Game Theory in Management

The saint at the heart of the Santa Claus story is Saint Nicholas, who actually lived in the third and fourth centuries A.D., and was known for his acts of philanthropy (ProjectManagement.com’s theme for December). One of the most widespread stories about him involves a poor man with three daughters of marrying age; however, this man did not have any money for a dowry, needed to secure an adequate husband, and carrying the real threat that they would be sold into slavery. Saint Nicholas became aware of their quandary, and would throw bags of gold coins through the family’s open windows at night (in a bid, no doubt, to remain anonymous), landing them near the fireplace where stockings were hanging to dry. This would become the Christmas tradition of filling stockings hung out on Christmas Eve with goodies to be revealed the following morning.[i]

Meanwhile, Back In The Project Management World…

Saint Nicholas’ story got me to wondering what Project Managers would wish for if they were to have access to a profoundly generous person with near-miraculous powers at their disposal. In creating such a list, I would like to remind GTIM Nation of my theory of the three different types of management, so:

• Asset Managers seek to maximize the return on their assets, with their primary information stream emanating from the General Ledger;
• Strategic Managers pursue the maximizing of market share, and can invoke a variety of strategies to do so (e.g., anything from hostile takeovers to funding an advertising campaign), and
• Project Managers work to complete their projects on-time, on-budget, while meeting all of the customers’ scope requirements.

Whereas “What do managers want?” is an extremely open-ended question, by specifying that we’re looking for what PMs want narrows down the list significantly. With respect to the other management types, my list for PMs includes:

• The organization’s accounting departments abandon the idea that the General Ledger is the source and residence of all management information expressed in currency, and especially reject the notion that the GL can provide insightful analysis on a given project’s cost performance.
• Similarly, PMs could wish for the Strategic Managers to refrain from assuming managerial supremacy, or even superiority, given the fact that any organization that consistently disappoints its customers is going to find acquiring more market share to be very difficult indeed, regardless of the quality of the advertising campaigns.

Then from within the Project Management community we could wish for:

• A return to the 1980s definition of Management Reserve and Contingency (see last week’s blog);
• Those advocating for widespread usage of the technique of re-estimating a project’s remaining costs, and adding this figure to the cumulative actuals as a way of asserting that project’s Estimate At Completion (the so-called “bottoms-up EAC”), have an epiphany (get it?), reverse their anti-Management Science stances, and instead advocate for the calculated version of the EAC.
• If we can’t get the previous bullet, then at least those “experts” should back off the dopey notion that the “bottoms-up” EAC should be time-phased.
• If we can’t get either of the previous two bullets, that these “experts” acknowledge that time-phasing the bottoms-up EAC is the same as creating an alternate cost baseline, reducing the original budget baseline to rubber. Showing my age once again, I can remember when the existence of a rubber baseline on a project was a very bad thing. Now, with time-phased EACs, I guess it’s required.
• If correlation is not causation, and science is all about establishing causal relationships that can be demonstrated and repeated in an experimental setting, how is it that assertions based on statistical correlation keep making their way into the codex of acceptable PM techniques? My next wish is for GTIM Nation at first, and then the PM community at large, to begin to insist that all risk management aficionados subject their analyses to the level of review rigor expected of any other scholastic endeavor, and to respect the results. In short, the future cannot be quantified, and I wish the risk management community would stop pretending to the contrary.
• For perhaps the biggest wish of all, analogous to a seven-year-old suburban-dwelling kid wishing for a pony, I would be made excessively happy if those organizations that present as advancing PM would embrace actual management science and legitimate scholarship, right down to falsifiability and genuine causal analysis.

I think I’ll ask my managing editor, Cameron, to put a GTIM stocking on the PMI® Headquarters’ fireplace mantle on the evening of the 24^th. Assuming Santa fires up ProjectManagement.com from time to time, and reads this blog, he now knows what I want. Who knows? It could happen…

[i] Retrieved from http://www.stnicholascenter.org/pages/who-is-st-nicholas/ on December 8, 2018, 21:22 MST.

Okay, GTIM Nation whippersnappers, listen up. Back when I was first learning the ropes of PM there was a simple elegance about the way the types of budget were established and managed. Once the proposal was won and the negotiations settled, the customer would fund the Contract Budget Base, or CBB (this is now often referred to as Total Project Cost). The CBB was everything – the actual budget, contingency, fee – like I said, everything (but not Over Target Baselines, nor Authorized, Unpriced Work[i]). This figure would then be broken out into its components:

• Unless the contract type was Firm Fixed Price (FFP), the award fee/fixed fee would be set aside.
• If the customer was funding the Contingency Budget, this was usually not included in the CBB; however, if the contractor was expected to fund the Contingency, it was part of the CBB, and had to be set aside. And, just for the record, Contingency was defined simply as “in-scope, uncosted.”
• If parts of the project’s scope were too difficult to quantify precisely enough to generate a detailed cost estimate, a rough order-of-magnitude (ROM) estimate would establish the amount set aside for Undistributed Budget (UB).
• The remainder was available for funding the actual work. This remainder was divided among the project’s Work Packages, which were then rolled up into Control Accounts, and ultimately formed the project’s Performance Measurement Baseline (PMB).

The more seasoned members of Game Theory In Management Nation are probably asking themselves “Hey! What about Management Reserve?” What about it, indeed? For on this very point many a project has met an inglorious end.

Alas, The End Of A PM Era…

In the era I’m talking about, the Management Reserve was established by going to each of the Work Package managers after they had submitted their “final” cost estimates, and asking/telling them to give back a certain percentage (usually around 5%) of their budgets to create the MR. This ask was usually expressed with the acknowledgement that, yes, the WP managers had only agreed to achieve the scope based on receiving their full budget request and, if they needed the 5% back towards the end of their task, no worries. If, however, they could actually attain their scope with a 5% savings, then that amount would become available to other, poorer-performing WP managers, and the overall project would gain a certain level of PM latitude going forward. This intra-project philanthropy functioned very much like society-wide philanthropy, in that it made everyone’s life a bit better. In addition, the use and management of MR was completely invisible to the outside customer, since their reserves had already been established, and this set-aside was explicitly from inside the PMB. Like I said, it was a simple, elegant approach to managing projects, their budgets, and variances.

We Should Have Left Well Enough Alone

But then bad things started happening to the simple, elegant approach. Lots of customers began to view the reserves as some sort of slush fund, where disingenuous contractors could pull monies to cover events that those particular reserves were never intended to address, such as scope creep. I don’t know, maybe the PM community at large brought this upon themselves through a few instances of questionable practices on larger projects, or perhaps some customers became overly meddlesome, or some of both. In any event, the proper use of the reserves got catawampus, with some unfortunate manifestations, the one with perhaps the largest impact being, what happens to variances-at-completion for the tasks at the reporting level of the Work Breakdown Structure?

The way it’s supposed to work is that, when a given Work Package was getting ready to underrun, they could simply notify the PM that they wouldn’t need their MR allocation back, and might even be in a position to push more into the account. Likewise, if a given WP looked like it was going to overrun, and the reason had nothing to do with customer-induced scope creep or a genuine contingency event, they could appeal to the PM for some of the Management Reserve. As long as the underruns/overruns came out roughly equal, all was well.

But when the definition and terms of usage of the reserve accounts were upended, that simple, elegant technique was thrown out. Rather than establish Contingency as a reserve for in-scope, uncosted work, and Management Reserve as a purely internal-to-the-PMB account, these two were redefined based on who funded them. When that happened, it suddenly became legitimate to pressure the PM to give back any underrun that the activities on the reporting level of the Work Breakdown Structure (WBS) had saved, but the overruns – well, the contractor was expected to absorb those, of course. The intra-project philanthropy device was exiled, making everyone’s project a bit more scrutinized.

What Now?

Of course, being the curmudgeon that I am, I think the best remedy would be to go back to the way it was done previously, but the realist in me realizes that’s never gonna happen. Maybe all that’s needed is a movement by the PM community towards an intelligent philanthropy element, starting with the definition of the term Management Reserve.

[i] See https://www.dau.mil/acquipedia/Pages/ArticleDetails.aspx?aid=cbb58d51-c988-4f97-b5a1-b1c59c076887 for an excellent article on these terms.

I don’t know what it is about Project Controls Specialists – the data gatherers, processors, and deliverers of information in the PM world – that their reports are often held in lower regard than, say, the data from the action item listing. It’s strangely reminiscent of the character of Cassandra, from Greek mythology. She received her prophetic capabilities from the god Apollo, in exchange for agreeing to have sex with him; however, she refused him after gaining her prophetic ability. Since Apollo couldn’t take back the gift, he cursed her by arranging for her to never be believed, even though she was always right (funny how she failed to foresee that particular outcome stemming from her refusal). She correctly predicted the fall of Troy, even down to the use of the Trojan Horse, and also knew beforehand the fates of key characters from The Iliad¹. However, true to the curse, nobody believed her.

On a mildly less epic scale, Project Controls Analysts (particularly the ones who read this blog, calculate their Estimates at Completion and eschew risk management techniques) are often noted for their ability to precisely call out which project tasks are doing well, but also which ones are in trouble, and are likely to finish over budget or late. Even though they are almost always provably accurate, the other members of the project team, customers’ reps, or organization’s execs will tend to not believe them until well after it’s too late to circumvent the unfolding disaster. I believe there are several reasons for this, including the idea that there’s always time to reverse the fortunes of poorly-performing tasks, the accountant’s take is always more accurate, or even that young priestesses inhaling volcanic fumes at Delphi provide better insights. I continue to be amazed that, each time I’m asked to perform a forensic analysis on a project that experienced a significant overrun or delay, but nobody seemed to know about the causal factors until it was too late to correct, the Earned Value Management System had correctly predicted which tasks were in trouble, had pegged the range of overrun/delay to within ten points, and had done so months in advance. Every. Single. Time.

As the fates would have it (get it?), I’m confident I have a solution to this ancient curse that the cost/schedule performance assessors have borne since the advent of the Cost/Schedule Control System Criterion, and it does not involve laying siege to the other business analysts’ walled cities for a decade. It is this: the Project Controls Analysts must provide their PMs with the information the PM wants, and in the format they want to see it. Nothing more, nothing less. Here’s why I’m confident this tactic will work.

Back during the days of the aforementioned Cost/Schedule Control Systems Criterion, or C/SCSC, the use of Critical Path and, specifically, Earned Value techniques was a requirement of the organizations performing major project work for the United States Department of Defense. All of those organizations had to “do” Project Management, and in a fairly particular way. Those who performed the role of collecting, processing, and delivering PM information didn’t have to justify anything – the performance of their roles was a condition of the contract. However, in those arenas where doing PM was not required, many organizations (foolishly, in my opinion) took advantage of the opportunity to cut some PM administrative costs, and downsized (or even eliminated) their Project Controls staff, leading into a cycle that repeats the following steps:

• A major project experiences a significant overrun.
• The organization has a re-dedication to PM principals in general, and Cost/Schedule Performance Systems in particular.
• These systems perform as expected, notifying upper management of where completion-threatening problems arise, and allowing them to be avoided.
• As months go by without another major overrun, management begins to question the need for the PM Administrative budget, at the same time as the unenlightened PMs begin to express frustration with having to provide status updates every reporting period.
• The budgets for the EVMS and CPM experts get cut, and those systems cease reporting.
• The organization experiences another major overrun, and the cycle begins again.

In order to disrupt this cycle, the head of the Project Controls organization needs only to supply their PMs with what the PMs want to see. Not only will this tend to keep their organizational costs low, but those PMs who eschew the cost/schedule performance information will have a notably higher incidence of failure, and will drop out of the pool for future project assignments. Those who believe the prophesies take advantage of this irreplaceable information stream will have a higher success rate, and will either continue to receive the preferred projects being brought in, or will be promoted into the higher levels of management where they can continue supporting the Earned Value and Critical Path folks. It’s a simple survival of the fittest, but it won’t work if the EVM/CPM-owning organization attempts to mandate the use of those tools. The PMs are the customers of these organizations, and won’t take kindly to an attitude of having to be educated on how to do their jobs.

In short, successful PMs already know who they should listen to. Unsuccessful ones will have epic poems written about the people they should have heeded.

1. Wikipedia contributors. "Cassandra." Wikipedia, The Free Encyclopedia. Wikipedia, The Free Encyclopedia, 17 Nov. 2018.

Okay, GTIM Nation, it’s time for another payoff grid (yaaaah)! Last week I discussed the problems presented when a project’s cost/schedule performance system was relaying information showing that a project was in trouble, but its principals rejected that insight, or even attempted to cover it up or refute it. Technically speaking, this is one of four possible outcomes from the cost/schedule performance system interacting with management, so:

So, under what circumstances would the cost/schedule performance system indicate that the project is doing fine, but the project team reacts as if it’s in trouble? Well, that depends on the information system.

In past blogs I’ve asserted that all valid management information systems have all of the following attributes:

• The information is timely. Depending on the data cycle, information has a relatively short shelf-life, after which it really can’t be relied on for making decisions of import.
• The information is accurate. Inaccurate information is useless.
• …and, perhaps most important but most elusive, the information has to be relevant.

Why is relevant management information elusive? Consider the following scenario. Midway through a medium-sized project, the Earned Value performance indicators Cost Performance Index and Schedule Performance Index (cumulative) are sitting at 1.02 and 1.05, respectively, meaning that, at the current rate of performance, you will wrap up under-budget and early. However, your friend the accountant, and that risk management fellow, are insisting on interrupting your project review meeting.

“Look you here, Ms. Project Manager!” the accountant begins, “you are spending more than your cumulative time-phased budget. At this burn rate, your project will overrun!”

The risk manager chimes in. “Also, the risk analysis indicates a 25% chance of weather interfering with some of the soon-to-start Control Accounts, which add up to $150,000 (USD – the RM wouldn’t actually say “USD,” that’s just for the international readers). Unless you have $32,500 in ready contingency, you will overrun, based on this analysis.”

You look over at your Project Controls Analyst.

“Jay, how confident are you in your figures?”

“You’ve got more than 90% of your tasks’ budget using one of the quantifiable EV methods, and the actuals have been verified. I’m very confident.”

The accountant speaks up.

“I’m a CPA, and the risk manager happens to be a recognized expert in the field. Whom are you going to believe?”

Whom, indeed?

Here’s the problem with those pushing management information streams that are essentially irrelevant: if they err in saying that a problem is nigh, and needs to be addressed, but that “problem” never actually materializes, there is rarely a downside to such alarmism. In fact, such ones as our fictional accountant and risk manager would be in a position to assert that it was their sounding-of-the-alarm that helped prevent the negative future predicted from unfolding. Conversely, should something really go south on the project, but their systems failed to provide any kind of an advanced warning, then not only would their systems’ vulnerabilities be laid bare, they, themselves, might very well have a harder time convincing the next PM (certainly the current PM) that the data they bring to the project review meetings is, well, relevant. If we assume that accountants weighing in on project cost performance or risk managers weighing in on, well, anything (a caveat: after the original risk analysis is performed, and contingency plans and budgets established), then these purveyors of marginally relevant information would be crazy to not constantly beat the sky-is-falling drum.

On the other hand, I have never – never – witnessed a properly functioning Earned Value Management System – even a very simple one – fail to give fewer than three reporting cycles’ worth of warning when any activity or task (with a duration of at least six reporting cycles) at the reporting level was headed for significant (>20%) overruns or delays.

So, whom are you going to believe? One more clue: the first party to point to their education level or depth of experience as a reason their arguments should be considered valid, as happened in our little story, is directly signaling that they have lost the relevancy argument on its merits.

They’re also signaling that you would be crazy to believe them.

I believe that we all overly dependent on our experiences. We become familiar with how certain scenarios unfold, and tend to expect similar results from analogous situations whenever we encounter them. Of course, each time these scenarios proceed as expected, it reinforces the notion that our formulaic expectations will be realized at the next iteration. I’ve written previously on how our experiences can become our worst enemy when it comes to selecting the optimal Project Management strategy for resolving a given problem, but what I’d like to address now is the phenomenon where we stick to pre-selected strategies even in the face of evidence that it’s a mistake to do so.

I think it’s fascinating how this effect influences management decisions and business analysis (ProjectManagement.com’s theme for November), because I’ve witnessed countless times managers making poor decisions in the light of evidence that their decisions are ill-advised. But probably the best dramatic example of this effect is beamed into households around the world here in Holiday Season, that example being George Baily’s behavior in the film It’s A Wonderful Life (1946). Briefly, George manages a small Savings and Loan (or Building and Loan) in the town of Bedford Falls, but his uncle Billy misplaces an $8,000 deposit to the local bank, which is run by the antagonist, Mr. Potter. As George contemplates suicide, his guardian angel appears and arranges for George to experience Bedford Falls as if George had never been born, in an attempt to get him to see how his life has had a positive influence on so many others’.

So what we have in this movie is George Baily experiencing an altered reality, but he spends almost the entire time rejecting his new circumstances, even in light of the following:

• By my count, Clarence Odbody, George’s guardian angel, tells George no fewer than twenty-seven times that the world around him has been changed to reflect what it would have been like had George never been born. Rather than accept Clarence’s information at face value, George constantly explains the things he is seeing and hearing as coming from different causal factors, including:
  • Clarence is crazy,
  • Clarence is playing some elaborate prank on George,
  • George drank some “bad liquor,” or something,
  • Everyone else is crazy,
  • Or George has gone crazy.
• The clues that Clarence should have been taken at face value far earlier in the movie mount quickly. These include:
  • George’s daughter Zuzu wanted George to fix her flower, which had shed some petals. George surreptitiously hides the petals in his pocket, and pretends to have fixed the flower. When George is transported to the George-less world, the petals have disappeared from his pocket.
  • In what has to be a blow to those who do not accept anthropomorphic global warming, it doesn’t snow in Bedford Falls on Christmas Eve if George is never born.
  • His acquaintances do not recognize him, and his home is obviously derelict.
  • The landscape around Bedford Falls has changed, since the Building and Loan was not in existence to build homes for its families.
• These and other clues point to the, frankly, inescapable conclusion that Clarence is right, and George is experiencing the world as if he had never been born. Even so, George feels compelled to track down his wife, Mary, to see what’s become of her. Still not completely accepting the alternate-world explanation, he confronts Mary, telling her that he’s her husband. She faints.

Meanwhile, Back In The Project Management World…

The well-known study in Cost Performance Index stability[i], performed by Captain Scott Heise, with attribution to Major David Christensen, represented, in my opinion, a significant event in Project Management Information system efficacy. Probably the most significant inference that came from that research is that, since a project’s Cost Performance Index (CPI) is fairly stable relatively early in its life cycle, then the common Estimate at Completion (EAC) formula of dividing the Budget at Completion (BAC) by that very same CPI will yield an EAC that’s reliably accurate to within 10 points.

However, even in light of this well-done and clearly relevant study’s results, there are many George Baily PMs out there who refuse to accept its implications, that the easily-calculated EAC is reliably accurate. They’ll see that the calculated EAC is indicating a future overrun for their Control Accounts (or even the project as a whole), but will convince themselves that they can correct the negative variance prior to project’s end, even going so far as to generate the so-called “bottoms-up” EAC to indicate that there’s no real problem, or it’s not as big as the despised calculated EAC is showing. I’ve been told of executives who order their PMs to perform this very tactic when the calculated EAC is an embarrassment to them. Cost performance report after cost performance report, project after project, and these PMs won’t accept the evidence in front of their eyes.

Just as George Baily’s ordeal would have been much shorter had he simply accepted Clarence’s explanations earlier in the movie, and proceeded directly to understanding how dreadful the lives of his family, friends, and associates would have been without him, I believe that many managers in the PM world would be far more successful if they were to rely on the calculated EACs, and deal with the indicated overruns more quickly and directly than attempting to deny or minimize their implications (or accusing the calculated EAC aficionados of being crazy).

But then, had George accepted that Zuzu’s missing petals constituted ipso facto evidence of the validity of Clarence’s assertions, It’s A Wonderful Life would have been a much shorter movie.

[i]Heise, Scott,  A Review of Cost Performance Index Stability, September 1991, retrieved from http://www.dtic.mil/dtic/tr/fulltext/u2/a246621.pdf  on September 12, 2018, 14:17 MST.