Game Theory in Management

I know what the ultimate Program Management Office is. It produces winners on a consistent basis, and is probably the closest to a pure meritocracy that any organization has ever, or will ever attain. Any American can join, but overseas GTIM Nation needn’t worry – analogous PMOs exist worldwide. Through this PMO, winning strategies and tactics are truly advanced, with no time given for unproven theories or modes of operation.

The organization I’m talking about? The United States Chess Federation.

“Wait just one minute, Michael!” I can hear GTIM Nation say. “The USCF isn’t a Program Management Office at all! It is, in fact, a collection of people who play a board game, albeit a very sophisticated, ancient board game.”

Fair points, all. But consider: what is a Program Office? Is it not a collection of PMs who manage projects sharing a common theme?

“Chess isn’t a project.”

Perhaps, but projects do share many characteristics with games (Game Theory in Management, anyone?). Besides, each game is unique, with a definitive beginning and ending date (time). Its scope is agreed to prior to its start (checkmate the opponent), and resources are dedicated to it.

But in my opinion what makes the USCF the ultimate PMO is the fact that its members’ value is definitively set, and on a purely objective basis: their score. You can be tall or short, skinny or rotund, beautiful or plain, well dressed or scruffy, and it makes absolutely no difference: if you win chess games, your score goes up. If you lose them, your score goes down. Period.

Compare and contrast this refreshingly objective evaluation basis to what happens in Project Management space. We all know of contractor companies that win large contract awards despite having massive overruns and lengthy delays in their histories. These guys lose, and lose big, but never seem to pay any kind of long-term penalty for doing so. Similarly, there are Project Management consulting organizations that deliver substandard services while charging inflated salaries, and yet, somehow, stay in business. Conversely, smaller, less influential contractors, with winning records, are often shut out from even competing for larger procurements.

Then there are the strategies and tactics that go into specific games projects. In the PM world, so-called experts can sit on the sidelines, pushing baseless assertions about the need to, say, time-phase Estimates to Complete, and how this “analysis” is absolutely critical to proper resource assignment, and print the codex of their idle speculations into guidance documents, and never pay any price for their error. Did any – any – of these people actually manage a project where the time-phased ETC prevented an overrun? No? Again, compare and contrast this to Chess Life magazine, which lists hundreds of Master and Grandmaster-level games. The reader can actually see which strategies and tactics worked, and which ones failed.

Chess players within GTIM Nation know that chess openings – canned strategies for the first 5-15 moves – have their own names, like Ruy Lopez, or the King’s Gambit. Indeed, virtually every named opening has at least several variants, each with their own name. Back when I was playing tournament chess, no player who had failed to commit to memory at least three openings (one when playing white, one when playing black and the opponent opened with 1. P-K4, and another when the opponent opened with 1. P-Q4), each with at least four variants (for those not counting, what was needed was a minimum of twelve game templates, played through around the tenth move) had any kind of chance. On the PM side, we have our own set of must-know canned strategies. The manager who does not know the difference between a Work Breakdown Structure and an Organizational Breakdown Structure, or isn’t familiar with how schedule logic can’t define a critical path within a network if most of the activities have been constrained, isn’t going to (or at least shouldn’t) last long in the PM profession.

In chess, winning strategies remain, and losing ones are abandoned. In Project Management, risk management is a multi-billion dollar per year industry, and yet has no real record of ensuring, or even aiding, project success. Ditto with communications management, or any of a dozen other, in my opinion, highly dubious assertions made in the theoretical quarter. These canned strategies simply seem to hang around and eventually become part of the conventional wisdom without ever having been shown to be the proximate, or even material cause of project success. At the risk of pushing my analogy past the breaking point, chess masters don’t have to deal with so-called communications experts, extolling them to keep their team mates (“stakeholders”) informed of what they’re doing, how, why, or when they do it. Nor do they have to listen to risk managers trying to tell them the odds that their opponent will employ a certain strategy (with an 80% confidence interval, don’t you know). Indeed, it’s illegal for a player in a tournament setting to be so advised.

If GTIM Nation comes away from this blog appreciating the strength of the analogy, great. If not, I understand. But to the latter category, let me add but this: The PMO that successfully enacts the USCF’s two main strengths – personnel status based on a pure meritocracy, and the ability to quickly and effectively reject losing strategies – will consistently out-perform the ones that don’t.

I believe that it’s axiomatic that the two most important questions that any Project Management Information System should be able to accurately answer are:

• At this rate of performance, when will this project be done?
• And how much will it cost?

Oh, sure, there are myriad others bits of information that can help shape the PMs’ overall strategy as well as day-to-day decisions, but I consider them secondary to reliably answering the two questions above. Why? Because a manager’s time is finite. The PM who scatters his energies among all of the activities and tasks within the project will be out-performed by the manager who knows which of her activities are doing fine without her attention, and instead concentrates on those tasks that are facing some kind of difficulty.

So, this being the case, has the optimal strategy for providing the answers to the two key questions been identified?

Yes, actually. In Dr. David Christensen’s paper Determining An Accurate Estimate At Completion[i] (National Contract Management Journal, 1993), he establishes that a calculated EAC is consistently and reliably within 10% of a project’s real at-completion costs once the project has passed the 20% completion point. This, in turn, was predicated on other studies that established the Cost Performance Index’s (CPI) stability. Since almost all EAC calculations involve the CPI (in fact, the most common EAC formula is total budget divided by the CPI), the establishment of its stability was, in my humble opinion, one of the great PM information system breakthroughs of the latter part of the 20^th century. It presented a quick, easy, and accurate way of answering the two main questions cited at the beginning of this blog.

For those of you who are newer to Earned Value Methodology, the CPI is calculated so:

CPI = Cumulative Earned Value / Cumulative Actual Costs

What’s Cumulative Earned Value?

Cumulative Earned Value = Total Budget * % Complete

The algebra whizzes within GTIM Nation probably already know what I had to be shown, that the Estimate at Completion formula, of dividing the Total Budget by the CPI, can be simplified to:

Estimate at Completion = Cumulative Actual Costs / % Complete

“Impossible!” the purists will say. “That critical piece of information can be accurately calculated with just two fairly-easy-to-acquire data points?” Yes, it can, and it gets better: the same formula works for duration! To know, within ten points of the final outcome, when your project will be finished, simply divide the cumulative duration by the percent complete to get total duration. Subtract out cumulative duration, and that’s how much longer you should expect the project to continue.

Now, has the GTIM blog just obliterated the need for comprehensive cost and schedule baselines, and complex Critical Path Methodology networks? Hardly. Those will always be in demand for advanced PM techniques and strategy development.

What the above assertions do (or at least should) overturn is the whole notion that EACs are best calculated using the “bottoms-up” method, where the PM (or cost analyst, or scheduler) re-estimates the project’s remaining work, adds this figure to the cumulative actual costs, and claims victory. Whereas the list of cumulative actual costs is usually reliable, that other component – the re-estimation of the remaining work – almost never is. Every single line item in the new estimate is subjective. It’s what the PM or estimator believes will happen in the future while ignoring what has happened in the project to that point, especially if a spate of poor performance has hit the project. That, of course, will be corrected going forward, the PM tells himself, as it influences away from reality each of the hundreds of data points going into the new baseline estimate.  The whole bottoms-up approach to the EAC is analogous to a person taking a one-mile hike, and being asked mid-point how long he thinks it will take to complete. Rather than look at his timer and multiply that amount by two, he instead tries to guess how far away he is from the finish line. If he had said prior to beginning the hike that he would finish at a certain time, the answer at the mid-point assessment will almost invariably be the same as it was at the beginning, available performance data notwithstanding.

I have no delusions that the above brilliantly-articulated analysis will bring a sudden halt to the practice of coming up with an EAC based on the bottoms-up approach. As a cliched method, it’s simply too entrenched in the commonly-accepted “wisdom” of PM aficionados and guidance-generators, who would rather embrace conventional wisdom than proven, repeatable management science.

And that preference is but one example of why PM is not as technically advanced as it could be.

[i] Retrieved from https://www.researchgate.net/profile/David_Christensen4/publication/228984762_Determining_an_accurate_estimate_at_completion/links/569f9d1908ae4af52546bb07.pdf on March 2, 2019, 13:28 MST.

I’ve relayed this story before, but it bears repeating for this week’s topic. I was part of a project controls team that had been asked to do some scheduling for a Human Resources department in the process of updating its software system. After pulling together and freezing the baseline, the team performed the first status pull, and took the results to the head of the HR organization. We pointed out that the project was likely to finish late, due to a few tasks on the critical path that were taking longer than expected. This communication was given on a Friday.

“Call in the entire department to work this weekend!” the manager demanded.

We were quick to point out that the personnel involved in the tasks that were pushing the finish milestone out were confined to just a few HR personnel, but mostly in the software engineering team providing support for the overall effort.

“No difference” he maintained. “This project is too important to risk delays while my staff is at home.”

“Just to be clear” we told him, “the vast majority of your department has nothing to do with the specific tasks pushing the project’s end date out. They will have literally nothing to do with accelerating the accomplishment of the activities we’re talking about here.”

The guy just stared at us, and pushed out the e-mail that demanded everyone in his (unfortunate) department work the weekend.

Of course, PMs readily understand the futility, and even the harm, of such a reaction. The project doesn’t benefit, not in the least. The workers’ morale couldn’t have improved upon seeing the e-mail demanding that they abandon kids’ soccer games, birthday parties, or even just relaxing time on their own, doing nothing at all – indeed, common sense informs us that that department’s morale probably cratered, especially since the everyone-must-work-free-overtime tactic seemed to be the go-to solution to this manager.

So why, do you suppose, did he do it?

I think the answer is clear. He probably spent considerable time in a college-level business school, where he learned that the purpose of all management is to “maximize shareholder wealth.” And what tactics make that happen? Well, getting the most out of your assets, of course! Are you the manager of a team of people, who get paid at a certain rate? Well, to advance the function of all management, by this teaching, one must get those assets to work over and above the level at which they are paid. It’s like getting free professional support, right?

I know that the vast majority of GTIM Nation – if not all of us – have experience working for organizations that expect or out-and-out demand “free” overtime. Once the initial thrill of becoming a staff member and having an actual salary fades, we quickly realize that we may have been better off back when we were hourly workers. I was laid off from the first company I worked for out of college after five years for the grave sin of charging two – two! – hours to an overhead account (when I had actually worked five hours on the overhead task), instead of charging every single hour to a direct account. My “manager” had given me the task, and provided the account I was to charge to, without disclosing the secret that this work was expected to be performed for free, and I paid a stiff price for her coyness.

Project Managers know that delivering the customers’ scope on-time, on-budget is the true point of management, and has far, far more weight on overall organizational success than, say, whether or not we should have rented or purchased the copy machine. Asset Managers can’t, or won’t, understand this, because they were subjected to business college professors who sold them on the “maximize shareholder wealth” stuff, including the notion that the return on investment (ROI) is the ultimate evaluator of management decisions. Don’t believe me? Engage in this quick mental exercise: imagine two identically-sized projects. Project A delivers a high-quality scope on-time, under-budget, and the customer is thrilled. The project team is comprised of highly-talented personnel who arrive late, leave early, and consistently fail to observe the company’s dress code, often in dramatic fashion. Project B is late and over budget. However, its Project Team arrives early, works late, doesn’t charge all of the hours they work (i.e., free overtime, conspicuously performed on weekends), and are dressed impeccably. If this happens in your organization, which team is likely to be commended? Disciplined?

You see my point. The Asset Managers have been ensconced in the middle of commonly-accepted business models for so long that they’ve lost sight of the actual point of performing Project Management. It’s kind of a funny coincidence that Generally Accepted Accounting Principles were first propagated at around the same time that Niccolò Machiavelli was writing The Prince. The business model pathologies that cling to the Asset Managers’ strategies like a pair of Tiberian bats have become so reflexive that it’s next to impossible to reason their practitioners out of using them and, by using them, they essentially oppose basic PM tactics.

There are lots of sources of frustration for Project Managers, and some of them are within our own ranks (cough, risk managers, cough). But PM practitioners should feel at liberty to, when confronted with some of the Asset Managers’ goofy strategies, say under their breaths “That’s not the way that works.”

Or, alternately, “This is why we can’t have nice things.”

Some years back I had accepted a job to head up a major program office. It had a huge budget, but no cohesive or even usable cost/schedule performance management system in place, and previous attempts at installing one had failed spectacularly. The fellow I was replacing had a Ph.D. in a technical discipline, but didn’t know a fraction of what he thought he knew about Project Management. The person who hired me asked me to attend a meeting of the Program Controls Office team prior to anyone knowing I was about to succeed this Ph.D., and so I did, but made it a point to sit in the back of the rather large conference room and be as small as I could.

The Ph.D. arrived, and strode up to the white board to draw four red triangles across the top.

“These represent the major milestones this program is on the hook for this fiscal year” he began. “Each of them has several subordinate milestones that are part of its scope” he continued, as he drew a few smaller triangles in a row beneath the first triangles, but in a different color.

“We have to provide status on our progress against these milestones to the customer, but the level two milestones are themselves supported by multiple level three milestones” he said as he drew more triangles along the bottom of the white board, in still yet another color.

“Now, these milestones have some relationships we need to discuss” he continued, in what I perceived to be a somewhat condescending tone. “These level one milestones depend on the level two milestones being completed on-time” he said, as he drew lines between the top-tiered triangles and the mid-level ones. “And the level two depend on the level three” as he drew lines between the mid and lowest-tiered triangles. “In addition, some of these milestones depend on the availability of the same facilities or other specific resources,” which he indicated by drawing lateral lines among the lowest level triangles, but in a different color than the lines indicating the hierarchical scope relationship. “And finally, at each level there are going to be milestones that need to be completed prior to the start of others,” leading to more lines being drawn but, since he had run out of different colors, he re-used the color of the level one milestones (red).

Since I was in the back of the room, I had a good vantage point for evaluating the whiteboard from an overall perspective. It was an incoherent mess, but with the level one milestones being represented in red sitting atop smaller triangles and multiple, crisscrossing lines it kind of looked like an immense serving of spaghetti and meat balls.

“As difficult as it may appear, we have to find a way of capturing the scheduling data for these milestones, and report on them accordingly.”

The room erupted in questions.

“Is there a way of prioritizing the level one milestones?”

“Do all the organizations performing the work on these milestones agree on that prioritization?”

“Are all of the projects involved using Critical Path Methodology software, or even the same package?”

The nature of the questions reassured me that the team I was to inherent in a couple weeks knew their stuff, but I became convinced that their (then) director was clueless.

As the meeting wrapped, an old friend of mine who happened to be on the program controls team invited the Ph.D. to sit with me for a moment to get my input. I had made it a point to not say a thing, but now I was cornered.

“Listen, George (not the Ph.D.’s real name), the relationships you just described are well-known aspects of a schedule baseline. When you say that the level one milestones are dependent on the level twos, and so forth, you are describing a decomposition of scope that’s nominally performed by establishing a Work Breakdown Structure. The description of the need to capture common or shared facilities and resources among the milestones is typically handled via a master resource dictionary. And the need to capture which milestones (he meant activities, but I thought it best to use his lexicon) need to be finished before others can start is quantified using schedule logic. Any CPM software package worth the name will be able to accommodate these parameters, and provide the kind of schedule performance information you are looking for.”

My friend who had put me in the hot seat was gesturing towards me as if to say “What he said!”, but George was having nothing of it.

“We’re already arranging a series of coordinated spreadsheets that list all of these milestones, and the relationships I just discussed. We’ll be able to handle it that way.”

I pretty much knew George would reject anything I had to say, but I felt the need to offer it up for rejection nevertheless. The serving of spaghetti-like diagram that he had drawn on the white board provided clear evidence that he was attempting to deal with PM concepts that he didn’t understand, rendering any attempt to convert those concepts into a workable strategy inchoate at best, and utterly misguided (and therefore wrong) at worst.

On the plus side, the whole episode served as a stark example of how, when authors, paper presenters, and, yes, bloggers push figures or tables that indicate entities, functions, or activities encapsulated in shapes that have lines or arrows in between representing some sort of vaguely-defined relationship, it’s a sure sign that the author has a poor grasp of the concepts underpinning the proposed process or strategy, which in turn means it’s probably going to be, well, wrong.

Another unexpected outcome of the meeting: I had to have Italian food for dinner that evening.

Going back to my Game Theory roots, one of the games that’s often used to model cooperation or defection behavior is Hawk-Dove. Imagine a population of 100 generic birds, who can adopt one of two strategies: the doves forage for food, and consume all they can collect, whereas hawks simply take the food of those doves who are unable to defend it. A fascinating aspect of Hawk-Dove is that the entire population’s payoff is maximized if all of them behave as doves. However, with the introduction of a single hawk, the Nash Equilibrium[i] quickly moves towards a ratio of 25% hawks, 75% dove, even though, again, the entire population’s payoff is maximized if they are 100% dove.

Of course, games like Hawk-Dove are only somewhat analogous to the real-life situations they are intended to model, and it is up to each individual PM to determine just how analogous such insights can be to their particular environs. However, the particular aspects of Hawk-Dove, of the movement of the Nash Equilibrium moving to 25/75 with the introduction of just one hawk, even though the payoff is maximized through 100% dove strategy selection, got me to thinking about a particular binary distinction I’ve often made in this blog about Project Managers, the Processors versus the Performers. In my version of this PM game, the Processors are those PM practitioners who care very much about process, and tend to devote significant amounts of energy and time into ensuring that cost and schedule baselines are properly set up and integrated, that the risk analysis is performed with only the best information available and by recognized experts, that there aren’t too many start-to-start relationships in the Critical Path network, that no more than 5% of the activities have their Earned Value claimed through the Level-of-Effort method … you know the type. But the sure sign that a PM is a Processor is the adherence-to-process behavior appears to (or actually does) take precedence over bringing in the project on-time, on-budget, with all of the scope particulars satisfied. Conversely, Performers devote the lion’s share of their efforts at bringing in the project on-time, on-budget, with all of the scope particulars satisfied, and do not place much emphasis on what others consider to be proper process. Indeed, Performers will often see formal procedures as onerous, and harmful to the attainment of their overall objectives.

Now let’s take a quick diversion over to Organizational Behavior and Performance space. In any meritocracy, the person who shows talent or an aptitude for performing the tasks the organization asks of them will be given tasks with greater levels of responsibility, including leading teams of others so that some level of knowledge/expertise transfer, or training can take place (they don’t call Project Management “the accidental profession” for nothing). In PM space, this typically occurs in one of two ways: the individual actually brings in the work by attracting new clients with their demonstrated level of expertise, or else they are promoted because their abilities are perceived as being advanced by the higher-ups in the organization. In other words, the Performers versus Processors dichotomy is somewhat natural in the way it manifests in the PM world.

Okay, let’s return to Hawk/Dove. In my analogy, the Processors get ahead without actually bringing in projects on-time, on-budget, and yet it seems they are invariably the ones who set policy for the macro organization(s). In this respect, Processors don’t go out and attain credibility by actually being successful PMs – they sort of pilfer street cred by presenting as if they have it, usually through academic qualifications or claimed experience. Conversely, the Performers get ahead by actually being successful managing projects, which often includes knowing which of the organization’s procedures to short-shrift, if not avoid altogether. To continue in the analogy, Performers are “doves,” attaining and collecting successful project completions; whereas, Processors are “hawks,” claiming to have the inside track on how such successes were accomplished in order to convey to others how they ought to, well, perform the function of Project Management. Consistent with the model, organizations are clearly better served if all of their PMs are Performers, since most (if not all) of their projects will come in on-time, on-budget, which is all by itself a source of strength for the Strategic Managers (who are seeking to maximize market share).  However, with the introduction of just one Processor, the Performers are forced into a position of having to defend the way they conduct Project Management in those instances where it deviates from the point of view of the Processors. I have no idea how to even approach the calculation of the Nash Equilibrium in this scenario, since the parameters are so numerous, if they could even be quantified at all. But you see my point – with the introduction of just one “expert” in PM who points to anything other than a stellar record of successfully managing actual projects as the basis for their expert status, and the macro organization will begin to drift away from a true meritocracy and towards a perceived-advanced model for conducting Project Management.

So, Where Do The Giant Magnets Come In?

Several guidance-generating organizations actually seek to attract Processors in order to churn out their highly subjective PM procedures, guides, and handbooks. But for organizations that seek to repel these “birds,” some useful strategies might be gleaned from Wile E. Coyote, the cartoon antagonist to the Road Runner from the Looney Tunes and Merry Melodies series of cartoons. Wile E. Coyote employs many devices and strategies for capturing the Road Runner, but always fails (usually spectacularly), meaning that a review of these strategies might provide an excellent primer for actually repelling the birds.

Now we just need the Processors to eat the bird seed with the iron pellets mixed in…

[i] Technically, a Nash equilibrium is a set of strategies, one for each player, such that no player has incentive to change his or her strategy given what the other players are doing. Retrieved from http://gametheory101.com/courses/game-theory-101/what-is-a-nash-equilibrium/ on February 9, 2019, 13:36 MST.