CPM: Constant Perfection Method

November 1963

I remember exactly where I was when it happened. It was November 1963 … and I was upstairs in my bedroom, lying on my bed and staring at the ceiling, counting the silver star-shaped decals I had proudly applied in the hour before.

From the date, you may be thinking that I am talking about where I was when President Kennedy was assassinated, but you’d be wrong. I was out on the front porch when that terrible moment happened just after I had finished lunch. Besides, that would not happen for another three weeks.

No, this Major Event in my life took place on November 3rd, a Sunday. I remember because we had just finished dinner and Mom was cleaning up the dishes in the kitchen, and Dad was settling into his easy chair in anticipation of an exciting hour with Ed Sullivan.

The news had been on just before dinner and the weatherman had given his predictions about the upcoming workweek. It was something he had said that I asked about over dinner, and that became the subject of a healthy discussion over roast beef and mashed potatoes.

“Dad,” I garbled as my tongue pushed aside a cluster of green peas, “the weatherman said ‘a 75% chance of rain tomorrow.’  Does that mean that it will rain for 75% of the time tomorrow?” Over the next 15 or 20 minutes my mom and dad took turns explaining to me how probabilities work. To my specific question I was told that the 75% pertained to the likelihood of rain falling on my head, but not the percent of time that rain will be falling.  By the time dessert came around I pretty much understood the concept of probabilities.

Or so I thought.  Upstairs, lying on my bed while spinning GI Joe by his arms, forcing him to do countless involuntary somersaults, I kept thinking about the weatherman’s face as he so proudly gave his predictions for how the week would play out. What occurred to me, as I lay there, was how the mere utterance of percentages had this magical effect of elevating otherwise speculative statement into one of great precision and credibility.

What occurred to me was that, as long as the continuum spanned from zero to 100, a 75% probability sounded quite precise … and certain! That is, is zero represents “I have no earthly clue,” and 100% represents “I’m absolutely certain beyond a shadow of a doubt,” then a 70% probability statement expresses a whole lot of certainty.

But the great revelation they hit me as I lay there was that it is not zero that represents “I have no earthly clue,” it is the 50% point. I mean, doesn’t “a 50% chance of rain” mean that there is as much of a chance that it will rain as there is that will not rain?  And isn’t that just a fancy way of saying, “I have no earthly clue?”  And a 40% chance of rain must mean “there is a slightly less chance of it raining than not,” while a 60% chance of rain must mean “there is a slightly greater chance of it raining than not.”

Suddenly, I was not so impressed by the weatherman. I now realized that his ‘75% chance of rain’ was only half way between “I have no earthly clue” and “I’m absolutely certain beyond a shadow of a doubt.” In other words, his 75% probability equated to only a 50% certainty. Weird, huh?

Precision versus Certainty

There is a common misperception suffered by those who create CPM-based project schedules as well as those who utilize them. There is a general impression that Critical Path Schedules reflect a fairly high degree of certainty. The mistake we make is confusing precision with accuracy. Our schedules contain precise durations, precise leads and lags (quantifying the stagger between overlapped activities), and precise start and finish dates. With all of this precision, surely the schedule can be trusted; surely the information contained within it can be relied upon with confidence.

But can it? That is the question of this week’s blog. Just how credible is a CPM schedule? Just how realistic could it ever be? And the operative word here is “realistic.” For, if the schedule is not realistic it is also not credible. So let’s talk about realism in our schedules.

Introducing Earliest Dates and Latest Dates

Activities represent the performance of Work scope. All performance consumes time. Accordingly, all activities are assigned durations, estimates of how much time will be required to perform the Work scope of the activity.

Activity dates indicate when activities start and finish, as determined by activity durations. It follows, then, that there is a direct correlation between the reasonableness of activity durations and the credibility of activity start and finish dates. In this article I will not address how to derive fairly reasonable durations; let us leave that for another day. Instead, perhaps we can assume that the activity durations are reasonable. The question that this article asks is, “are start and finish dates inherently reasonable as long as the activity durations upon which they are based are inherently reasonable?”

The answer might surprise you.

Because not all readers of this blog have the same level of knowledge of the Critical Path Method, let me clarify which dates we are talking about. There are two sets of dates: Earliest Dates and Latest Dates. Earliest Dates for activities in a CPM schedule are derived through a process known as a Forward Pass, in which the durations of the activities are cumulatively aggregated such that the start and finish dates of a predecessor activity necessarily contribute to the determination of the start and finish dates for the successor activity. As you might expect, a Backward Pass calculates the Latest Start and Latest Finish dates for all activities.

Earliest Dates Declare What Is Possible

The way to understand Earliest Dates is to think of the expression “if everything goes perfectly!” For that is what Earliest Dates really mean. So that you can fully appreciate this point, imagine someone asking you to schedule the next three hours, down to the minute. It is Sunday and you are about to take your son to a movie. It is 1:30 p.m. and you have just made the decision to attend a movie on the other side of town that starts in 45 minutes.

You still need to change clothes. There’s the drive to the theater, and your car needs gas. You need to allow time to purchase the tickets including possibly standing in line. You need to allow time to buy your refreshments, and make your way to your seats. After the movie, the two of you will go to a nearby hot dog stand where you will buy two brown bag dinners and head over to the park on foot. After chomping down the hot dogs and feeding the bread to the pigeons, you will make your way back to your car, and eventually arrived back home.

Again, the assignment is to schedule all of this down to the minute. Obviously, for all of the above tasks, you must estimate reasonable durations. Then, by cumulatively aggregating the durations, starting at 1:30 p.m., you can calculate a start and finish time for each and every activity. Let’s try:

  • Suppose you can throw on some shoes and be in the car in four minutes. Start: 1:30; Finish 1:34
  • Suppose that fueling the car will require seven minutes. Start, 1:34; Finish, 1:41.
  • Suppose that the drive to the theater is 18 minutes. Start, 1:41; Finish, 1:59.
  • Suppose that parking the car, walking into the theater, standing in line, and finally buying your tickets can all be done in eight minutes. Start, 1:59; Finish, 2:07.
  • Suppose that you can stand in line at the refreshment stand, purchase your refreshments, and make your way to your seats and another eight minutes. Start, 2:07; Finish, 2:15.

If everything goes perfectly you can be seated, with food in hand, when the curtain opens and the credits begin to play. If everything goes perfectly!

This is how the Critical Path Method works. When we perform a Forward Pass, we make the assumption that all previous activities will have occurred in precisely the amount of time allowed by their durations. The important thing to notice is that, except for the activity durations themselves, there is no opportunity to inject realism into the schedule.

But Project Time Management “best practices” strongly discourages the padding of activity durations. That is, generally-excepted scheduling practice is to estimate and activities duration based on a reasonable amount of time to perform the Work scope of the activity. For example, if we think that we can fuel the car in seven minutes, then we should not assign an activity duration greater than seven minutes.

The Take-Away from the example is that all of those start and finish times reflect the earliest possible points in time for the start and finish of those activities, assuming that everything goes perfectly. This is the true meaning of Earliest Dates in a CPM schedule. They present the most optimistic, hopeful, and (some might say) Pollyannaish expectations possible.

The question I’m asking you is: how reasonable are we being if we expect each and every activity to occur precisely as we have scheduled them?

Earliest Dates clearly suggest what is possible … if everything goes perfectly. But, is “what is possible” the same as what is probable? What is likely? What is certain? And the better question, the one I would ask any seasoned Project Manager, is: would you reasonably expect all activities in a project schedule to attain each and every one of their Earliest Start and Earliest Finish dates? I’m asking, here, about what are reasonable expectations.

Latest Dates Formalize Procrastination

For those of you who are unfamiliar with how a Backward Pass works, the process begins at the end of the project schedule where we expect to find some ultimate deadline by which time the project must finish. Then, by working backwards through the activities, we cumulatively subtract all activity durations in the schedule. Ultimately, we arrive at a start time for the very first activity, which represents the theoretical last possible moment that we can delay commencement of the entire schedule … and still meet the project deadline.

To illustrate how a Backward Pass works, using our movie example, we need only imagine that it is not 1:30 in the afternoon, but instead eight o’clock in the morning. The family is having a leisurely breakfast and discussing what to do in the afternoon. Mom suggests that Dad and Junior go to a movie. Out comes the iPad, movie listings are consulted, a theater is chosen, and a movie start time is noted: 2:15 p.m.

The same list of activities as was developed in our earlier example begins to form on a sheet of paper as Dad itemizes all of the steps that need to occur from getting dressed through settling into the movie seats. Activity durations are estimated for each separate activity. Then, starting with 2:15 p.m., the activity durations are cumulatively subtracted. When the Backward Pass is finished, it is determined that the latest that father and son can start getting dressed is … 1:30 p.m.

So, what does this tell us about Latest Dates? For one, it is blatantly obvious that Latest Dates mean that we are waiting to the last possible moment to start our project. Second, just like the Forward Pass, it assumes that each and every activity will occur precisely as scheduled, and that no unexpected surprises will befall the project. Once again, the assumption is that everything will go perfectly.

Given that life is full of surprises, and given that all of the activity durations were, at their best, just estimates … how prudent wouldn’t be for a Project Manager to wait until the last possible minute to start a project? Would any reasonable, responsible individual late until the last possible moment to start a project and then hope and pray that everything will go perfectly?

Should CPM Perhaps Stand for Constant Perfection Method?

When we speak of perfection in a schedule we are describing a situation in which activities are completely, utterly, absolutely, and flawlessly performed. How often does that happen in your life? And yet, the Critical Path Method gives us only four calculated dates to work with: Earliest Start, Earliest Finish, Latest Start, and Latest Finish. That’s it!

Two sets of calculated dates. One set represents absurd, almost naïve, optimism — the kind that no life-experienced person would embrace. The other set represents the ultimate in procrastination, staking the entire success of the project on the assumption that, once begun, everything will occur perfectly.

View at from either perspective, it would seem that CPM believes in a fairy tale called Constant Perfection; that, not just for a few moments but for the entirety of the project, perfection will prevail with constancy. We are beginning to see the Critical Path Myth.

As If the Schedule Weren’t Incredible (Not Credible) Enough!!

“Okay, Murray,” you think to yourself, “perhaps the points you are making are technically valid … but aren’t you really just living in the theoretical world? I mean, do your arguments have a substantive relevance in the real world?”

Well, let’s see. Certainly you have heard the term Critical Path … and as in Critical Path Method? The critical path is that string of activities having the least Total Float. And just what is Total Float? Are you sitting down? Total Float is a theoretical value derived by subtracting Latest Dates from Earliest Dates. For example, if we were to subtract 1:30 p.m. (Latest Start for getting dressed) from 8 a.m. (Earliest Start for getting dressed — you can’t go back in time), we would determine that we have 5 1/2 hours of Total Float for our movie “project.” That’s how Total Float works.

In the real world of Construction Project Time Management, well-structured and highly formalized “best practices” dictate that, on a regular basis (most typically, monthly), the project schedule should be updated and evaluated to determine how well the project is performing against temporal commitments. In practice, reports are generated that compare actual performance with planned performance.

There are a number of standard performance measurement reports, virtually all of which incorporate Earliest Dates into their calculations, either directly or indirectly.

  • Start Date Variance Report: Compares the actual start dates for activities with the planned Earliest Start dates. Flags and reports any variances between what was planned and what actually occurred.
  • Finish Date Variance Report: Compares the actual finish dates for activities with the planned Earliest Finish dates. Flags and reports any variances between what was planned and what actually occurred.
  • Total Float Report: Monitors and reports changes in Total Float from one reporting period to the next. Management concern usually heightens as Total Float diminishes in availability.
  • Earned Value Report: This decades-old performance measurement technique compares actual consumption of key resources to planned performance based on Earliest Dates.

It would seem, then, that after all of the good and noble effort to create and maintain a credible project schedule, its ultimate use (at least, as a project monitoring and performance measurement tool) boils down to an unspoken assumption and expectation that the project’s progress will coincide with those pie-in-the-sky Earliest Dates. And, criticality will be defined in terms of those wait-to-the-last-minute Latest Dates.

Realism: The Elusive Element of Credibility

We need our project schedules to be credible. We want to trust them, to rely upon them to give us meaningful, timely, and useful information. We may not be able to do much to improve the inherently speculative nature of activity durations: they are, after all, provided by humans — who are known to err from time to time.

And the logic that we draw between activities, to symbolically depict the interdependency of activities one to the other, are also inherently speculative to some degree. And there will always be those surprises that continually and unpredictably afflict every project. We can do little about that either.

But the one thing we can do to salvage as much credibility in our schedules as possible is to be aware, and forever remember, that the two sets of calculated dates given to us by the Critical Path Method represent two extreme operating contexts, neither of which any sane Project Manager would choose in which to manage a project. The prudent Project Manager would no sooner expect everything to occur flawlessly at their earliest possible moments then he would wait to the very last possible moment to start the project’s work.

I chose this topic to write about with the sincere hope that it might reach those key members of Project Leadership who most often misunderstand how CPM works and most often harbor expectations of the schedule that exceed its inherent ability to deliver.

In this sense, at least, ignorance is not bliss.

4 Responses to CPM: Constant Perfection Method

  1. Zach Reed says:

    One must start somewhere, right? If we cannot start with our best estimate, then where should we begin? A project execution strategy is created to help break the project down into manageable (and individually attainable) parts with a set of assumptions and stratagems—given X many years of data we estimate the number of weather days per year to be ___, or we will use one crew per work type to perform each section of the project, etc. Granted these things can change, for example, maybe it’s a particularly wet year (more than anticipated), or perhaps you have an extra crew that can jump in and cut down work type durations. But one must start somewhere.
    This illustrates the communicative and dynamic nature of a project schedule. When things do not go according to plan (perhaps exemplified by a baseline variance report) options are discussed and a solution agreed upon. The schedule is then updated, and the current trajectory analyzed with the understanding that humans are involved.

  2. Sue Backiel says:

    Being new to the field, I don’t have any practical experience but I agree with Connie. There must be a better way. Using a “Probable” date as well as Earliest and Latest date would give more realistic and accurate reporting.

  3. Michael Neal says:

    [C] Connie, you are right on track. We often use varience reports to give out comparing the planned finish to the current projected finish. The works great for the super and management, though not needed in the field as much.

  4. Connie Bremer says:

    This blog is very thought provoking, and a little frustrating. Why in the world would all these variance reports be based on the earliest dates if the probabilities of meeting these dates is in most cases improbable. It sets up the key players to always be chasing that golden carrot to catch up. Why has no one incorporated maybe an earliest, likely, and latest date where you could run variance reports on these dates to see just how much realistic jeopardy you are facing in meeting commitments? Or base the reports on the likely dates, then if you make the earliest date you are ahead of the game. You could add/overwrite column headings in P6 as an example to incorporate another date column, but the canned reports will still use the planned dates. Yes you could customize a report, but this is just a work around that could be construed as not completely honest considering so many just accept the defaults and dominant methods. It’s a failure before it has even begun. I wonder if anyone has ever done a study to determine a percentage of activities that do finish on or before their earliest dates. No wonder so many PMs pad so much in building the schedule, if they provide what they presume to be reasonable durations, it would likely come back to bite them in the behind. I know I don’t have years and years of experience at scheduling, but there has got to be a better way. Perhaps I am just being naive.

Leave a Reply

Your email address will not be published. Required fields are marked *