This action might not be possible to undo. Are you sure you want to continue?

In this paper we draw on the knowledge of risk theory to explain why the risk of Complex Projects is significantly different from the risk of Simple projects. The interdependency of contributors, which is the mark of complexity, leads to mathematical properties of the resulting risk which is often overlooked by decision makers: higher probability of large failures, and a high influence of correlation. Teachings from statistical theory: probability of the outcome of simple versus complex systems What are the main differences in the resulting probability distribution?

The main difference in the resulting distribution is that the tail of the power law distribution is much wider than In statistical theory, the two extremes of systems the tail of the normal distribution. In other words, in simplicity and complexity lead to significantly different complex systems, the probability to have a significant statistical distributions. Complexity is related to the event far away from the average is much greater. On the interdependency between the different elements of the contrary, the normal distribution leads to very, very low system under consideration. The more a system is probabilities quickly when one moves away from the complex, the more interdependent are all the different average by a few standard deviations. Not so with the elements. power law distribution!. For simple systems where all the different elements are For example, the probability of an event distant from the fully independent, the limit value theorem states that the median (P50) value more than 3 times the standard overall statistical distribution of their sum tends to be a deviation is less than 0.1% for a normal distribution – normal distribution (also called Gaussian or “bell curve”). and still about 2% for a power law (exponent 2.5). An interesting property here is that even if some This observation has a lot of elements show a non-symmetrical consequences, and fits much (skewed) probabilistic distribution, In complex systems, the better real life observations. the resulting sum tends to be much probability to have a significant Unfortunately, because normal more symmetrical. event far away from the average is distributions are much easier to Complex systems tend to behave much greater handle mathematically, and are very differently and the resulting more intuitive, they are used in a risk is generally a power law curve number of instances where they (also called Pareto curve). It is a do not represent reality. This has resulted in a lot of distribution where typically, events that are 10 times less criticism with regard to the traditional equations derived frequent are 10n times more severe, where n is a from power law assumptions supporting various financial parameter. For example, for earthquakes magnitude, n≈3. products, for example, or representing the fluctuations of It works the same for popularity: the category of n times markets ([2], [3]). It is even more the case in the field of elements that are 10 times less frequent are 10 project management, where the few quantitative risk more in number or size: for the distribution of family approaches have all been based on assumptions of names, n≈1,9; for the hits on webpages, n≈2.4, for the independence of the different factors at play in a project. population of US cities, n≈2.3 [1]. It is a phenomenon Hence, the actual risk of project management has been that can be observed in a large variety of man-made or constantly undervalued in traditional quantitative natural situations. The famous ‘long tail’ (Zipf law) which methods, which is confirmed by represents the distribution of actual observations: while Montepopularity of any product, the Carlo based quantitative methods The actual risk of project probability/impact of for project risk give a variability of management for complex projects earthquakes or the famous 80/20 the project cost in the range of has been constantly undervalued rule of prioritization, are all based ±5% or maybe at most ±10%, we in traditional quantitative methods on this power law distribution for can observe at the same time a complex systems. The power law significant number of complex distribution tends to represent a lot of real-life situations, projects with variances far above these numbers, as long as they correspond to a complex system with a lot reaching sometimes 2 or 3 times the estimated cost with of interaction. a measurable probability (as they do visibly happen in all Actually, without knowing much about the system other industries, and every few years in project companies). than it is a complex system with a lot of interdependent elements, we know the outcome probability distribution is a power law curve. This powerful insight will be used in this paper.

© Project Value Delivery, 2012

2012-24 rev 0

Discover more on:

www.ProjectValueDelivery.com

comes to complex projects. This effect outcomes worse than average. etc. For power law distributions. Benoit Mandelbrot. and sometimes level does not really work when it organization or branch. and sometimes outcomes better than of project risk at the organization thus. Let’s suppose are somewhat correlated. The reality of complex projects is common processes. It can be represented statistically by a power law. with projects considered to be personal. more on that in situation are far reaching. the probability that the aggregate exceeded © Project Value Delivery. The project risk at the organization level does not really work when it comes to complex projects. with a much higher probability of particular organization or industry. The consequences of this mathematically distribution. even if the [2] The Failure of Risk Management: Why it is broken probability for one project to be loss making is 15% and how to fix it. This is the Summary usual justification of project-based organization – the aggregation of many projects makes a relatively The reality of complex projects is that really disastrous predictable overall performance. because not so good outcomes are generally much worse than good demonstrated situation are far reaching: it shows that the outcomes. Richard 0. The probability of loss making for the aggregation of projects becomes very small. They are that the outcome of a project we correlated through the use of consider is its cost. some research simulations have been [1] Power laws. the different projects significant deviation from the mean values. If we consider occur far more often than in similar locations. Douglas Hubbard. if the probability for Hudson one project to be loss making is 15%. Pareto distributions and Zipf’s law. As an illustration. This is because there is a high probability that one of these projects will be very significantly loss-making. amount of correlation varies. in the same order of magnitude will evaluate the consequences for the organizations and than for a single project. it is not so much the case. organizations. We have now established that mathematically. 2009 (after utilization of contingencies etc). policies. projects occur far more often than predicted by common For complex projects that follow the power law knowledge. relatively rare occurrence of a project that is really very References: bad. but we’ll use coefficients. because in a power law distribution. and that has a significant impact on the overall outcomes and diminish the overall risk to the risk level of the aggregate. increase the overall risk to the average. because it is driven by this for risk management policies.Page 2 of 2 How different distributions affect the risk of the aggregate of several projects the total revenue by more than 25% was still 1. remains limited in the case of For simple projects which have a normal distributions. erasing all the gains from the other projects. good and become quite dramatic in the case of power law not so good outcomes cancel out nicely. For normal distributions. but can normal distribution. the probability of [3] The Misbehaviour of Markets: a Fractal View of the sum of 6 of these projects to be loss making was only Financial Turbulence.edu/~mejn/courses/2006/cmplxsys899/ powerlaws. organization. the addition of all these projects produce a traditional economic protection offered by aggregation of distribution that is skewed toward worse outcomes.5%. The that their outcome is fully predicted by common knowledge. Even with weak correlation the next paragraph. and the final distributions: correlation increases the probability of distribution ends up to be narrower in relative terms: the freak events appearing simultaneously on several addition of several projects is a good way to average out projects. 2012 2012-24 rev 0 .5%!). the probability of the sum of 6 of these projects to be loss making was still 12%!! (and while the probability that one single project’s loss exceeded the project average revenue by more than 25% was 5%. we independent (which is not the The consequences of this will discuss here the general effect case in actual project mathematically demonstrated of correlation. A series of papers probability of the aggregation of these projects to be loss making is still very high. the effect of this simplification here as a first The traditional economic correlation is to increase the spread approach).pdf independent. paper done by PVD with representative distributions for http://wwwretrieved at project outcomes. there will be protection offered by aggregation of the aggregate of the projects. people Let’s consider a number of that really disastrous projects or equipment. the The reality is much worse: correlation outcome of complex projects need to be represented by a In reality. because they happen similar projects.umich. the situation is even worse.

- [David C Sim] How Many Jews Became Christians in t(BookZZ.org)
- No Punctuation is Funnier
- RR Layouts
- Interrogation Techniques
- [Martin Montgomery] an Introduction to Language and Society
- David Ludlow - Railroad 101 01-22-15
- Do Not Retire Baby Boomer Manifesto
- Grading Concrete Roadbed _ Garden Railways Magazine
- Your Life in Weeks _ Wait but Why
- Duplicate Cosmetics
- Math Ics
- 27218 Web BASICStampManual v2.2
- [Edwin P Alexander] American Locomotives a Pictor(BookZZ.org)
- 226838560 Gary Allen What a Way to Run a Railroad Railroad Monopoly in America 1976 (1)
- How to Wire a Layout for Two-train Operation _ ModelRailroader
- Math Ics
- [Bartelmann M.] Cosmology(BookZZ.org)
- Plant Invaders of Mid-Atlantic Natural Areas
- 107.03.GrowingUpJersey
- 18981 Training Students to Extract Value From Big Data
- MMS Salary Survey 2014
- Fundamental Errors in the General Theory of Relativity
- Drawing Basics_free Beginner Drawing Techniques
- 2014 Phrma Profile
- Collaborative Project Management Guide eBook

project management PMO project selection

project management PMO project selection

- Decision Making
- US Federal Reserve
- US Federal Reserve
- A Coherent Framework for Stress-Testing.pdf
- W4INSE6220
- Eurostat Integration Methodologies
- Risk Response Plan Template
- Monte Crl
- EXAM risk
- Probabilistic Credit Scoring for Cohorts of Borrowers - R Mosquera
- Introduction to Bayesian Networks
- Lec2
- Information Ferret Raid Log Example
- Nassim Taleb Journal Article - Report on the Effectiveness and Possible Side Effects of the OFR (2009)
- Risk Analysis
- Seven Myths Slides
- SuccessProbSPDM
- Selection of Technique for Assigning a Target Safety Integrity Level
- Investment Part II-1
- Advance Technique for Contigency
- Probability
- ASM Template Import R1
- Decision Tre Analysis
- HAZOP
- 311E-C02 Tversky Kahneman 1981
- Paper MID
- bm_aktienkurse_e.pdf
- eur19056IIen
- Return and Risk
- 2012-24 True Risk Complex Projects v0

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd