out the impossibility of capturing the Return on Investment (the Asset Managers’ favorite evaluation parameter for determining whether or not a given course of action is “worth it”) on a proposed advertising campaign. Marketing initiatives, of course, do not benefit the PMO, and cost money, often times serious money. So, if they don’t benefit PMs or Asset Managers, who gains? The Strategic Managers do, since they seek to acquire more market share for the organization. Successful ad campaigns can (and do) bring in far more new revenue than their basic costs, whereas poor ones not only fail to make back their costs, they can actually harm the brand name they attempt to promote. Given that there is no reliable way to quantify the go/no go decision on launching a particular marketing initiative, what we have here is essentially a roll of the dice, and Asset Managers are not known for being fans of rolling the dice.

The last COI-revealing structure, the idea that, of Quality, Availability, and Affordability, one must choose two, is largely self-explanatory. The assertion that any viable business strategy must select which of these goods/services-providing tactics receive attention at the expense of another is simply a different way of stating that the attempt to do all three is futile.

Well-run organizations will realize that these inherent conflicts of interest among rival teams and schools of thought should be managed with the acknowledgment that there are no solutions, only tradeoffs. Less-than-well-run organizations will allow the principals of each team to battle out their points of view in highly contentious conference rooms, where the optimal approach to the organization’s problems is rarely discovered, much less implemented.

In the latter instance, though, some of those battles can become legendary.

[i] Retrieved from https://www.wharton.upenn.edu/about-wharton/ on October 25, 2021.

As anyone who has attended a PM-themed seminar or symposium can readily attest, training is a popular way to monetize PM expertise. There’s training for scheduling or cost performance software, training for risk management (NIC), training for PMP® Certification, and on, and on. I have also noticed that, when reviewing lessons learned or incident reporting documents, part of the get-well plan will involve some form of an enhanced training program a majority of the time (that, and updating procedures). But I have to ask: is training the most appropriate way of ensuring something bad that the organization has endured never happens again?

Meanwhile, Back In 1881…

To help illustrate my point I want to transport GTIM Nation to the years 1878 to 1881, in New Mexico Territory on the American frontier. One of the most notorious outlaws from that conflict (if not the most) was Billy the Kid, who was credibly accused of killing eight men, one of whom was a deputy sheriff. Pat Garret had arrested Billy once, but Billy managed to escape from that jail, killing three men in the process. Garret then re-initiated the manhunt for Billy, and, after hearing that he might have returned to the Fort Sumner area, went to see his old friend and employer Pete Maxwell on July 14, 1881. Garret and Maxwell were talking in a darkened room at around midnight when Billy himself unexpectedly walked in. At first, Billy didn’t see Garret, but upon realizing a third person was in the room started asking “Quien es? Quien es?” (“Who’s that?”) Garret, recognizing Billy’s voice, drew his gun and fired twice. The first bullet fatally hit Billy in the chest, the second one missed.

One of the things I find remarkable about this story is that Garret had no training in police work or constabulary. He hadn’t even been in the military[i]. Indeed, the first police training facility wouldn’t open for another twenty-seven years[ii]. Nevertheless, Garret tracked down an escaped outlaw – who had already been convicted of murder – on a first hunch, and shot him fatally on the first attempt, in a room so dark that Billy couldn’t make out Garret’s features in time to recognize him. Quite a feat of law enforcement in a difficult and hostile environment for somebody who had had zero training, no?

According to the Peak Performance Center website (https://thepeakperformancecenter.com/educational-learning/learning/principles-of-learning/learning-pyramid/retention-rates/), the National Training Laboratory’s Learning Pyramid quantifies learning retention by method of instruction so:

• 5% Lecture,
• 10% Reading,
• 20% Audio-Visual,
• 30% Demonstration,
• 50% Discussion,
• 75% Practice Doing, and
• 90% Teaching Others.

Virtually without exception, the PM-themed training offered today is never more advanced than the Audio-Visual category from this pyramid, and even that level assumes that a well-done slide deck can be considered to be “Audio-Visual,” as opposed to an enhanced “Lecture.” This fact, taken together with the previous observation that Incident Reporting/Lessons Learned documents often lean heavily on enhanced training as the go-to remedy for preventing accidents or problems analogous to the one that’s being reviewed, and we have a management scenario where the go-to remedy for even our biggest problems has no more than a 20% effectiveness rate. And, before GTIM Nation reminds me that I also pointed out that procedure modification is another oft-used tactic, I would like to preemptively highlight that procedures are documents. Documents are read (theoretically). And the Reading component of the Learning Pyramid has an even lower learning retention rate than the slide deck-enhanced Lecture.

I would like to propose that the preceding paragraphs have presented us with two extremes of the same scale: Pat Garret experienced improbable success with virtually no training at one end, with the assertion that using a training program as the first line of defense against the recurrence of large-scaled problems is probably flawed at the other end. In-between these two extremes lies the place where the use of training is an appropriate way of advancing an organization’s capability maturity. Note, however, that, away from this suitable range, training per se becomes either irrelevant, or ineffective.

Don’t misunderstand – I’m not knocking training, unless, of course, it’s training in risk management (NIC), among other trendy but ultimately useless management topics. I’m just urging GTIM Nation to be clear-eyed about its efficacy. Far from a universal panacea, it could be little more than a shot in the dark.

[i] Wikipedia contributors. (2021, October 10). Pat Garrett. In Wikipedia, The Free Encyclopedia. Retrieved 23:46, October 18, 2021, from https://en.wikipedia.org/w/index.php?title=Pat_Garrett&oldid=1049124810

[ii] Bykov, Olga, Police Academy Training: An Evaluation of the Strengths and Weaknesses of Police Academies, Themis – Research Journal of Justice Studies and Forensic Science, Spring 2014.

Before I launch into my problem with In Search Of Excellence (Harper Collins, 1982) by Tom Peters and Robert Waterman, I want to state up-front why it’s awesome. It’s refreshingly free of a priori assertions common to so many other business writings. Instead of pumping up their academic credentials and telling us how they thought the business world ought to operate, Peters and Waterman went out to organizations that were actually successful, and asked them to what they attributed that success. Distilling the results into eight characteristics of successful companies,

• a bias for action,
• staying close to the customer,
• autonomy and entrepreneurship,
• productivity through people,
• hands-on, value-driven,
• stick to the knitting,
• simple form, lean staff,
• simultaneous loose-tight properties,

…the book relays what has already been observed to work. Up to the time of its publishing, In Search of Excellence’s predecessors as the must-read books in business schools everywhere were written by Peter Drucker, whose writings had a far more top-down, way-it-ought-to-be tone. By taking a scientific approach to the business world, Peters and Waterman not only reset the conditions by which organizations were considered successful, they redefined what constituted successful writing about the management sciences.

“So, Michael” I can hear GTIM Nation say, “for a book that you admire, a massive best-seller, a virtual must-have for anyone with the word ‘manager’ in their title, what on Earth can you have against it?” The same thing that afflicts so many other business theories, books, articles, or paper presentations: it does not appear to have a limiting factor.

Back when I was writing The Variance Threshold column for PM Network, I once made a snarky comment that, to fully embrace Peters’ and Waterman’s theories, the business owner should give away all of the company’s assets to the next person who presents themselves as a “customer.” I’m in total agreement with the notion that successful organizations need an intense focus on fulfilling the customers’ wants and needs – I’m just wary of the fact that there does not appear to be a limiting factor. Retailers, for example, are infamous for operating under the axiom “the customer is always right.” And yet, even here they will (usually) not accept for refund or exchange a product they sold that has clearly been abused to the point of breakage. Obviously some level of limiting factors are in play, even if they’re not actually codified into some sort of procedure.

The absence of limiting factors is another reason why I pay the risk managers (no initial caps) no mind. In the March 2008 edition of PM Network, I engaged in a point/counterpoint discussion with Dave Hillson on a variety of risk management (NIC) topics, particularly whether or not it also encompassed “opportunity management” (whatever that is). After a segment where David discussed his usage or definitions of three different terms, I noted:

David, you redefined risk as uncertainty that matters. If the PMI Risk Specific Interest Group or special experts such as yourself have that latitude to redefine those terms in that fashion, then essentially, there's no aspect of project management that could be fairly said to lie outside the purview of risk management.[i]

Essentially, if the experts can’t tell you what something isn’t, then they really can’t tell you what it is.

The same is true of the Asset Managers’ favorite maxim, that the point of all management is to maximize shareholder wealth. If that were the case, then there would be no aspect of management – particularly and especially Project Management – that was outside of their scope, and that’s simply not the case. Nor is it true in the realm of Strategic Management, which seeks to maximize, not shareholder wealth, but market share. Since things like competitor buy-outs definitely help increase market share while hitting the profit-and-loss statement pretty hard, it’s fairly clear to me that the realm of Strategic Management provides a rather distinct limiting factor to the efficacy of at least one of the Asset Managers’ tropes.

Which brings us back to Peters and Waterman. Their message was sorely needed at a time when the Asset Managers’ rules permeated the business world, when PMI® was gaining a truly global standing by establishing a management model based on the customers’ expectations of Scope, Cost, and Schedule. An enhanced focus on the customer is excellent, but to have its desired effect the axiom “Quality – Availability – Affordability: pick any two” must inform the business model. Otherwise, it’s impossible to know at what point the limiting factor has been reached.

And you just might end up crashing the organization in the name of staying close to the customer.

[i] Hatfield, M. & Hillson, D. (2008). Danger ahead?: Some project managers contend there's a silver lining in risk management; others say it's called opportunity. PM Network, 22(3), 76–80.

The risk management (no initial caps) practitioners are very fond of setting up elaborate risk management (no initial caps) baselines and risk registers, collecting data, processing it using various Gaussian Curve-based techniques, and delivering … well, what, precisely? Now, before said risk management (no initial caps) practitioners flood the comments section citing their preferred artifacts, let me add in just one criterion: the output has to actually inform a project-changing decision or strategy.

Some years back I came across a story about one of the earliest college-level statistics courses ever offered, where the instructor started off the first class by producing a coin from his pocket, and announcing that, when he flips the coin, there can only be one of two outcomes: it lands with the “heads” side up (showing, in relief, the image of a person), or the not-heads side (in American parlance, this would be “tails”). He then proceeded to flip the coin. According to the story, the coin landed and stayed on its edge. I don’t know if it managed to find a crack in the floor, or if it simply spun on edge until it came to rest in the vertical position. I don’t even know if the story is true, but it does illustrate something about risk management (rather than keep typing “no initial caps,” I’m just going to use the acronym “NIC”), that, no matter how hard we try, we can never capture every possible outcome to a given situation, even one as controlled as a simple coin flip.

The same phenomenon often comes into effect when discussing Game Theory. One classic example involves the Ultimatum Game, where a researcher approaches two people (Person 1 and Person 2), and offers Person 1 this deal: the researcher will give the Persons $100 (USD) if Person 1 can provide an arrangement for how the money will be split between the two Persons, and that proposal is accepted by Person 2 on the first try. The Game Theorists calculated speculated that Person 1 would maximize their payoff by proposing $99 for Person 1, and just $1 for Person 2, under the theory that Person 2, looking at receiving either $1 or nothing at all, would (naturally!) approve of this distribution scheme. In reality, such offers were almost always rejected by Persons 2. Perhaps the Persons 2 felt that the Persons 1 were being unfair, or didn’t deserve such a lopsided distribution of unearned largess, or was somehow being disrespectful. Interestingly, when the actual evidence of the Game Theorists’ wildly inaccurate calculation speculation of the optimal distribution scheme was shown to them, a common reaction was to blame “cultural factors” rather than admit that their core assumptions were flawed.

Something very similar occurs when risk managers (NIC) become aware of unforeseen negative events happening to the projects to which they are assigned that receive absolutely no mention in the risk register, or any other document, for that matter. Instead of “cultural factors,” though, the risk managers (NIC) will invoke the “unknown unknowns” category of risk events to escape such obvious evidence of the absence of validity undergirding their analyses.

But I want to take a look at what happens even on those occasions where something happens to the project they’re working, and that something is referenced in the risk register. In the simplest analogy, a project to flip a fair coin would feature a risk analyst informing the PM-coin flipper that the odds of either pre-flip choice coming about is 50/50 (assuming that no coin-holding cracks are present in the floor). How does this inform the decision to pick one side or the other? Keep in mind that this example is one of the most limited-outcome scenarios that could possibly exist. Also, such outcomes are almost always driven by randomness, unless the coin flipper has acquired the skills needed to flip the coin so precisely as to arrive at a pre-determined outcome. While randomness does afflict Project Management, if that was the sole determining factor of project outcome there would be absolutely no reason to have an educated or talented Project Team. 

So, in the PM environment, we’re looking at two aspects that absolutely blow to smithereens any credibility the risk managers (NIC) would otherwise have left:

• The PM’s actions are not random, and therefore do not match any pattern that would otherwise inform an analysis based on Gaussian curves, and
• The environment is far from controlled with respect to its potential outcomes.

With these two factors irrevocably in play, there is simply no output, no report derived from risk management (NIC) or risk analysis (ditto) that can be specific enough to inform a particular decision or tactic. But a lot of time, effort, and expertise goes into such analyses, doesn’t it?