Game Theory in Management

In last week’s blog I discussed the difference between the structure of valid management information systems, and one of the versions of invalid systems, which are essentially polls. This week I want to discuss the middle ground between light and shadow, between science and superstition, and it lies between the pit of man's fears and the summit of his knowledge. This is the dimension of imagination. It is an area which we call – no, not the Twilight Zone! Whoddaythink I am, Rod Serling? It is an area (cue the creepy guitar music) I like to call…

The Management Illusion Zone.

You’re about to meet a project controls analyst, one Mr. Sam Earnest. His project is entering a critical phase, where just one misstep could lead to disaster. He provides one of two competing management information streams, the valid one which could his project to success; the other, a risk management baseline and analysis. Mr. Earnest is about to enter… the Management Illusion Zone.

Scene I: setting, Dirk Bailey’s office.

(Knock, knock)

Bailey: Come in!

Enter: Sam Earnest, Project Controls Manager (who looks suspiciously like William Shatner), and Wanda Devlin, Risk Manager (who bears a strong resemblance to Julie Newmar).

Bailey: Do you two have your monthly reports ready?

Both: Yes.

Bailey: We’ll begin with you, Sam.

Sam: On your Cost Performance Report, overall you’re okay, but there’s this one task that’s showing both a negative cost and schedule variance. I’ve drilled down through the WBS, and the specific task having trouble is the Finalize Design activity.

Wanda: I performed a risk evaluation of the entire baseline, that task included, and my report indicates that it’s perfectly fine – very little risk, if at all.

Sam: This isn’t about whether or not anybody recognized that it was risky beforehand. This is about what a negative variance will do to the project right now. That activity is on the critical path, with at least a dozen other activities not being able to start until its completion.

Bailey: I’m afraid I’m with Wanda on this one, Sam. What could possibly go wrong with getting the final approval of the design? It passed its preliminary reviews with flying colors.

Sam: I’m not sure, but I’ve heard rumors that its Work Package Manager – Henry Bemis (who, incidentally, looks a lot like the late Burgess Meredith) – spends so much time reading requirements that he isn’t connecting with the right people to get final design approved.

Wanda: The odds of that happening are approximately 700 – to – 1. I’m telling you, I’m looking at the exact same task, and there’s no problem there.

Sam (becoming visibly agitated): And I’m telling you there is! If you guys don’t do something, it’ll wreck the whole project!

Wanda: Sam, relax. We’re all on board the same project team. By the way – how are you coming along with your recovery? Didn’t you have some sort of breakdown six months ago?

Sam (caught off guard): Well, yes, Wanda, thanks so much for bringing that up. I’m getting along fine. Now, back to this variance…

Bailey: It simply defies common sense that something so trivial as getting a final design authorization could endanger the whole project, Sam. It’s really just a formality.

Sam: Formality or not, if we don’t get it squared away, we’re looking at blowing past the project’s early payment milestones.

Wanda: The odds of that happening are…

Sam: How can you possibly compute that?

Wanda: It’s from the Monte Carlo analysis of the baseline.

Sam (becoming more agitated): It makes no difference what kind of formulaic approach you took, if the feeder data represents unquantifiable events! What did you put down as the odds that an internally-approved design wouldn’t get final external authorization until after the scheduled milestone?

Wanda (looking over her reports): We gave that the typical administrative-type odds, 15%.

Sam (almost hysterical from frustration, now): But don’t you see, that’s exactly what happened!

Bailey: Sam, you’re a little too involved here. I think it would be best if you were to be removed from the project, and go someplace where you can rest. Wanda can generate all the management information I need until you’re calmer, more comfortable.

At this, two men dressed in scrubs come in, and take Sam away in a straight jacket.

Narrator: The work of Mr. Sam Earnest has ended now, reporting not only from task through to project, but also from the fear of recurring project breakdown. Mr. Earnest has that fear still... though, for the moment, he is alone in this concern. Happily, his conviction will not remain isolated too much longer, for the project does end up late and overrun, tangible manifestation of the dangers of relying on contemporary risk analysis techniques, even from so intangible a quarter as the Management Illusion Zone.

As ProjectManagement.com transitions from its October theme – leadership – into its November theme – data management – I can’t help but notice that the two have something in common: they fail utterly when they become dependent on polls.

British Prime Minister extraordinaire Margaret Thatcher once said “To me, consensus seems to be the process of abandoning all beliefs, principles, values and policies. So it is something in which no one believes and to which no one objects.” She could not have been more prescient, for if the role of leader could be readily accomplished through a pure democracy, then Rome would have never conquered Athens.

But what, pray tell, does this have to do with data management? Well, a few things. Recall my oft-asserted proposition that the bottom 20% of managers who have access to 80% of the information they need to obviate a given decision will always out-perform the top 80^th percentile of managers who have access to only 20% of the information so needed. Now, here’s the second piece of this central puzzle.

All valid management information systems are predicated on this basic structure:

·         Raw data is gathered based on some type of discipline.

·         It is then processed into usable information based on some type of methodology (for we PM-types, this is almost always Critical Path or Earned Value, but that’s a different discussion).

·         This information is then made available to decision-makers in both a timely and intuitive fashion, so that it’s readily actionable.

Graphically, these three steps can be represented as three sequential events.

Conversely, invalid management information systems will often assume the structure of a spider. They appear as a central data repository, surrounded by input/output nodes – essentially, polls, masquerading as legitimate information systems. These are often named action item lists, or milestone performance systems. But make no mistake – they are structurally invalid, and any usable information that comes from them is purely by coincidence, the way a broken clock is accurate twice per day.

These polls are as damaging to leadership as they are to informed management decision-making, if not more so. Leadership based on consensus at least has the quality of representing the desires of the followers, though one has to wonder if, reduced to being directed by polls, Patton’s Third Army would have stormed across the Rhine or have gone back to France for a little extended R&R. On the management information side, though, the backward effects can be even more insidious, since the poll-based systems invariably displace their legitimate cousins. After all, why go through all the trouble of setting up a Critical Path Methodology-capable baseline for a project’s schedule, when taking a survey of the task managers on whether or not they felt they would achieve the milestones they were responsible for can do just as well?

Let’s now return to the desk of the Prime Minister, and that of the Project Manager. The politician has access to opinion polls, the manager can tap into a so-called management information system that is, in actuality, a compilation of the opinions of the team members. In what way are these two decision-makers different? Are they not both presented with a basic choice, of simply selecting the path that the consensus indicates, or else choosing to implement a specific approach to solving the problem(s) before them, regardless of the opinions of those who are – let’s face it – not in a position of leadership?

Next week, in Part II, we’ll discuss feed-forward management information systems, and the level of chicanery that comes in to play with them.

It was yet another cold and stormy night. I was sitting at my desk, reading the etching on the glass of my office door, rotagitsevnI etavirP, yrrebpsaR ylnatS, when a very sinister shadow came across it (the door, not necessarily my name on it).

The knob slowly turned, finally unlatching, and the door’s hinges creaked as light flooded in from the hallway. Backlit in the door was a tall figure, raincoat collar turned up to the point that it almost touched his low-slung fedora.

“You Raspberry?”

I put my head on my left hand’s fingertips in exasperation.

“Why does every person who walks through that door ask the same bloody question? Who’s name in on the door you just walked through?”

“Raspberry.”

“And who did you come here to see?”

“Raspberry.”

“And, according to the building’s directory, who occupies this office?”

“Raspberry.”

“Given all that, who do you think I am?”

“I don’t know. That’s why I asked. You Raspberry?”

I knew two things instantly. One, this was not a hit man, at least not a very good one. Two, this guy was not street-wise at all. He must have been sent by my old nemesis, Monolithic Corporation.

“Yeah, I’m Raspberry. Waddayaneed?”

The stranger looked behind and around him prior to closing the door.

“I represent a client who runs a major corporation. Some of our lower-level managers have been disappearing suddenly. At the same time, our market share…”

At this his voice trailed off, as if he were suddenly aware that he was about to divulge something he shouldn’t.

“We shouldn’t have this level of competition” the stranger restarted, with sudden vigor. “We’re the biggest game in town, and these little pipsqueeks…”

Again his voice trailed off, for the same reason. Although I knew which corporation he referenced, I wanted to hear him say it.

“Which corporation do you represent?”

“Why do you need to know that?”

“So I know where to not start looking.”

“I’m not going to tell you that. But I can tell you that you should start looking over at Acme. Here’s a list of names. Good luck.”

“Aren’t you forgetting something?”

“Of course – your retainer.”

The stranger scribbled out a check, tore it out, and pushed it across the desk.

“Alright, then. I never thought I’d do any business for Monolithic, but I’ll get to the bottom of your disappearances.”

“How did you know I was with Monolithic?”

“You just handed me a check.”

“Yeah, so?”

“Just a hunch.”

* * * *

I came across some old friends as I pulled into the parking lot at Acme.

“Stanly! How the heck are you?”

“I’m great, Marcus. You sure seem happier since the last time I saw you.”

“Yeah, you know what, I finally got out of that mid-level management position at Monolithic, and took a team lead job here at Acme. I took a cut in pay, but I’m much happier.”

“How so?”

“The short answer is, if you don't recognize merit, the meritorious start to disappear. They were so big, though, that the bad leadership calls will take a while to catch up to them. Here at Acme, if you can deliver, you can move up. We’re a fraction of Monolithic’s infrastructure, but I think we’re going to make some serious inroads into their customer base.”

“You may already have.”

* * * *

I met the Monolithic guy back at my office.

“I still can’t figure out how you realized I was from Monolithic!” he began. “I know I didn’t give myself away! How did you do it?”

“You gave me a check, remember?”

“Yeah, so?”

“Are you, by any chance, a senior-level manager?”

“How did you know that?”

“’Nother hunch.”

A lot of what’s written about leadership has to do with consideration for the followers, communications with stakeholders, maintaining a high level of integrity, etcetera, etcetera. I would like to cut straight to the heart of the matter: where, exactly, are we being led?

Have you ever had the experience of following someone who had no idea where they were going? For long-distance runners in a competition on a poorly-laid-out course, this can be especially frustrating. If the appointed leaders make a wrong turn, it can add a lot of distance to an already difficult run, meaning that those who would otherwise be in the forefront can suddenly find themselves significantly – if not prohibitively – behind where they thought they were. In short, when we’re talking about leadership, one (literal) misstep can lead to dramatic and frustrating failure.

It’s true in running, and it’s also true of project management. The problem with managerial leadership is that the appropriate course is almost never known (or even knowable) beforehand. The team simply has to tackle the scope in front of them, really without a clear picture of where they will end up, since we’re no longer talking about a geographic destination. We’re talking about an anticipated end-state that takes place in the future, and the future is notoriously difficult to map out.

(Just as an aside, “the undiscovered country” is NOT the future, at least not as referenced in Hamlet, Act III, scene 1. As Hamlet himself uses the term in his famous soliloquy, it actually refers to the afterlife – which, on second thought, might have been the real intent when Councilor Gorkon quotes it in the Star Trek movie of the same name. Perhaps it is better rendered in “the original Klingon.”)

Look, I’m not saying it’s fair, or easy; but the truth of the matter is that followers have an expectation that leaders can see into the future, at least to some extent, and have selected the most appropriate strategy to deliver them to this desired end-state. This is why leaders of all stripes are so often attracted to soothsayers, fortune-tellers, or others who claim to be able to peer into the future and return the results that will be realized should the asker continue on the chosen course of action. On the managerial side, these people are known as risk managers (Ahh, c’mon! It’s been weeks since I took a swipe at those guys!). All these so-called analysts have to do is to shoe-horn in some statistical jargon into their predictions, and they automatically assume the aura of insightful, must-have information generators. Consider:

Palm reader: “I sense a dark force, one that you can’t stop, damaging your path going forward.”

Risk analyst: “If the weather closes in, you will have a 15% lower chance of completing those tasks on-time.”

Crystal ball-gazer: “I see a tall, thin man coming into your life, and changing many aspects of your comfortable routine.”

Risk analyst: “Did you see that last Baseline Change Proposal?  I’ll have to completely re-do my ‘risk baseline’ if this goes through.”

It’s my contention that true leaders spend less time attempting to quantify the future than they do making their teams more robust and better prepared to deal with the unexpected events that inevitably find their way onto – and obstructing – the path to success. That’s not to say true leaders don’t plan – on the contrary, they make it a point to plan, and do so to the best of their ability. But it’s like Eisenhower said : “In preparing for battle, I have always found that plans are useless, but planning is indispensable.”

And so it is in project management. To have a clear view of the project team’s ultimate destination , and the best way of attaining it, is one of the most – if not the most – crucial aspect of managerial leadership.

That, and knowing when Klingons are mis-interpreting Shakespeare…

For my next installment on October’s theme, leadership, I’m going to remind my readers of the blindingly obvious: contentious project teams do not perform nearly as well as harmonious ones. On those occasions where your team does not steadily progress through the clichéd stages of Forming-Storming-Norming-Performing, what strategies are available to combat project team infighting?

For this analysis I’d like to turn to one of the “games” – actually, an analogous set of stated interactive parameters – that I referenced in the book that this blog is named after, Game Theory in Management, titled Hawk/Dove. As the name implies, this game involves two types of birds, or, more specifically, fairly generic birds that can choose a passive strategy (“Dove”), or an aggressive one (“Hawk”).

In all variants of the Hawk/Dove game, the overall population’s payout is maximized if every bird selects a Dove strategy, i.e., they each forage for food, and consume or keep all that they gather. However, if the available food supply is less than can support the population of birds, then the only way for a given bird to survive would be to adopt a more aggressive Hawk strategy, one where the survivor either takes the food other birds have collected, or else stops them from foraging in the first place.

Even in those instances where there is more than enough food to support the population, the introduction of just one Hawk – or one bird that consistently selects the Hawk strategy – will result in a change of behavior in the entire population. Collectively they will arrive at what’s known in Game Theory as the Nash Equilibrium, which is that point at which there is no advantage to be gained on the part of any one player by altering their strategy or strategies. In the 100-bird variant of Hawk/Dove, the Nash Equilibrium is 25%/75% Hawk-Dove, meaning that either 25% of the birds act like Hawks all the time, or else all of the birds act like Hawks one-quarter of the time.

I can hear y’all now – “Hey, Michael, enough with the birds! My project team members are giving each other the bird! You said you had a simple trick…”  Okay, okay, back to the project team.

Your team’s members may not be hurting for food (or a paycheck), but promotions, raises, bonuses, and other forms of recognition are generally more scarce, scarce enough to induce some aggressive behavior in certain team members. The aggressive strategies do not include stealing other team members’ lunches out of the refrigerator (on second thought, though, they might); rather, they involve attempts to change management perceptions. This strategy involves four tactics:

·         Maximize the perceived value of the contributions of the instigator,

·         Minimize the perceived impact of the errors of the instigator,

·         Maximize the impact of the mistakes of the target(s), and

·         Minimize the value of the contributions of the target(s).

What do these four tactics all have in common? They require the instigator to communicate with the PM. So, what’s the simple trick that defeats them all simultaneously?

Forbid ex parte conversations.

Any time any team member wants to talk to you about how they see the performance of another member of the team, interrupt them, and arrange for the other person to be in the room prior to continuance of the conversation. It’s pretty automatic: if the instigator has a real problem, he won’t mind airing it out in front of the person perceived to be the cause. If, however, he’s simply trying to cast himself in a better light at the expense of an innocent comrade, he will immediately realize the game is over, and refuse to proceed. It’s not a silver bullet, but by not allowing ex parte conversations, your team will quickly learn to trust you, trust each other, and should begin performing better.

Identifying the person stealing the lunches? That’s a different problem…