Game Theory in Management

Innovation has a way of changing everything, and usually in utterly unpredictable ways, which is why I’ve described it in this blog’s title as the ultimate change agent. Want to be a true agent of change? Come up with a way of providing a service or manufacturing a product better, faster, or cheaper, and the proverbial world will beat a metaphoric path to your doorstep. But what about those instances where the change sought isn’t of recent invention or discovery, but the target organization is nevertheless deficient in that particular capability, like, oh, I don’t know, Project Management?

This is where it gets fairly complex. GTIM Nation is familiar with my respect for Thomas Kuhn’s The Structure of Scientific Revolution (University of Chicago Press, 1962), where he points out that, while people tend to look back on the progress of invention and innovation as being somewhat even, what actually tends to happen is better described as advancing in fits and starts. An example Kuhn uses is the field of cosmology, and the transition between the Ptolemy model (Sun and planets revolve around the Earth) and the Copernicus model (Earth and planets revolve around the Sun). When our ability to collect more and more accurate data leads to observations that appear to contradict (or even overturn) the commonly-held theory in a given field, addendum to the prevailing theory, named cycles and epicycles, will be produced that appear to explain the anomalies in order to keep the prevailing theory from being completely overturned. Eventually, though, enough new data accumulates to initiate a crisis leading to a “paradigm shift” (yes, Kuhn was the first person to coin this phrase). It’s at this point in the transition that things get really interesting, at least as far as parallels to management science are concerned. In the run-up to the paradigm shift, many of those holding to the conventional wisdom will aggressively resist the newer theory, and do so in ways that suggest that their objections are not purely scientific in nature. After all, to cite the late Michael Crichton[i], real science has nothing to do with consensus – if one researcher in one laboratory can show reproducible results from a valid experiment based on Theory N, and Theory N is incompatible with Theory X, then Theory X becomes discredited.

Meanwhile, Back In The Project Management World…

All of which leads back to the at-first-glance strange phenomena of a management science world that is often unwilling (unable?) to adopt well-established business model innovations, specifically in Project Management. To be fair, I hold that much of the blame for this highly uneven adoption rate lies within the PM community itself. A person can tolerate just so much eat-your-peas-style hectoring in seminar paper presentations before the threat of being tagged as not “doing PM properly” ceases to have any effect as a motivational factor. That having been admitted, I believe that the majority of the energy against the acceptance and implementation of PM techniques resides with those who have something to lose if such techniques were enacted. Of course, I’m talking about our friends the Asset Managers.

Is there a focus on Asset Management techniques in the business world now? Almost certainly. The assertion that the point of all management is to “maximize shareholder wealth” is still being widely taught at the college level, despite the influx of data challenging that little axiom’s validity. Even if such a focus could be adjusted, we would still have the legal requirements of a comprehensive accounting system, quantifying virtually every transaction in the organization, remaining. Clearly, advancements in PM capability have nothing to do with challenging, much less overturning, Generally Accepted Accounting Principles (GAAP), but it’s equally obvious (at least to us PM-Types) that a project’s Estimate at Completion (EAC) is best calculated using an Earned Value Management System. Trying to use projected average spending rates to derive it just doesn’t cut it.

When we’re talking modifications to an existing business model, we must acknowledge that those who are comfortable with the existing processes and systems stand to lose if such changes are successfully implemented, either with respect to ease of work or being seen as occupying a certain status within the current paradigm – hence massive resistance to change, even if the erstwhile change agent has discovered a superior (or even optimal) technical approach, and even if the existing problems represent a clear danger to the organization’s continued survival.

So, what’s the solution? Come up with an innovative implementation strategy. Genuine innovation is a powerful (ultimate?) change agent, especially when the commonly-employed devices of haranguing management to do better or getting them to sign some updated policy or procedure that mandates proper PM techniques have been tried, over and over, and have largely failed. There are other ways of advancing PM capability within the macro-organization.

And a real change agent/innovator will find them.

[i] “If it’s consensus, it isn’t science. If it’s science, it isn’t consensus. Period.” Crichton, Michael, Aliens Cause Global Warming, Caltech Michelin Lecture,  January 17, 2003.

Buckle up, GTIM Nation. I’m about to take you on a series of mental exercises that may very well change your perceptions about change control/change agency (ProjectManagement.com’s theme for December), a concept central to virtually all medium-to-large projects. Proper management of the change control process is often key to project success, and I intend to show it in a rather novel light.

First stop: why do we even need change control? The obvious answer is because the project in question isn’t proceeding according to plan. Of course, no project goes entirely according to plan, but we don’t convene an emergency meeting of the Baseline Change Control Board each time the PM has to take an unplanned afternoon off to see the dentist. A change has to be significant to justify initiating the Change Control process, right? The real question then becomes, how big of a change are we talking? Consider the following graph:

The dotted line represents the original cost, scope, and schedule baseline. The dashed lines show the inherent flexibility in the baseline – it’s how much scope, cost, or schedule can deviate before some adjustment to the baseline is deemed necessary. Finally, we have the solid line, showing the actual experience of the Project Team. Note how, while October is within the bracket for determining the necessity of changing the baseline, starting in November the Project rides the upper bracket until February, where it exceeds the upper limit. Within this metaphor the BCCB would need to process a Baseline Change Proposal in its March meeting. But, as fate would have it, progress snaps back to being consistent with the baseline in April, and continues within acceptable limits to Project completion. There was no way that the March BCCB could have known beforehand that this would end up being the case and, in any event, they didn’t change the baseline for whatever reasons. Now consider this graph of the same project:

Alert GTIM Nation members will readily perceive that the only changes have to do with the values of the upper and lower limits of the brackets where the Change Control function would be invoked. In other words, this plan was robust enough to handle larger fluctuations in scope, cost, and schedule when compared to the first example.

So, what about a baseline leads it to be more robust, or at least less brittle? Note that nothing in the risk managers’ (no upper case) bag of tricks set of techniques could expand the tolerance brackets in the above graphs. The risk analysis document, reduced to its core, is basically a list of things that could happen to a project that would lead to the Team laying claim to additional budget, whether that budget has been funded or not, under the rubric that such events had been predicted and somewhat quantified. Indeed, the more extensive the risk analysis, the more the original baseline should be considered intolerant of the changes that will inevitably occur during project execution. To be fair, for any project awarded on a competitive basis (the majority of them, presumably) there is going to be Marianas Trench-levels of pressure to prepare a Basis of Estimate that meets the bare minimum of the Request for Proposal’s scope requirements, since most winning bids are also the lowest ones. This being the case, even bad weather events (for construction projects, among others) can and do receive their own line item in the contingency budget estimate.

On the other side of this strange little change control phenomena is one of the deadliest project influencers, scope creep. If bad weather makes an appearance in the risk management plan (no initial caps) due to its capacity for inducing delays and overruns, how much more so would the stripped-to-its-core cost or schedule baseline be harmed by the introduction of non-aligned additional scope? I would go so far as to assert that the propensity of certain customers to request some additional capability or detail that wasn’t in any of the scope baseline documents points to an assumption that the original baseline was fairly robust in the first place, not in need of additional budget or time for “just” this one little addition (I wonder how many instances of scope creep would be eliminated if the word “just” was prohibited at project review meetings).

All of which points to a couple of observations about the whole Change Control process, specifically that the more the original cost/schedule baseline is intolerant of nominal performance deviations, the greater the need for a Baseline Change Control Board, and that these are a natural result of the competitive bid process, particularly when the lowest bid that appears to be able to meet the scope requirement(s) is all but guaranteed the win.

Interestingly, Firm Fixed Price contracts are largely immune to the anomalies of BCCBs, or the basis of the Contingency Budget, or the documents associated with risk management, but that’s a topic for another blog.

No self-respecting blogger on the topic of Game Theory can go long without referencing the Nash Equilibrium. Named after Thomas Nash, the central character in the movie A Beautiful Mind, it refers to

…a concept in game theory where the optimal outcome is when there is no incentive for players to deviate from their initial strategy. The players have knowledge of their opponent's strategy and still will not deviate from their initial chosen strategies because it remains the optimal strategy for each player.

Overall, an individual can receive no incremental benefit from changing actions, assuming other players remain constant in their strategies. A game may have multiple Nash equilibria or none at all.[i]

So, what does this Game Theory theory have to do with Change Management/Change Agency in a Project Management context (ProjectManagement.com’s theme for December)? Plenty.

I, like virtually every other Game Theorist out there, believe that the business world is highly analogous to some grand game, with economic rewards, certainly, but other prizes as well. Many aspects of the business world inform Game Theory hypotheses, such as the notion of the “logical man,” or one who will consistently (if not exclusively) act in their own observable self-interest, whether or not that self-interest is confined to economic benefits. Similarly, much of Game Theory makes an appearance in business environments, such as commodities trading, which often takes on the characteristics of a zero-sum game. Heck, even the brilliant Michael Maccoby named his most successful business behavioral archetype “the Gamesman[ii].”  I think it’s obvious that a substantial amount of overlap exists between the two realms.

Now, lets take another look at the Nash Equilibrium definition. “…there is no incentive for players to deviate from their initial strategy.” (emphasis mine) For those PM-types who have been brought into a medium-to-large organization to set up a Project Management Office, and (presumably) advance that organization’s PM capability, success involves way more than getting your PMP®, memorizing EIA 748, and knowing how to administer Critical Path Methodology software’s databases. You can flawlessly execute the following steps:

• Get the accountants to agree to collect costs by Work Breakdown Structure elements,
• Send the Project Controls specialists to Earned Value and Critical Path school (or hire ones with that expertise already),
• Train the PMs and Control Account/Work Package Managers in the basics of PM,
• Establish usable Scope, Cost, and Schedule baselines,
• …with Baseline Change Control Boards, staffed by willing participants,
• Arrange for monthly Project Reviews, where Gantt Charts and Cost Performance Reports are the main discussion points,

…and still fail miserably if your implementation strategy isn’t near-perfectly matched to the host organization. Why is this so? Because of the Nash Equilibrium. The players Project Team members have been playing this game working here for some time, and they will often perceive “no incentive … to deviate from their initial (non-PM-oriented) strategies.”

Also, consider the influence of our friends, the asset managers. Like us, they exert considerable effort in shaping the structure of the business model, and the Management Information Systems that make it work. Unlike us, their charter is, to a considerable extent, mandated by law. An Earned Value Management System is key to a successful PMO, but it doesn’t help at all when it comes to paying corporate taxes, and not paying taxes sends even very powerful people to jail. And I haven’t mentioned the very strict standards that go into preparing a profit-and-loss statement, or a balance sheet, or any of the other exhibits that form the basis of the asset managers’ function. The idea that the purpose of all management is to “maximize shareholder wealth” was pretty standard fare when I was attending business school. Long-term members of GTIM Nation know of my many objections to that little maxim, so I won’t rehash them here. I’m merely pointing out that, while basic PM-oriented cost performance practices and techniques are not incompatible with Generally Accepted Accounting Principles, in practice they do represent an alternative way of processing, presenting, and making decisions based on management information streams. My experience and education inform me that the introduction of project performance information systems that are not dependent on the general ledger represent something of a rival to the asset managers’ influence on the overarching business model. In other words, most of the organization has probably already adopted a series of non-PM-related management strategies, and there is no real incentive to change.

So, essentially, the advance-PM-capability game can be arrayed against us PM-types from the start, no doubt accounting for many failed attempts at doing so. Blaming such failures on such amorphous aspects such as “resistance to change” doesn’t do any good, particularly for those tasked with becoming change agents. This leaves us with an overwhelming (?) question: are we ready to win at effecting organizational change, even in those cases where the odds in the game are against us?

Well, are we?

[i] Retrieved from https://www.investopedia.com/terms/n/nash-equilibrium.asp on December 5, 2022, 18:33 MST.

[ii] Maccoby, Michael, The Gamesman, The New Corporate Leaders, Simon and Schuster, 1977.

Agile Project Management is very popular among Information Technology (IT) portfolios, and for good reason. It can be reliably employed to significantly increase the odds of any given IT project coming in on-time, on-budget. It is also a source of frustration for me, since it does have some notable shortcomings, but its afficionados tend to be oblivious to them.

Problem Number One is that, while Agile does streamline the configuration management problems that tend to accompany traditional PM and its Baseline Change Control Board (BCCB) lethargy, it can also impede the discovery and implementation of the optimal technical approach. To help illustrate this, let’s turn to the Game Theorists’ favorite tool, the payoff grid. Consider two scenarios (simultaneously), where a given IT Project is headed by a manager who has identified the optimal technical solution, and is pursuing it, versus an analogous Project where the PM is rather unsure or unclear about the best way of addressing the Project’s central problem. Add to this two environments, one Agile, the other traditional PM. Here are those scenarios in a payoff grid:

Let’s dispense with Scenario D first, since this is likely the environment that hatched Agile PM in the first place. If the PM has selected the optimal technical approach to the central problem of the IT project, the last thing she needs is an ossified BCCB curtailing her latitude of movement when it comes to configuration management. A key characteristic of Agile PM is that it reduces the number of opportunities for, ahem, stakeholders to influence the technical agenda in the name of expediting progress (take that, Communications specialists!), meaning that, if that very technical agenda is the best fit for the organization, most superfluous input isn’t even allowed a venue. Conversely, if the best technical approach is unclear or even unknown, a traditional PM approach (Scenario A) can provide greater opportunities for more voices to be heard in identifying it, assuming the PM is open to the greater number of stakeholders.

The remaining two scenarios are where we can easily get into Project Management problems, with no nominal capacity within Agile to correct. The most readily-available boogey-man for those wishing to avoid traditional PM techniques in IT projects, Scenario C, is rather difficult to refute. IT projects tend to be far too dynamic and fast-paced to have use of a BCCB that meets every month, or even bi-monthly, where almost any member can slam the brakes on a recommended or needed scope update. Recall the old saw that there are no solutions, only trade-offs. What’s being traded in Scenario B? Well, if the optimal technical approach is either unknown or unclear, the removal of the BCCB carries with it fewer stakeholders providing input. This input can be inessential, but every now and then it can also be game-changing in its capacity for steering management decisions in the right direction.

Then we have the problem of implementation. Probably the only genre of software that isn’t in need of some sort of consideration of how it is to be implemented to the wider organization is games, whose customers purchase it for entertainment. I believe most everyone else is looking to enhance a management capability, which means we’re tinkering with a business model. Make no mistake: no matter how blatantly obvious that an extant Management Information System (MIS) is failing to perform adequately, there will almost always be someone within the organization who will not want it to change. In a mid-to-large organization, you can even expect some members to actively work against any transition, which leads us to Problem Number Two: while Agile may enhance and streamline the Configuration Management process significantly, it really doesn’t address the implementation strategy. As stated earlier, this is irrelevant in the development of game software, but for virtually everything else it’s going to be an issue.

“But Michael!” I can hear GTIM Nation say, “why are you laying the implementation issue at the doorstep of Agile PM? Don’t other projects have similar problems?” No, not really. You generally don’t need to convince the members of the client organization to occupy a new office building, or municipality to hook up their power grid to a new dam. Software is different like that, and, since Agile lays claim to being an adaptation of traditional PM for Information Technology specifically, I think it should be at least somewhat on the hook for this oft-encountered barrier to successful project completion.

For those who insist that simply mandating the usage of a new or updated system, on pain of discipline, is sufficient for IT project success, I would counter that a lack of an appropriate technical approach to the actual implementation will more than negate any advantage from using Agile over traditional PM in the first place. The only question left is, what are you going to do about the other half of the Agile equation?

Nicknamed “The Prince of Paradox,[i]” philosopher and writer G.K. Chesterton asserted the following:

“In the matter of reforming things, as distinct from deforming them, there is one plain and simple principle; a principle which will probably be called a paradox. There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, I don't see the use of this; let us clear it away. To which the more intelligent type of reformer will do well to answer: If you don't see the use of it, I certainly won't let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.”[ii]

The shorthand version of this quote is “Don’t ever take a fence down until you know the reason why it was put up,” which is close to capturing the overall meaning, though I would add the observation that, where Chesterton thought the modern type of reformer would go up to the fence “gaily,” I would substitute “naively.” It’s an easy conceit that informs younger generations that all that was set up before with which they take exception was done so out of ignorance. Properly trained auditors know the basis for their comparisons, and appropriate methods for evaluating against any given audit standard. Poorly (or even un-) trained auditors simply show up and start condemning each and every condition or technique with which they disagree.

Meanwhile, Back In The Project Management World…

A common and frustrating characteristic of reformers, in this case those who would implement some more “advanced” PM capability, such as Agile/Scrum (ProjectManagement.com’s theme for November), is that they will often approach their work with the confidence that their proposed business model updates will work for their new organization, if only the targets of the reform would (a) recognize the superiority of the reformers’ approach, and (b) cooperate with it fully. This is pretty ironic, considering the practice of software engineers to conduct what’s known as A/B Testing. In situations where older code is undergoing significant additions or modifications, or when, say, a website isn’t performing as anticipated, a standard practice is to avoid implementing the modifications all at once, and instead update characteristics incrementally, constantly comparing the new version’s performance with the existing. By approaching the update in this fashion, the team avoids the scenario where they find themselves having worsened the situation from where it was when they began, with no idea precisely which change or combination of changes caused the reversal, and, therefore, no cogent path to reclaim their losses. Without a comprehensive knowledge of all of the other modules that interact with the one being changed, it’s almost guaranteed that a significant number of these modules will fail, perhaps causing the modules depending on them to also fail, initiating a cascading effect across the entire application. And, just to be clear, it is extremely difficult to get a handle on each and every connection among nodes in even a relatively simple network.

Somehow, though, this caution in tinkering with existing nodes in a network tends to evaporate when the discussion turns to implementing a novel PM strategy, such as Agile. Make no mistake: business models are highly complex networks, whether they came into existence through a master plan, or entirely holistically, by piecemeal. One might be able to change out, say, the software package for the general ledger with relatively few implementation issues, since the functioning of the GL is highly regulated. Not so with Project Management software packages or approaches, where even the appropriate level of system rigor tends to be a highly contested matter. If an organization finds itself in the software marketplace through, perhaps, some Project Team members developing an app that solves an acute problem afflicting entire portfolios, but that organization was ill-disposed towards setting up Work Packages and Control Accounts, it’s highly unlikely that they will suddenly embrace that very change just because they are now being asked or required to create Epics, Stories, and Sprints. And even in those circumstances where the organization has a fairly robust traditional PM capability, there may very well be legitimate reasons for not streamlining the configuration management/ change control process. It would, then, be a mistake to attempt an Agile management implementation until after those reasons can be discovered, evaluated, and addressed, lest they manifest in the new business model.

To be sure, all of this can be done. It simply requires the flexibility to be cognizant of the reasons behind the configuration of the existing business models. In short, it will require the Agile PM reformer to be, ahem, agile enough to hop over Chesterton’s Fence.

[i] Retrieved from https://www.christianitytoday.com/ct/2001/augustweb-only/8-27-52.0.html?paging=off#bmb=1 on November 11, 2022, 17:31 MST.

[ii] Chesterton, G.K., The Drift from Domesticity (1929).