Game Theory in Management

After excoriating the risk management types over the past couple of blog entries (they deserved it!) for asserting the efficacy of their techniques far, far beyond their legitimate limits, I’m now ready to do the same thing for a group of management information custodians who have dominated the best-decision-in-any-given-situation dialogue for hundreds of  years, and yet shouldn’t. That’s right – I’m talking about our friends, the accountants.

What we now know as double-entry bookkeeping and accounting first came into existence along about the time of Machiavelli (coincidence?), and has largely monopolized the executives’ information stream ever since.  So prevalent is the suspender-clad ones’ take on the management universe that leaders who have no idea how to read a profit-and-loss statement will make regular reference to the “bottom line.” For my readers who have had occasion to (or even regularly) attend high-level meetings within your organization, let me ask you this: have you ever seen your company’s top executive cede ultimate veto authority over any course of action – no matter how subtly – to the chief financial officer? I have, and I find it appalling. The accountants have successfully manipulated the management information debate into a narrative where the general ledger is the ultimate arbiter of information for high-level decision-making, a sort-of oracle at Delphi dressed in clip-on suspender-supported togas, and I think it stinks.

What kind of information can the general ledger provide? The set of valid general ledger-based management information consists of:

·         How the assets are doing.

…and that’s it.

Of course, how the assets are doing is the basis for such decisions as how much your organization pays in taxes, how much it owes or is owed, how much profit can be claimed, etc., etc. But it is undeniable that data pertaining to assets is the limit to general ledger information. Naturally, your organization’s CFO team will never, ever admit this, but it is undeniably true. Of the information streams they claim to be able to provide, but can’t, the more prominent include:

·         Project cost performance (their pretending to the contrary is one of the more egregious management science frauds perpetrated on board rooms everywhere)

·         Schedule performance (making a list of objectives with associated dates and person(s) responsible is to schedule management what alchemy is to nuclear physics)

·         Estimates at Completion (EACs)

·         Return on Investment!

Yes, I know that last one will make even the most timid CPA sputter with rage. After all, the claim to be able to calculate that variable on any particular asset (or group of assets) serves as the underpinnings for almost all of their subsequent assertions that they can generate the brass ring of information tidbits, the ultimate definer of good business decisions, or bad ones. And it’s all completely wrong.

The basic formula for calculating ROI is:

ROI = (Gain From Investment – Cost of Investment) / Cost of Investment

so that any asset’s ROI greater than 1.00 is assumed to be a good investment decision, appropriate for pursuing.

Do I have to say it? This is sophistry of Cecil B. DeMill proportions. Except in extremely rare cases, the Gain from Investment (better stated as the anticipated Gain from Investment) variable is impossible to calculate. What was the Gain from Investment on the Titanic’s lifeboats? They were assets to the White Star Line, were they not? Why can’t the suspender-and-bow-tie-clad ones calculate it, then? Because if the Titanic doesn’t sink, the lifeboats have zero value, or even negative value if you take the character Cal Hockley’s assertion from the movie that they were a “waste of deck space on board an unsinkable ship.” However, since the Titanic did sink, then the existing lifeboats were literally invaluable.

And yet, how many business decisions are based on this ROI calculation? Even the Project Management Institute® published books defending the need for creating a Project Management Office (PMO) based on … its Return on Investment! 

How many other allegedly useful information streams are out there that are, in fact, either stretched way past their nominal limits of efficacy, or are out-and-out fraudulent? To find out, you can either stay tuned to this blog, or, if you are impatient, then order my recently-released, must-have book, Game Theory in Management.

In my previous post, I put forth a few reasons why I believe that large parts of risk management theory, as currently practiced, are of relatively little (or even no) value, and the time, effort, and resources devoted to its analysis and output are a complete waste. To completely bury any remaining opposing arguments, I want to delve a little bit into the realm of management information system epistemology – but don’t worry, I won’t get all fancy-smanshy in making my points. After all, that’s the approach the risk managers take all the time, and it makes me crazy.

As I discuss at length in my recently-released, must-have book, Game Theory In Management, while information is the life blood of the organization, not all information is valuable. In order for a given piece of information to have value, it must possess three characteristics:

·         It must be accurate. Inaccurate information is worse than useless – it’s actually misleading.

·         It must be timely. Among competitors, the one with the most timely information will win every match.

·         Finally, valuable information must be relevant.

Okay, Hatfield, you say, the first two bullets are measurable. But how does one know which information streams are relevant, and which aren’t? To make this assessment for any given information stream, you are going to have to purchase my previously-mentioned, recently-released, must-have book. For this blog, though, we only have to evaluate the risk management-flavored information stream.

Management information – especially actionable management information – isn’t free. It requires time and effort to collect raw data, process it into info, and deliver in a readily-understandable format. For project managers, one of the most valuable analyses available is the output from the critical path methodology. Let’s say you are the PM for a construction project, and you have an estimate from your concrete pourers that it will take 20 days for them to finish pouring your building’s foundation. Since the framers can’t start until the foundation is done, you have scheduled them to arrive on-site on day 21. After day 10, however, your scheduler has learned that the concrete pourers are only 25% done. A quick calculation from your CPM-capable software reveals that, at this rate of performance, the foundation will not be finished until day 40. If the framers don’t get a call telling them to not arrive until day 41, they will spend 20 days standing around, unable to do anything other than incur actual costs. Clearly this information is highly relevant, since its unavailability would have lead directly to negative cost variances.

Can risk management methodologies make the same claim? Before we evaluate RM techniques for relevance, let’s do a quick check to see if they meet the other two requirements put forth for information value. Is, say, a Monte Carlo cost and schedule contingency analysis timely? I suppose it can be, so I’ll pass it on that count. Is it accurate? It may or may not be, and this is where a particularly insipid piece of MIS legerdemain is allowed to creep in. If the Monte Carlo analysis happened to quantify a potential scenario that actually unfolded, then the risk managers claim victory. What about all of the potential scenarios that didn’t come about? Shouldn’t they represent system failures? And, if something happens that wasn’t captured in any of the risk managers’ analysis, they just claim it was an “unknown unknown.” Convenient, huh?

Finally, the relevance evaluation. Let’s say in the previously discussed construction project that your customer insisted on a robust risk management program, but didn’t care about critical path scheduling. Since you had access to only one project management analyst, you assigned her to perform a risk analysis. Your framers show up on day 21, but the foundation is only half complete. Your analyst comes to you with one of the following two reports. Which one is relevant?

·         “I know you wanted a complete risk analysis by now, but I’m having trouble with the software. However, I did have a feeling that something like this might happen.”

·         “According to the completed risk analysis, we projected a 34% chance that this would happen, and that the impact would be the cost of the framers having to stand around for 20 days, or $100,000. So, we put $34,000 into the contingency reserve, since that’s $100,000 times 34%.”

If you said that neither of these reports is relevant, go to the head of the class.

And, once you are standing at the head of the class, inform your risk manager that his information is only tangentially accurate, completely irrelevant, and that he needs to go away and stop pestering you. Next up: when to tell your accountant to sit down and shush.

The incomparable Walter E. Williams has written several prescient columns on the topic of the idiocy of raising the minimum wage, and he has furthered several interesting arguments in his criticisms. I think one of his most effective points is that if the minimum wage is $7.25 (USD) per hour (which it is), but any given (usually unskilled) employee is incapable of generating that amount in wealth, then no organization that seeks to turn a profit would ever hire or retain such a one.

According to Salary.com’s Salary Wizard page, risk managers typically earn between $72,033 and $132,106 (USD) per year. Forgive me for being imprudent, but I have to ask:

What does that buy you?

Obviously risk managers do not create that amount in wealth. So what, exactly, do they produce? They produce information. What kind of information? Ahh, there’s the rub. In order for risk managers to lay claim to justifying their salaries (not to mention their offices, computers, overhead accounts, and colleagues having to put up with their statistical jargon-based haranguing), they would have to show that the information stream they generate leads to, if not the realization of that amount of economic wealth, then at least the avoidance of being hit with fines or overruns equal to or exceeding that amount.

Okay, so what kind of information, specifically, do risk management-types provide? (I promise to stop writing as if some unseen person is asking me questions.) Typically, Decision-Tree or Monte Carlo analyses are based on a review of the existing scope, cost, and schedule baselines for a given project. The point of the review is to try to articulate and estimate alternate scenarios to the project execution plan documented in the baselines, and then to estimate the impacts of those scenarios should they come to pass. This fails on several levels:

·         Unless the projected alternate scenarios alter the project team’s response to events unfolding in that manner, they remain nothing but rank speculation. So what if the alternate scenario came about? Other than “I told you so” bragging rights, nothing has improved from the project managers’ point of view.

·         The possibility that something – anything – could go wrong on a project is the typical PM’s ubiquitous consideration. The quantification of this consideration is highly subjective, and adds nothing to the decision-makers’ information base.

·         Consider the risk managers’ own terminology. Events are either “known-unknown,” or “unknown-unknown.” What’s the difference? The former were documented in the risk managers’ analysis, the latter were not. It’s an inherent get-out-of-stupid-analysis-jail-free card. Did we predict it, and it came about? We’re geniuses! Did we predict it, but it didn’t come about? Well, it’s a good thing we at least thought about it! Did we fail to predict it? It was inherently unpredictable!

Now, I understand this sort of pseudo-management science has a vast following, its own ISO standard, a revered place in the PMBOK Guide®, and is entrenched in many guides on how to conduct project management. But a lot of law enforcement bodies employ psychics on major cases, especially when they’re desperate, but that doesn’t make psychics a valid avenue of forensic crime investigation. “Pet Rocks” created quite a sensation in their day, but that doesn’t make common rocks an appropriate object of our attentions or affections.

Risk managers are particularly useless in information technology projects. Will technology change in the middle of your IT project? Probably. Will the scope change, in ways both subtle and profound? Undoubtedly. That’s why Agile and Scrum were developed in the first place. Has your risk manager ever fed you information that tangibly helps mitigate, or even ameliorate, these happenstances? The successful project manager is prepared to respond in a robust manner to the unexpected events that unfold in the course of pursuing her project’s scope requirements. Attempting to quantify the odds and cost/schedule impacts of those things that may or may not go wrong is nothing more than institutional worrying, described in Gaussian terms.

I’m not even close to exhausting my criticisms of the risk management field, as currently practiced. A more complete discussion is contained in my recently-released, must-have book Game Theory in Management (Gower Publishing, 2012). For now, the simple fact remains that the future cannot be quantified. Anyone who claims to the contrary is either engaged in deceit, or in quackery. For that reason alone, you should fire your risk manager.

The other day I made the mistake of surfing over to a popular search engine/ news web site' business page, and read an article about the dangers of using some of the trendy clichés that have become prevalent in the vernacular of the modern office. As with virtually all of the articles that deal with cliché usage, this one admonished readers to avoid them, but not because their usage may lead to misunderstandings, and not because their usage is a clear indicator of lazy or fuzzy thinking. No, this article wanted people to avoid using trendy techno-managerial clichés because – wait for it! – you might not come off well in a job interview.

I have a couple of problems with this, and they both point to a troubling trend in articles that are putatively about how to manage better. My first problem is with the original subject line of the piece. From a writer’s point of view, taking on excessive cliché usage is (ironically) in and of itself a cliché.  But, for all the nose-upturning towards the linguistic hoi polloi out there repeating pat phrases ad nauseum, the simple fact of the matter is that clichés can convey specific meanings, often in ways far more efficient than their more verbose, complete cousins. They only become useless (and then straight on to highly irksome) when overuse leads to the blurring of their original reference or meaning.

Take (as this writer did) “It is what it is.” Bill O’Reilly himself sneered at this phrase, on his top-rated cable news analysis show, no less. Consider what I believe to be the more complete definition of this phrase:

The (person, concept, asset, thing) we are discussing is not changing, and, if we are counting on it doing so, then we will fail. We had better evaluate our future alternatives in light of this unchanging parameter.

Hmmmm. Five words versus thirty-seven. If for no other reason than sheer expediency, one would believe that “It is what it is” would not receive such sneering condemnation. Now, I must admit that one of my friends, a brilliant and capable project manager, tends to use this phrase often. The problem with this fellow is that, genius that he is, he was raised on a farm/ranch in Oklahoma, and he has this Oklahoma accent going that makes him sound like he fell off of the grunion truck yesterday (you thought I was going to say “turnip truck,” didn’t you? Well, I can’t, because that would be a cliché, and might anger certain business writers!). I and some of his other friends recommended he use the Latin version, Id eccum id ecca, in an attempt to sound more sophisticated. Alas, not even Latin spoken in an Oklahoman accent sounds sophisticated.

Another undeserving target is the phrase “going (gone) viral.” Allow me to take another stab at this phrase’s more complete meaning:

The (person, concept, asset, thing) we are discussing has received more exposure, attention, and acceptance than anyone could have expected beforehand, based on its relative merits. It is now so common as to be nearly ubiquitous.

An even more complete description of the phrase would have to include a brief discussion of Metcalfe’s Law. As I discuss in my recently-release, must-have book Game Theory in Management (Gower Publishing, 2012), Metcalfe’s Law deals with a characteristic of networks also known as the Butterfly Effect. In essence, Metcalfe’s Law posits that the power of networks grows exponentially as the number of connections between its nodes increases, but this also increases the network’s vulnerability to cascading events, or episodes of small changes to a limited number of network nodes leading to large, or even catastrophic impacts to large segments of the network removed from the initially small changes. The “viral” in the phrase “going viral” refers, of course, to a disease organism, which can potentially be transmitted to a large population through very few initial carriers. But using the dreaded cliché “going (gone) viral” negates the need for the discussion contained in the entire preceding paragraph. People who have no idea of any aspect of network theory, much less Metcalfe’s Law, know in an instant what is being communicated when that phrase is invoked.

Which brings me to my heartburn with the ultimate point of the original article, the idea that, if you don’t do as the writer recommends, you will somehow suffer in a job interview. Many articles have been published around this theme, mostly along the lines of Human Resource directors spilling their deeply-held secrets about who does or does not get job offers. These sub-rosa decision criterion almost always turn on what most would consider trivial aspects of the candidate’s resume, dress, or demeanor. It should go without saying that focusing on such trivial aspects of employment candidates, instead of their capacity to generate or protect more wealth than their salaries, dooms the organizations these HR geniuses work for to an inevitable, but well-dressed, demise, clichés and all. From a business writer’s point of view, it’s a cop-out. It’s a defense of an arbitrary and capricious practice, based on nothing that could remotely pass for “management science,” but dressed up in such a way as to present as helpful information for the reader. Yeah, showing up for an interview and using clichés might come off poorly – so does being overly verbose while answering direct questions, and the selection of linguistic approach depends on the circumstances.

So, yeah, managers and projects fail. So do business writers.

In my previous blog I used the dopey literary device of furthering the idea that a management science concept could be legitimately evaluated in a fictional setting. However, many business insights can be gleaned from analogous non-fictional events, especially war events, which is perhaps why battle analysis is such a popular exercise of management writers. In wrapping up our November theme of information technology and its impact on management, I want to evaluate some common but wrong-headed management science concepts, and how they are shown to be invalid by the way that events unfolded at the Battle of Midway.

Midway is an atoll, as the name implies, approximately midway across the Northern Pacific Ocean. After the Doolittle Raid against Japan in April 1942, the Japanese high command sought to extend the defensible perimeter around the home islands, and elected to invade and conquer Midway. And the Imperial Japanese Navy was fully capable of doing so – they had six large attack aircraft carriers, and multiple smaller carriers, battleships, cruisers, and support craft, manned and piloted by highly-trained and experience crews. Conversely, the United States had only three aircraft carriers, and one of those, the Yorktown, had been badly damaged at the Battle of the Coral Sea in May. The American aircraft could not perform as well as their IJN counterparts, and there were fewer of them. In fact, by any measure of pre-battle analysis, the situation for the U.S. Navy was virtually hopeless. But, the USN had one key advantage: they had better information.

The Japanese Naval code, nicknamed JN-25, was extremely sophisticated and considered unbreakable. However, American cryptologists had made impressive progress in reading IJN communications, and were becoming aware of an IJN target named “AF,” though it was unclear where AF was. A junior naval officer suggested that the commander of Midway transmit a plain signal that they were low on drinking water, and it was approaching crisis levels. Soon after, a JN-25 message was intercepted indicating  that AF was low on drinking water, and the USN knew that Midway was the target. By interpreting further communications, the Americans became aware of almost the entire Japanese order of battle, including major units assigned and an approximate timetable.  Admiral Chester Nimitz decided that his best response would be to place his only available carriers, Enterprise, Hornet, and the hastily and only partially-repaired Yorktown, in a position north and east of Midway, to counter the Japanese assault.

Conversely, Admiral Chuichi Nagumo had no idea that the Americans had been tipped off to his intent, much less his approximate order of battle. Nor was he aware that the USN had any units at all in a position to counter his mission. As he approached Midway, Nagumo had his four carriers prepare for an attack against ground targets, with some reserve aircraft prepared to attack ships in the unlikely possibility that the Americans had naval units in the area. On 4 June 1942 IJN carrier-based bombers attacked Midway, but failed to damage its facilities sufficiently to allow the commencement of the ground assault. As if to punctuate the need for another strike against the island, a wave of Midway-based high-altitude bombers arrived over the Japanese fleet, and dropped their bombs. Even though none hit, the need for another hit against Midway was apparent. As the first wave of Japanese dive bombers were returning, needing to refuel and requesting to be re-armed with more ground-assault ordnance, the scout seaplane from the cruiser Tone radioed in with the worst possible news – it had sighted an American fleet, including what appeared to be an aircraft carrier. Nagumo finally decided that the planes that had already been armed for ground assault be re-armed for naval targets, and that the returning aircraft would be recovered and refueled prior to the launching of the next assault.

It was at this moment that the American carrier-based planes converged on Nagumo’s task force. Once the torpedo planes were virtually wiped out by patrolling Zeros, the American dive bombers had scarcely-defended approaches to the Japanese carriers, three of which were reduced to blasted, burning metal within minutes. The Japanese would go on to lose all four of the participating carriers, along with over 3000 valuable pilots and crews, while the Americans lost the damaged Yorktown, destroyer Hamann, and 307 pilots and crew.

 So, what are the management science lessons here? As I discuss in my recently-released, must-have book, Game Theory in Management (Gower Publishing, 2012), I believe that no asset manager in the world would had advised Nimitz to even try to defend Midway – the imbalance of material was simply too stark, and the opportunity for exacting enemy losses remote. Similarly, which commonly-practiced risk management technique would have yielded a go-ahead advisory to Nimitz?  Note that the battle-winning intelligence had nothing to do with estimating the odds of the Japanese selecting a particular approach – the intelligence simply rendered an accurate picture of what was actually happening, real-time. In reality, the only advantage the USN had was in their pre-battle information – in every other category, the Americans were hopelessly overmatched, and yet the value (or even existence) of a superior information stream, providing timely, accurate, and relevant information, is not even taken into account in the business models of the asset or risk managers. In my thinking, this reveals the asset and risk management information streams to be far more confined in their range of efficacy than has been advertised.

What’s the value of information technology to management? It changes history.