It is not my intention to reopen the discussion about the controversial article from Nicolas G. Carr. I can understand the intention of the Author, and the reasons for his opponents to contest.
There is however a difference between other business-supporting utilities (like electric power) and IT in terms of business process support. Or to define it better – the difference is getting bigger with every year after 2003 when the original article was published.
The reason is the increasing scale, the IT systems are responsible for the processing (and continuity) of the organization’s business processes.
I could imagine the following stages in IT development in relation to business process support:
- Storage & calculators
- Functional support (i.e. document issue)
- Sub-process support (i.e. invoice processing)
- Complete implementation of the core business processes
In the early stage of IT, computers have been used mostly as storage and fast calculators. Computer failure caused the data to be retrieved and the calculations to be repeated.
The downtime costs were countable and adequate to the application area.
The same downtime cost calculation can be done with further IT development when IT implemented sub-processes (such as loan calculations and payment).
Nowadays we face a situation, where – in order to compete in the rushing business – we use IT systems as complete core business processors.
Additionally, we reduce or even eliminate human interaction through automation.
We don’t need to look very far for examples of the business advantage – Amazon AWS, Google, and Netflix, are companies whose core business is implemented totally in IT, and they don’t need manpower to scale (at minimum in the major income area).
Besides the social costs of such a business delivery model, which are not predictable and are widely discussed now (i.e. robot tax), there is a problem with the IT downtime cost.
The total cost of IT failure grows with the degree of business-from-it dependency. When an organization having critical business totally implemented in IT suffers an IT total failure it does not process its business processes.
When the downtime exceeds the business process lifetime tolerance, the business process chain is broken and the costs grow probably in a non-continous and non-linear way (I would love to see any reasonable research documentation, but as I noticed in my previous posts there is no science in IT by now).
Netflix identified this problem and invested a lot in implementing Chaos Monkey and afterward Chaos Kong – to prevent itself from business breakdown. It seems Google is going the same path.
So, we can believe, that commercial companies, whose core business is based on IT will prevent themselves from long-running IT breakdowns – we can call it a self-preservation instinct. And the automation and related business-from-IT-dependency are properly managed.
The bigger problem is in the utilities or public sector. Because in this area the costs – especially the cost-to-recover are not countable at all.
- Let’s imagine the total cost of IT failure in a nuclear power plant if the control of a kernel cooling system fails (almost happened).
- Or something from my professional garden – what is the total cost of long running pension payment system downtime? (in relation to downtime duration)
It sounds like s-f, but there is an important issue when copying the business implementation models from commercial sectors into utilities or public sectors. (Especially, when AI is going to replace, the still-existent but reduced number of human interactions).
To summarize, IT does matter and the more IT implements a critical business process, the more it matters.
When the IT implementation of critical business processes supports life processes or life-supporting processes, the IT importance is crucial. Such IT systems cannot be designed and constructed in a standard way.
Thinking about the future of IT it will be for sure one of the next coming drivers – the total cost of downtime and the preservation techniques.
For the public sector, I believe we would need law regulations in this area, and hopefully, they will appear before the first disaster and not afterward.