Stephen Freeman Rotating Header Image

August, 2007:

I'm a carpenter, not an architect

Jimmy Wales, interviewed by “The Oregonian”: (right at the end).

bq. Five years ago, I was just this guy sitting at home in my pyjamas, typing on the Internet like everybody else, and now I get asked these big guru questions about technology, and I have no idea. I’m a carpenter, not an architect.

(Attribution corrected by Aslak Hellesoy. Apologies and must get my glasses fixed.)

Almost Lean

I recently bumped into a McKinsey report on “Applying lean to application development and maintenance”: (free after registration).

There are some good points in the report, such as cleaning up the prioritization process, cross-training staff, and pushing quality throughout the process. They also recognise that it takes a while to convert, and that it’s important to look at the underlying principles rather than just copy someone else’s practices. The conversion of a large financial institutions sounds like a success (of course) and I’m sure it’s better than what went before.

What they’ve seem to have missed, however, is the people aspect of Lean—that the people who do the work should be intimately involved in improving the process. The tone of the piece sounds like good ol’ Business Process Reengineering. Managers “use” people and “tap offshore resources”, and they intervene when a developer’s work is running late (rather than providing a culture where the developers themselves ask for help).

Another doubtful aspect is the talk of rewards and incentives. Demming was very clear about the perniciousness of linking bonuses to specific targets, but that’s not a shift that’s likely to happen in most financial institutions. It’s also noticeable that the new “incentivized” metrics they proposed were to do with reducing waste, rather than adding value.

Finally (and I know this is only a short piece), they only talk about Lean Production—the perennial application of manufacturing metaphors to software development. There’s some much more relevant material in the Toyota Product Development process that is not recognised widely enough.

In the meantime, we have some Lean Software initiatives in our world that seem to have deeper understanding of Lean’s roots. Apart from “the Poppendiecks”:, David Anderson’s been writing up an interesting application of “Kanban”: to software development

Efficiency vs. resilience

Here’s a posting from Rod Van Meter to comp.risks. on 23rd July 2007. I’ll just quote it.

bq. By now everyone has heard of the M6.8 earthquake up in Niigata last week, a couple of hours north of Tokyo by shinkansen. Ten people were killed (all in their 70s and 80s, living in traditional-style houses with heavy ceramic tile roofs that collapsed), 6,000 homes and buildings destroyed, roads cracked and/or covered by landslides, a fault slip that came to the surface and displaced a section tens of kilometers long by something like a meter. Net effect was (if I recall) to push one plate 16cm north. […]

bq. One small company in Niigata, Riken (no relation to the research lab with a similar English name, I’m sure) makes 60% of the piston O rings used by *all* of the car manufacturers in Japan. Their plant was badly damaged.

bq. Japan’s auto makers, of course, are famed for their “just in time” supply chain management. I know people who have worked for subcontractors, and the penalty for being late in supplying a critical part can easily exceed $100,000 A DAY.

bq. Toyota was forced to idle at least 27 plants, Daihatsu four, Honda and other manufacturers several each. Toyota is still shut down, as of this writing (Monday, a week after the quake), and has an output loss of 46,000 cars or more. I haven’t seen a breakdown of the percentage intended for domestic consumption versus export.

bq. One interesting part of the response is that the auto manufacturers sent teams of their idled workers to Niigata to help Riken clean up and get back in production. They were there helping by Thursday, despite the transportation disruption, general shortages of goods including water, food, and electricity, and risk of aftershocks.

bq. One point and one question:

bq. * A disaster it is, but a relatively local one, in a mid-level city where events rarely make the world news. And yet it will affect car prices around the world, no doubt. Just one more data point that the world’s economy is one large web.

bq. * Toyota is a very well-run company, but they let this happen to them with an important single-sourced part. How good is YOUR disaster plan, whether personal or corporate? How good are your suppliers’ disaster plans, and their suppliers’?

Maybe the manufacturers had done their sums and figured that the cost of lost production from an earthquake was less than the cost of providing a safe alternative? In the meantime, we also should take a look at the risks inherit in our organisation: the one person who survived the layoffs who knows how to update the market data, the system administrator who works secret miracles to keep everything running, or the one person in the building who still knows C++.