Wednesday, August 17, 2005

Bus Lane Woes

Blogging can be fun. It’s even better when you don’t have to do the work. Today’s blog is a guest posting from Martin Geddes. Martin lives in Edinburgh, Scotland, where he resides in downtown walking distance from the beautiful city centre. He doesn’t own a car, so when he flies off to see his top tier clients on other continents, he has to take a bus to the airport. Here is his story…

I was on the bus to the airport yesterday - front of the upper deck.

Two-lanes on each side of the road. There’s a bus lane that starts about 70 yards before some traffic lights, and continues on after the lights. We’re about to reach it. It’s rush hour to go home, and the car traffic is stacked up in the right (non-bus) lane. Cars are in the left lane which is about to become a bus lane (usage change posted on signs along the road), and are trying to pull right into the car lane. But they can’t because it’s jammed. So the bus is stuck behind a car which has miles of green-painted empty bus lane in front of it, but isn’t allowed to use it. The lights are green, but soon go red. We’ve missed a cycle.

We almost missed the green light second time round because the cars couldn’t clear the bus lane quick enough. 50% loss of capacity. All because of a policy-based system constraint. The solution? Align the policy with the overall system throughput goal—cars can enter a bus lane at any time of day, but will be punished if you ever prevent the progress of a bus during the rush hour. Put the monitoring equipment on the bus—just a video recorder. Simple!

Tuesday, August 16, 2005

Definitely Not Mickey Mouse

I have to congratulate the organizers of the PMInAction event - the PMI Orange County Chapter - for a 1st class event. It definitely wasn’t Mickey Mouse despite the fact that we were hosted at the Anaheim Community College and the speakers were lodging at the Holiday Inn adjacent to Downtown Disney and right outside the Disneyland amusement park. Hal Macomber, Greg Howell and I had a hearty dinner at one of the Disneyland restaurants and talked agile and lean project management over a few margaritas. I thought that might be the best bit of the weekend but I was wrong.

You might ask, what is a conference for project managers run by project managers like? And the answer would be, well organized! If ever there was evidence to underscore Peter Drucker’s observation that volunteer organizations have highly motivated people then this was such an event. I was picked up at the airport and dropped at my hotel. I was picked up at the hotel the next morning with my fellow speakers and delivered to the venue 10 minutes before breakfast officially opened. At the door of the venue I was met by my speaker liaison. He would then escort me, personally - the other speakers had their own liaison - to the speakers’ ante room. Everything from my point of view ran like clock work. Meanwhile, I’m still getting a kick out of the idea that members received a PDU for their PMP to sit and hear me suggest that they stop estimating and start agile estimating.

Here are my slides for download. As usual these days they are short on bullet points and big on pictures. You have to hear the words. I believe that local PMI members can buy the CD from the organizers.

#1 The influence of W. Edwards Deming on Project Planning and Tracking in MSF v4.0

Traditional project management success is measured as conformance against plan. W. Edwards Deming taught that conformance to process and his theory of Profound Knowledge were preferable to conformance to plan or specification. He believed that focusing on process, productivity and variation created a culture of continuous improvement and ultimately led to better economic results, whilst conformance to plan encouraged heroic effort and a lack of repeatability. Deming’s work is widely admired and implemented in manufacturing and production processes. How might it be adopted into the project management body of knowledge and how would it affect the way we manage and run projects. Microsoft has adopted Deming’s thinking into its Microsoft Solutions Framework (MSF) for CMMI Process Improvement methodology. This talk will explain how conformance to process is reconciled with iterative project planning, tracking and reporting and how project managers can avoid making what Deming called Mistake #1 and Mistake #2.

[Download in PDF 2,670 KB]

#2 Focusing on Bottlenecks: How flow, variation and constraints affect project scheduling

Eli Goldratt has long since argued that traditional Gantt or PERT project scheduling is flawed. His Critical Chain scheduling method is based on his Theory of Constraints which first grew to prominence in alleviating bottlenecks in manufacturing operations. It’s poorly understood in project management circles that Goldratt’s manufacturing solution, Drum-Buffer-Rope is a pre-requisite to Critical Chain and that modeling a project a flow through a set of functional operations is essential to understanding the power of Critical Chain. This presentation will explain the theory of Agile Management for Software Engineering and show how through the incorporation of Lean queuing theory to software engineering projects, it was possible to enable both the Drum-Buffer-Rope and Critical Chain solutions for use in software projects. Microsoft has adopted the use of Lean cumulative flow tracking and reporting into its new Visual Studio 2005 Team System Product and the Microsoft Solutions Framework v4.0 methodology enabling the use of Goldratt’s theories, for better economic results in software projects.

[Download in PDF 2,110 KB]

Monday, August 15, 2005

Operating at 20% of Capacity

I’m in the mood for updates this week. So here is another one.

Today I biked to work. Nothing amazing in that. However, what is odd is that I haven’t done it for 7 weeks. For 5 of those weeks I was away - not working, not in my office, not even on email much. So today, I was getting back into the habit for the first time in a while. I intended to bike only to Montlake (about 8 miles) and take the bus along the SR-520 across Lake Washington to Redmond. However, when my bus, the route 545 due at Montlake at 0740, arrived, it was fully loaded with 3 bikes. So I waited a bit longer.

And now for a quick aside on exploiting a constraint…

The King County Metro really understands constraints - I’ve blogged about ways I see them exploiting constraints before, a couple of times. So here is another one. Each bus has a two bike rack on the front. [Recently, they’ve elevated the constraint by adding triple bike racks to the 545 buses, but we won’t get in to that.] However, during the nice weather of the summer months, long queues of cyclists form at the bus stops at either side of the lake. It can be frustrating as bus after bus goes by and is already fully loaded with two bikes. So what to do?

Well it turns out that a lot of the rolling stock gets displaced. Particularly in the morning rush hour, the metro bus company has to move buses across town. At certain peak times, as many as half the buses crossing Lake Washington on the SR-520 are out of service heading for an East-side bus depot. So, the metro company exploits the SR-520 floating bridge bottleneck by instructing drivers of out of service buses to stop for cyclists and transport them across the lake. Regular passengers aren’t allowed to get on. It’s a classic exploitation strategy. It keeps the line of cyclists to a minimal level.

However, this morning, I didn’t want to bike from Evergreen Point to Microsoft’s Redmond campus - another 6 miles. I wanted to break myself in gently. So I waited as several out of service buses passed. Then another in service bus came. Again full. More than 10 minutes had gone by. Another out of service bus. Another two cyclists got on it. Looking at my watch, my resolve to wait broke down. I decided to bike the East-side. Right behind that bus was another. I waived it down and boarded. As we emerged from the bus stop into the freeway traffic, we were only 6 vehicles behind the bus in front.

Regular readers will recall my earlier post on the SR-520 bottleneck. You may even have read Glen’s comments based on his undergrad experience as a physicist. I didn’t know what to make of Glen’s comments at the time and didn’t reply. What I do know is that interesting problems attract physicists who think they can solve any problem. It’s no accident that Eli Goldratt is a physicist. I also assumed that the physicists had gone away failing to solve the problem. They had been attracted to something more solvable. I was also amazed that the throughput-based line of thinking didn’t have more traction. Well it turns out that it does. And as James Surowiecki explains in The Wisdom of Crowds, there has actually been a long academic rivalry between the physicists, who want to model the problem as grains of sand flowing down a pipe, and the engineers, who want to model it as a throughput problem where the optimal throughput is achieved at 30 cars per lane per minute - a takt time of 2 seconds. [So, it’s no coincidence that I’m an engineer and that’s how I saw the problem.]

So remember - the optimal throughput of the SR-520 floating bridge across Lake Washington with two lanes in each direction is 60 vehicles per minute. This is achieved with a gap between vehicles of 2 seconds and is true regardless of the speed of the vehicles.

So my bus set off in the busy rush hour traffic across the lake - just six vehicles behind the other bus, which also contained two cyclists. When we got to the other side and stopped in the bus stop, the other bus was nowhere to be seen. I got off, unloaded my bike, and set off up the bike path expecting to find the two cyclists in front of me. Nothing! I certainly wasn’t fast enough to catch them. I estimate that the other bus had gained a full minute on mine, despite being only six vehicles in front. So do the math! At optimal throughput, that bus should have had a 12 second lead on us. It should have been in the bus stop while I was unloading my bike. However, variation in traffic flow had increased that gap to around 1 minute. The SR-520 at 0750 in the morning was only operating at 20% of its theoretical capacity.

The Wisdom of Crowds does leave us with some hope. Page 156 reports that academic researchers Helbing, Treiber and Huberman have developed two simple schemes which would greatly increase the flow of traffic. One of them, interestingly, only needs to be deployed in 10%-20% of vehicles in order to become effective. In other words, only 10%-20% of drivers need the extra information enabling them to make more informed driving decisions and magically traffic flows much better. So there is hope! [I wonder whether the local city planners currently intending to completely replace the SR-520 bridge with a bigger and better version have considered a tax subsidy for this new equipment as an alternative? The theory of constraints teaches us that it is better to first invest smaller amounts in exploiting the constraint than a very large sum in elevating it.]

Sunday, August 14, 2005

Raising the Loaf

A quick update to my earlier story on the $9 Dollar Loaf. As if to underscore the need to embrace change and stay current, that story is already out of date. A source close to the action sent in this report over the weekend…

I just got back from the U District market. Peggy has gone from the ridiculous to totally insane. She is raising the price to $15/loaf, and then to $25 on 1/1/06. The stuff sure does keep the bowel movements regular, but Geez!

It would appear that Peggy is giving her customers sufficient warning so that they can arrange a home refinancing in order to continue enjoying her artisan crafted delicacies. I wonder what Seth would make of that?

Wednesday, August 10, 2005

CMMI Webcast Scheduled for August 18th

Do you want to hear me present our MSF for CMMI Process Improvement material live on the web? And then followup with a chatroom session immediately afterwards where I will answer all your questions? Then be sure to tune in to our webcast at 11am Pacific Daylight Time on August 18th.

Abstract

Microsoft Solutions Framework (MSF) for CMMI Process Improvement is a highly iterative, adaptive planning, agile software development process which meets the requirements for the Software Engineering Institute’s (SEI) Capability Maturity Model Integration (CMMI) level 3 and provides a smooth transition all the way to level 5. In this Webcast, David Anderson (PM for MSF for CMMI Process Improvement, author and award-winning blogger) will introduce this feature Visual Studio Team System. Immediately following the Webcast, there’ll be an online chat with David and other members of the MSF team.

[Update: To register click here]