Innovation and model boundaries

When designing an information system, a piece of software or a law, experts work from a model. This model must have boundaries. When these boundaries are violated, life may become difficult:

  • After over two years in lock-out, the journalists of the largest Montreal newspaper decided to bow to their bosses and accept whatever they could. What happened? In Quebec, we have a strong anti-scab law. In theory, therefore, you cannot publish a newspaper without your newsroom. Except, of course, that the law assumes work happens in the offices. In practice, anyone can work for a newspaper from home, sending documents by email. The boundary of the work model in this case is the (physical) office. Entrepreneurs are smart enough to realize that they do not need the conventional journalists, working in the newsroom. Alas, consumers are now realizing that they do not need the newspapers in the first place. We have broken models all the way up.
  • Email was designed to work within a small and  non-commercial network. It was never meant for the world we live in. And I attribute much of the rise of social networks (e.g., Facebook) to the need to recreate these small non-commercial networks. The irony, of course, is that people who are investing in Facebook right now tend to forget that the boundaries of the Facebook model have much to do with avoiding spam.
  • Entity-relationship models were designed in an era when corporations had few databases which were relatively small and simple. Moreover, you could isolate various databases: there was no need for your databases to ever be compatible with the databases of another company.   Today, many corporations have databases with hundreds of attributes, all designed by different experts. The semantics can vary widely within the corporation. What is worse is that you must often reconcile the semantics across several companies inside supply chains. The net result is that if  you wanted an actual entity-relationship diagram which represents what is actually going on, it would never fit in anyone’s head. We have done much work on schema evolution and data integration but we fall short of solving the problems without pain.

The boundaries of our models are constantly being broken. These broken pipes represent vast opportunities. A new CEO who wants to turn a company around should look for breached boundaries in the company’s model. Competitors should look for broken water pipes as a way to identify soft targets.

I believe this calls for an entirely new field of expertise. Some people should become experts are recognizing broken models.

Here is my favorite broken model: E-commerce is pathetic and inefficient because it is based on the pretense that we need stores. Like we do not need newspapers (just news), we do not need stores (just products). In the nineties, people were talking about software agents that would go out and shop for us. Why do I have to do so much manual labor to find the best product and the best price? Sometimes, it almost looks like sabotage. Consider this scenario. You go see your doctor. Instead of handing you out a prescription that you will need to present in some drug store, the doctor just tells you “the medicine will be delivered to your house tomorrow morning, it will be automatically debited from your insurer’s account”. See? No store. Ironically, the future of e-commerce might be to do away with stores entirely. Oh! And it won’t be invented by people running e-commerce stores.

Daniel Lemire, "Innovation and model boundaries," in Daniel Lemire's blog, February 28, 2011.

Published by

Daniel Lemire

A computer science professor at the University of Quebec (TELUQ).

2 thoughts on “Innovation and model boundaries”

  1. Amusingly, I often go to stores to look at things (like laptops) that I plan to get online because of the lower prices.

  2. @Steven I think it is a common experience. It has become easy to deal directly with many manufacturers (think: Apple or Dell) and it will just get easier over time. Ultimately, cutting the middle men reduces costs. Of course, you need a payment infrastructure (think:VISA) and shipping infrastructure (think:UPS). But we have that already.

Leave a Reply

Your email address will not be published.

You may subscribe to this blog by email.