Mythical Man-Month, The: Essays on Software Engineering, Anniversary Edition

Oddly, I was reminded of this classic work whilst reading Chris Date's otherwise quite unremarkable tome, "The Third Manifesto". Date and Darwen cite this classic text admiringly. And this may be the most important contribution to have emerged from their efforts. Having toiled in the Information Technology field for decades, I was, of course, familiar with many of the gems of wisdom that were first articulated by Brooks in this classic book. But it was a true joy and revelation finally to read the book itself from cover to cover. Among the pearls of wisdom contained within these pages are the following: Adding people to a late software project tends to make it later. While it takes one woman nine months to give birth, nine women cannot accomplish the same task in one month. (Hence, the concept of the mythical man month. People and time are not interchangeable commodities.) The factor most dispositive of success in software engineering is conceptual integrity. The first duty of the manager is create a concise and precise written plan. Communication, and its attendant, organization, require as much skill and careful consideration as any other aspect of technical project leadership. There are many, many more wonderful insights contained within the corpus of this outstanding book. While dated, no doubt, the truths that emerge from careful consideration of this important work are that overcoming problems of human interaction are really paramount to success in any task as complicated as software engineering and that the discipline of software engineering is perhaps one of the most wonderfully rewarding career paths open to creative and serious folks even today. This outstanding book rightly deserves an honored place in the library of any person who would succeed in a career in information technology now, or in the future. Yes, it deals with human factors that some may argue can be overcome by technology. But, as Brooks so cogently demonstrates in his wonderful essay on the "silver bullet", the search for the final solution to the problem of software engineering is very much like the hope to slay the mythical werewolf with a silver bullet in that it is a search for an enigma to deal with a chimera. It can't realistically hope to succeed. Finally, in assessing the timeless importance of this classic, we are reminded of the sage advise of that great philosopher, Arnold Schwarzenegger, that, when working with people, everything is political. Yes, the human factors always do matter. And Dr. Brooks has illuminated those human factors of software engineering in a manner both satisfying and edifying. Pick up this timeless classic. Absorb the teachings. And watch your productivity and effectiveness in the discipline soar. God bless.

The Mythical Man-Month is Frederick Brooks' seminal collection of essays vis-a-vis software engineering. From the title, one would imagine that the tome's unifying thesis revolves around the discredited idea that adding more engineers to a project will enable the project to be completed in fewer months, or, to put it another way, that the length of a project's schedule is a linear function of the number of workers assigned to that project. Using graphs based on mathematical formulas and on research conducted by other specialists, Brooks neatly dismantles the person-month myth - demonstrating, in fact, that in many projects (particularly if complex interrelationships are required or if the project is behind schedule), adding more bodies often increases the time required for completion. Despite what the title suggests, however, the above-mentioned topic is but one of many covered by this work. Other topics include the distinction between the "essential" and "accidental" elements of software design; the distinction between building a computer program vs. designing a "programming a systems product" (and the ninefold difference in complexity and time between the two); the quest for software engineering's elusive "silver bullet"; the importance of documentation; the surprisingly small percentage of time that actual writing of code occupies on the timeline of a typical software-development project (as contrasted with time needed for testing and debugging); large teams vs. small "surgical teams" (and why the latter isn't always the answer for all projects); the "buy versus build" dilemma; and many others. Much of the material in the first several chapters of the book appears obsolete (although there are still valuable principles that can be gleaned). However, in chapter 19 (a kind of "retrospective" chapter added 20 years after the original publication date), Brooks amends much of the out-of-date material, e.g., his earlier views on program size and space metrics (rendered all but irrelevant in this age of multi-gigabyte memory), and the degree to which the (albeit hard-to-predict) personal computer explosion and the growth of the Internet. However, even since the time of the book's revision (1995), further explosions have taken place in the computing industry - most notably with regards to Web 2.0, the ubiquity of data-driven Web applications (these even obsoleting many shrink-wrapped products), Web services, and development methodologies such as Agile and XP - that even chapter 19 may seem a little out-of-date to the modern developer. In spite of this, the principles of the book are still applicable: the chapters on estimation, team size, and the dismantling of the person-month myth are enough to make this tome required reading for developers and managers alike - especially the latter.

This is an excellent compendium of knowledge about software development, particularly in relation to project management and efficient team organization. Though it covers more topics than just those, it really demystifies and sheds light on why managing software development is so different and so much more difficult than any other industry. If you have any interest in philosophy, computer science, or good writing, this book is well worth your time. If you are interested in two or three of them, it's a must-read. This is a classic in the software development space and has been extremely influential for many years. Mr. Brooks' writing style is impeccable; he carefully dissects and examines each topic, with the wit and wisdom merited by such a technical field, yet he does it without using a lot of double-speak or unnecessary "fluff" - not a true text but rather a collection of essays, each chapter comes across as a polished, finished product, well-focused on a single topic. This particular edition is also highly recommended. It contains four additional chapters: No Silver Bullet, yet another influential essay by Brooks that was not in the original edition; an overview of all his points (the entire book) in an easy-to-digest format; his thoughts 20 years on from writing the original, and how the industry has changed in that time; and finally, his responses to various criticism he has received over the years specifically in response to the "No Silver Bullet" essay. This is an excellent purchase and a great read.

The one constant in the technology world is change. The biggest complaint that I have as a reader of technology books is that their material seems outdated by the time you pick it up off the shelf. It is nothing less than amazing that a book originally written almost 40 years ago is still applicable today. I can't count how many times while reading the book that I forgot how old the material is and found myself stunned when Brooks throws in a reference to System 360 as if it were a recent release. The central idea behind the book is that as software projects grow, you have to change your approach to managing them. The myth is that you can scale a project linearly. Every resource you add to a project does not shorten development time by a constant rate. In fact, in many cases it could lengthen it. I've seen the result of ignoring this warning a number of times (not pretty). I've also had the privilege of working on a team that respected the axiom. Brooks' idea of the surgical team metaphor for software development is intriguing. I have yet to see it fully implemented as he describes but I've come close to it with a core group of "lead" developers pairing with junior developers to augment their abilities. To the point, this should be the first book you read before taking the helm of a software project.

This classic on software engineering made its mark as a study of how to manage large complex projects. Most of the book applies to management in general rather than to software engineering in particular. Here are two examples of the more general insights the author, former IBM manager Frederick P. Brooks, gives readers: an instance of a straightforward insight immediately applicable, and another one we obtain by carefully (!) reading the text. Brooks looks at different types of projects; large projects that can be split into many simple independent tasks will be completed faster if we add more staff to carry them out. However, engineering projects are seldom so simple. Developing software for a new machine requires the machine, which is itself being developed, as well as documentation, which is being prepared! All these tasks relate to each other and require all participants to communicate with each other. The number of communication lines within a team grows exponentially with respect to the number of team members. At some point, adding more men and women actually delays project completion, shattering the myth of the man-month. However we must be careful as we read the Mythical Man Month. This book was written in the seventies about the author's experiences in the sixties, so to understand Brooks correctly, we've got to read him carefully. For instance, Brooks praises PERT charts, saying outright that "there is no substitute" for them. I can't believe this is a blanket endorsement for mindlessly turning out slides and charts by using Microsoft Project! Brooks didn't have MS Project or PowerPoint in the sixties: PERT charts were carefully drawn, often by hand, they were expensive, and they were prepared after thinking things through. We find the true insight a little further down page 156: "The preparation of a PERT chart is the most valuable part of its use". Some of the book is of course more relevant to software engineering. For instance, Brooks's correct 1986 prediction that off-the-shelf, shrink-wrapped software would become the standard way to implement solutions. Just look at Microsoft Office or at SAP's R/3 to see the truth of this. (I'd even say that because powerful software is now so cheap, we've created a glut of output.) Read properly, The Mythical Man Month remains as insightful today as in the seventies and eighties. Brooks's style is friendly but professional and business like. Budding project managers will find many useful insights. Vincent Poirier, Tokyo

This book is a mix of some awesome insights that comes only through real world project management and some dangerous preaching that comes only through not writing code for years. Any arguments with or against this book is in vain because what this book discusses is not mathematical proof but rather empirical observations and extrapolations on few types of software development projects carried out with few types of methodologies. One of the most dangerous things that this book does is recommending and preaching elitist groups, aka Architects. The author actually goes out and claims that architects should never write a line of code!! Madness. I've seen and worked with those kind of "Architects" and to say the list, a person who is blueprinting your million dollar project without knowing the strength and weakness of his tools intimately is bringing disaster to your project. Your policy should be to hire architects who must also prove themselves to be the best coder in the team. I can't overemphasis this point. I see people often stating that they spend less than 10% time on actual programming because they are architect. They start out giving analogy of the person who designs the building but doesn't put his hand son brick and mortar. The analogy is obviously misleading. The bricks that make up building are all same and doesn't require any skills in setting them one over another while a program is made of several different complex constructs, components and libraries and using one thing over another can make a huge difference. Another bad thing that this book does is to draw a picture that big teams are bad and should be avoided at all costs if possible. I think, one has to keep open mind about this. I've always believed that using right methodologies, tight control on recruiting and management structure with isolated smaller groups it is possible to create teams with thousands of developers and deliver the product on schedule. The essential part of methodologies included componentizing the entire product, forming hierarchy of small sub-groups each of who owns these components and talks to each other through well defined APIs, using test driven development and continuous integration. The point is, large teams aren't evil anymore because current state of programming tools and methodologies have evolved to support it very well which wasn't the case 3 decades ago. This book has probably done more damage than good. Still, this book I would rank as one of the must read for any software developer/manager provided you read it AFTER you have significant experience of developing the software with the intention of enhancing your understanding rather than acquiring. You would be most benefited if you look at the book with an eye of a critic instead that of a student.

The first half of the book is a case study of the development of OS/360 in the 1970s: what the problems were, what was tried, what worked and what didn't. While I (and probably many others) snicker at the state of technology then compared to what it is now, I feel that the lessons Brooks learned (and happily relays to the reader) are still relevant and valuable. You certainly will have to abstract the methodology to the current technology we have today, but managerial lessons, as I said, are still relevant, mostly because people haven't changed that much. Basically, adding more people to already-late projects makes things worse. All of the communication and documentation that goes along with large projects are 100% necessary, and the documentation should be about 90% complete before coding starts. I think a wiki would solve both of these issues in one shot, but that's me. The last half of the book is mostly an inner dialogue by Brooks about what he thinks of the lessons he preached, what other people in the industry have said about his book, and his responses to it. I think this is a definite must-read for anyone that programs on large software projects or manages large software projects. Brooks comes right out and says at the beginning that other engineering disciplines already know about all of the project management overhead, which I agree with, because I am in one of those other disciplines. Apparently the programming people don't see it necessary to teach project management as part of a bachelor's degree program, which might explain a lot of the larger programs in the past few decades. I have to admit though, the entire computer industry, both hardware and software, has been through a tumultuous and extraordinarily rapid history. Other disciplines have a much longer history book from which to reflect and design better processes, management or otherwise. Finally, the prose is dry sometimes awkward, which I suppose is typical of the professor types with delusions of eloquence. Despite that, I thought it was overall an easy read, though not as humorous and engaging as some of the other software books I've been through.

Although it was written in 1975, much of the book's information remains surprisingly relevant today. For instance, Fred discusses Artificial Intelligence in Chapter 16, along with expert systems and inference engine technology. It’s understandable to wonder if this information might be outdated, but it still holds value. For small projects, diving into these details might be more than necessary. However, for larger projects, whether you’re a programmer, engineer, project manager, or part of senior leadership, understanding these concepts can be incredibly beneficial. Big projects involve more than just building something — they require the right team, organization, coordination, effective debugging tools, and strategies for deployment. Many of us have experienced OS or system issues and might be curious if developers conduct thorough testing before releasing updates. It’s important to remember that when changes are made, they often need to be validated from scratch, which takes time, resources, and additional costs. A lot comes down to discipline and judgment — skills we can all learn from this book. When you build something, it’s essential to have ways to debug it, and to ensure that users have access to clear and comprehensive documentation, not just a few quick instructions. While mainframes might seem outdated to some of us, it’s interesting to note that many of the core components, like memory and disks, are still in use today. To get the most out of this book, focus on the underlying concepts and proven methods. And most importantly, appreciate Fred's foresight.

Letto molti anni fa all'inizio della carriera, l'ho visto in offerta usato e ho voluto comprarlo per la mia libreria. Racconta molte cose che dovrebbe essere ovvie a tutti, ma purtroppo ancora oggi si incontrano professionisti del settore a cui mancano decisamente le basi.

Essential reading for all Software Engineers

Too difficult for a new graduate

Mr. Brooks and his team needed 5000 man years from 1963 to 1966 to create OS/360. Peak head count was 1000 persons. Today we ask the question: was this massive amount of labor necessary? I think so. Today the art of operating systems is much advanced. But at Brooks time, he had to break new ground, fast and in good quality. He and his very large team did write history, like Prof. Corby (Corbató) did with Multics. Next to Brooks law I like his scheduling rule most: 1/3 planning, 1/6 coding, 1/4 component test and early system test and 1/4 system test, all components in hand. In my projects I used the surgeon team, Mill's Proposal. It worked again and again.

Emballage légèrement endommagé. Coup sur la couverture du livre