Friday, July 17, 2009

Just What Is a Digital Millwright?

The name of this blog is a play on the title of a chapter in The Big Switch: Rewiring the World, from Edison to Google called "Digital Millworks". Author Nicholas Carr draws an analogy between the complex mechanisms that directed the power of rushing rivers to early factories, and the complex datacenters many companies maintain specifically for their own operations. The book, which I highly recommend for all IT professionals and business IT users, draws a historical analogy between the transition from local electrical plants to the public utilities and the coming transition from corporate data centers to computing utilities. Carr discusses the nature and workings of these emerging utilities and then transitions to the social impacts of "One Worldwide Computer".
As an IT professional for the last 25 years that has both developed software products for complex corporate datacenters and helped companies choose, customize and integrate these products, I think of
myself as a digital millwright.

Do any other IT users or professionals feel a kinship with an earlier or more mechanical profession?


  1. On Carr - Martin Campbell-Kelly has some thoughts on his thesis, in the May edition of the Communications of the ACM (I'd include a link but the article is available only to ACM members). A computer scientist and historian of computing, Campbell-Kelly notes that there have been at least two failed attempts to centralize computing, and that centralization is attractive only when the cost of centralization is lower than the cost of non-centralization. Basic stuff, really, but an interesting perspective on the current SaaS craze.

    I'm working through The Big Switch now, thanks for the recommendation!

  2. Patrick,

    Thanks for the comment. I read the article as well, and I have a number of comments:

    First of all, I don't think utility computing or SaaS is the same as recentralization. I think that most organizations will receive varying services from different organizations, either directly or through brokers, just as no two IT shops seem to use the same product set today. With the ascendancy of SaaS and other remotely hosted services, the probably of a medium to large business using As standards continue to develop the financial and logistical barriers to service provider diversity will be easier to surmount than the barriers to software supplier diversity.

    Secondly, the authors suggest that the prospects of hosted services are overblown. This is a common occurrence for all types of technology, and does not portend anything in particular for SaaS and hosted solutions. In fact, the Gartner Group regularly publishes hype cycles for groups of technologies that track where they are on a cycle that begins with a steep ascent to a peak of optimism about the game-changing nature of a particular innovation, but then market perception descends just as steeply into a "trough of disillusionment". But then, just as all seems lost, a steady re-ascent agrees, until the technology reaches a stable plateau of market acceptance.

    Thirdly, hosting services are subject to the same dynamics of any good or service that is required over a wide geographic area. As technology or other factors changes the actual and perceived cost and risk of remote production and long-distance distribution relative to the risk of local production and short-distance distribution, the market responds as might be expected. For example, factory farming and refrigeration made food a lot cheaper, but now consumers perceive certain social, environmental, and health costs and risks, and buying local is becoming more popular. Similarly some people are generating their own electricity using wind or solar power, and even selling it back to the grid. I think as certain computer solutions compete to have richer interactions with their users environments, there will always be a need for local processing and storage, especially as these resources continue to decline in cost, but for a given application, a boost in network bandwidth and speed might cause the pendulum to swing in the other direction. Consider the processing power on today's smartphones, for example. Also, consider the fact that today's WAN optimization devices increase local processing (protocol optimization and date encoding) in order to reduce the need for network resources.

    In summary, I think that almost technology "fails" to follow its original path, since market innovation and economic circumstances almost always intercedes.