As an IT fossil, I‘ve learned that whenever I hear “Methodology” I can peek around the corner and find a hoard of teenage consultants who know nothing about our business, nothing about our existing systems, nothing about the politics, nothing about the history of how things got to be the way they are, and little about human nature. They don’t have those things, they can’t get those things, and so they have to use a methodology. The slick principal consultants know a lot about sales and sucking up to the right folks though. And of course, they know how to blame the existing IT staff. Sometimes, it works though.

If I don’t find the teenage consultants, the next I look for is my own managers who have spent the last 3 years reorganizing. Because they’ve been in staff meetings for the last three years, they’ve come to believe that they don’t need to manage any more. All they need is the right methodology and the right organizational structure, then they can get back to reorganizing.

Hey, I like process, methods, predictability, structure, and quality. I wish amateurs could ask standard questions, plug the answers into equations and get the perfect system.

But the missing link is always good management, experienced management, proven management, folks who know where methods work and where the fail. In FogCreek’s case good management probably means the boss has to know enough to be able to anybody’s job. Not scalable.

The big companies have to learn that humans figured out the management thing before they built the pyramids.

-- Anonymous, January 18, 2001

Answers

Your article is a good description of the scalability problem. With most of it, I agree. I only partially agree with 'Beware of Methodology'. Sure, beware of *bad* methodology, but some methodology is good.

Your example, for example, talk only about methodologies of procedure (fry patty, toast bun, slice of cheese, squirt condement, insert into foam box). These tend to produce the kind of results you describe.

Methodologies of communiation, however, are very useful for teams to get a grip on a large problem. Methodologies of use anaylsis are also useful, by examining a consulting project in terms of the end users *goals* you make something that makes them happier. Methodologies of risk assesment and mitigation help you understand what problems to tackle first, and the feasability of a goal.

I guess in the end it's caveat emptor and allways keep you eye on the ball and all that. If you spend millions on project that a talented kid can do in a day for a pizza and twenty dollars, well, tough luck. Maybe you can flip meat for a living.

-- Anonymous, January 19, 2001

A question: Are you saying that you can have Big Macs or the Naked Chef but not both? In context: is it possible to grow a small development company into a larger one? What are the traps to avoid? What things need to be in place?

-- Anonymous, January 19, 2001

One programmer, one project? Fine, throw the methodologies out the window. Let creativity run wild. Get three programmers in an office and already you'll be in big trouble. Without methodologies, each programmer will create software that is completely different. This is far more frustrating than simply following the methodologies. What you should do is get everyone together every three months and review all of the methodologies. On a different note: If you hire sub-standard programmers, your company will fail somewhere in the future.

I find methodologies to be a pain.

-- Anonymous, January 19, 2001

It is not impossible to scale up McDonalds methodology. For instance Smith & Wollensky steakhouses http://www.smithandwollensky.com/ are as consistant as McDonalds, at the same time serving gourmet food. They have a huge operations manual and a consistent system for tracking and procuring ingridients that allows their manager to account for every single potatoe and 1/2 lb of meat. Of course they are not Peter Luger Steak House, but still very respectable.

-- Anonymous, January 19, 2001

What's being described is known in manufacturing as TQM - Total Quality Management. I think the examples miss a couple of points.

The first principle is that Quality is defined as conformance to requirements. Saying that quality of anything produced by methodoglies will be low is not true, but it is also an irrelevent statement. McDonalds creates a standard for what a burger should be, and then eliminates non-conformance to that standard. If they wanted, they could make the standard equal to gourmet food. It would be a lot more expensive, and most people wouldn't like it anyway. Does Jamie Oliver make better food then McDonalds? Some people will say yes. If I go to his resturant will the meal cost the same as a Big Mac? I doubt it.

The other problem is that writing software can become a whole lot more complex than making food. Could one programmer create Microsoft Excel alone? What if you had to do it in C instead of C++? Object Orientation is a methodology for programming. If you understand it, it makes your software more reliable and easier to create. If you don't understand it, you have a mess.

-- Anonymous, January 19, 2001

Just for the hell of it

Useless trivia: The Naked Chef was trained at Westminster Catering College--in the UK. It's the oldest culinary school in the UK. He has a band.

More useless trivia: Canada, Sweden, Norway, and Singapore came in top 4 in this year's Culinary olympics (hosted in Germany--it's a 100 years tradition) this year. There were 25 countries. Canada alone send in 200 cooks and chefs.

Even more trivia: Credible (but not world famous) cooking school in Vancouver, Canada cost around USD$5000 per semester. Normal computer science school cost around the same. Both fields pay less than people working in the waste management sector--on average. Unemployment rate for cooks and chefs is expected to be at 10% this year--as oppose to 5% or less in North America. Workers in both fields work long hours and little over time pay for decades at a time and have little chance of fame or owning a world-class scalable franchaise. Exceptions get all the attentions.

-- Anonymous, January 19, 2001

In the case of restaurants, at least, another route to cloning is to locate the clones where the customers will be grateful, even if it's not as good as the original.

Case in point: Wolfgang Puck's restaurants in the terminals at O'Hare airport. They cut down the menu to just the most popular, easiest, lowest-common-denominator items, like pizzas, and serve those. Compared to most airport food, it's much better. And tourists get the novelty of dining at Puck's.

How this translates to software, I've no idea.

-- Anonymous, January 20, 2001