I’ve just finished Cities by John Reader, on the history of cities, and it’s chock full of information and great stories.
This story, on Japanese manufacture, is lengthy but so good I have to quote it in full:
Bicycles were extremely popular in Japanese cities at the end of the nineteenth century, when the import of goods that Japanese manufactures could not compete with on price — or could not make at all — was damaging the national economy. Clearly, if bicycles could be made in Japan, both the massive demand for an individual means of transport and the national economy would be server at the same time. As Jane Jacobs points out in her book The Economy of Cities, Japan could have responded to this challenge by inviting foreign manufacturers to establish plants in the country — though this would have brought little profit to the Japanese themselves. Or they could have built a factory of their own — which would have required large investments in specialised machinery and the training of a skilled labour force. The Japanese followed neither of these options. Instead they exploited an indigenous talent for ‘economic borrowing’ — or imitation, as non-specialists would call it. It worked like this:
Not long after the importation of bicycles had begun, large numbers of one- and two-man repair shops sprang up in the cities. Since imported spare parts were expensive and broken bicycles too valuable to cannabalise, many repair shops found it worthwhile to make replacement parts themselves — not difficult if each of the shops specialised in making only one or two specific parts, as many did. In this way, groups of bicycle repair shops were in effect manufacturing entire bicycles before long, and it required only an enterprising individual to begin buying parts on contract from the repairmen for Japan to have the beginnings of a home-grown bicycle manufacturing industry.
So, far from being costly to develop, bicycle manufacturing in Japan paid for itself at every stage of its development. And the Japanese got much more than a bicycle industry from the exercise. They had also acquired a model for many of their other industrial achievements: imitation and a system of reducing complex manufacturing work to a number of relatively simple operations which could be done in small autonomous workshops. The pattern was applied to the production of many other goods, and underwrote the soaring economic success of Japan during the twentieth century. Sony began life at the end of the Second World War as a small shop making tubes on contract for radio assemblers. The first Nikon cameras were exact copies of the Zeiss Contax; Canon copied the Leica; Toyota Landcruisers were powered by copies of the Chrysler straight-six engine.
Here are the reasons this is great:
- It’s distributed manufacture, a network of independent units operating as a single factory, but in a more agile way.
- It reminds us that the idea of interchangeable parts is relatively new–and was a world-changer. It parallelised and distributed manufacture. Are we at the level of interchangeable parts in software yet? Despite common protocols like HTTP, I don’t think so, not quite.
- It points to an alternative to the mass manufacture and assembly line of Fordism. The parts can be accessed separately from the assembly, we can build our own neighbourhood factories for custom goods! Mass manufacture doesn’t imply treating workers like interchangeable parts too! What’s more, it bootstraps off mass manufacture and makes something different out of it.
The most exciting reason?
This pattern is happening, right now, in India with mobile phones. 100s of small shops repair and rebuild phones with generic components and reverse-engineered schematics, supported by a developed training and tool-production infrastructure.
How long before we’re seeing cheap-as-chips kit phones, assembled by entrepreneurs harvesting the market stands of Delhi?
5 Comments and Trackbacks
1. Henrik said on 9 January 2007...
While it is an interesting history lesson, I don’t understand the attempt to relate it to software.
In all areas of software development I have been involved in to task has never been to produce many identical pieces of software. I only see such patterns on a very abstract level such as,
* Produce an admin app as a Windows client working against an SQL server
* Produce an interactive brochure as a web site
* Produce a corporate news portal/document hub
* Produce a personal profile web page
In all these cases the process has definately been automated and streamlined by making dedicated development tools and software frameworks.
If you want to look more fine grained you have to look at the patterns for making and using a cog, rod, lightbulb, sprog, tire. The equivalent I can see atm is list, input field, table, field. Car manufacturing revolutions taught us that standardisation and reduction of inventory is vital. That might be applied to software by condensing the number of concepts developers/designers need to master. So in that sense I think we can learn something, but that wasn’t the period you described.
When it comes to phones we are back in the physical world and manufacturing is directly relevant. I wonder though if there are that much margin to squize on the labour side. Quality control is going to be a major challenge if you don’t have a homogenous labour regime.
2. Matt said on 9 January 2007...
I think the key idea from this period, for software, isn’t standardisation and automation (though you’re right to point those out as important). It’s the interchangeable parts, and yup, I should have been clearer about what I meant there.
On a macro scale, it’s still not really easy to swap out my email server for a different one, where all email server manufacturers compete for (say) reliability and speed. Or on a micro scale, perhaps I want all HTML rendering to go through my own, custom-written interchangeable parser, which would be doing its own special functions too.
If it was possible to piecemeal replace/upgrade Microsoft Office, maybe we could make a patchwork office suite out of the parts.
I think this was the promise of technologies like Taligent, OpenDoc and OLE (originally? I might be misremembering)… perhaps those ideas will come round again.
3. gavin said on 9 January 2007...
There’s more to it than just copying…. In both the Japan and India examples, whole new ways of doing business have sprung up in their post-copy/manufacuring histories.
In Japan, you’ve got Keirin bicycle track racing, with incredibly regulated part manufacturing and racing standards with vast sums of money changing hands in gambling. Don’t think Keirin would exist if it wasn’t for the local bike repair shops.
And Jan Chipchase has all those great photos of the Indian mobile phone “shared use” businesses too. I think he mentions someone figuring out a way to do mobile atm/banking over shared cell phones reliably and safely as well. Ripe for some emergent biz plans.
I love the tension between “manufacturing process” and “business plan.” Great stuff. Thanks.
4. JamesB said on 12 January 2007...
Great post and i love both Reader’s observation and your analogy with software. In many ways we have an engineering history that mirrors software development in that many of the smaller engineering companies are off-shoots from bigger organisations for whom increased specialisation became uneconomic [because of the need for flexible production and their inherent unflexible labour], rather allow talented individuals to leave and sub-contract out to them for their niche products. We’ve become victims of our own economic success in that products are now often more cheaper to replace than to repair in the West, so we’re missing out on craft skills and creativity that could provide a next gen industry. Hacking, a hobby rather than a need seems to be the only exception to this trend [in terms of products – we’re great at being ‘culturally’ creative!].
5. Henrik said on 18 January 2007...
Ok fair point Matt.
I seem to remember the same on Taligent/OpenDoc/OLE
Web Services, ActiveX Control, EJB are other attempts to accomplish this.
The main challenge seems to be clearly defining the product silo’s. After all a sproket with 80 teeth cannot be replaced by one with 81. What set of products could be used to create all software?
Unfortunately I think you quite quickly end up with a huge list of component specifications, and have a hard time cutting it down to a manageble size. Even then you wouldn’t cover all bases. I suppose that is what some might have said back then, but I really believe that software is much more complex.
Apart from Exchanges & Notes, E-mail servers seem fairly interchangable to me. But they are after all restricted by protocol standards, which obviously limits the possible differences. On the HTML front you can make your own Apache modules(not easy and not industry wide I know) on the server side and FireFox plugins(again .. I know).
It just dawned on me that distribution cost is a major influence as well. A small Indian shop cannot sell a product in Europe on their own, hence they have to join a network and adopt it’s standards. On the other hand a small Indian software vendor can sell to anyone, without adopting any standards.
The way I see it is that you are absolutely right, but until there are market mechanics that make standardised components the most viable way, it wont happen. People have tried to create a market place for components several times with little success.