In my opinion, there is only one main question when solving problems through computers. Before I tell you what I think that question is, let us first define IT as two fases:
- distilling the essential from the information stream
- building a generic solution that will remain useful
Fase 1 depends on recognising the abstract behind the deluge of data. A person experienced in defining requirements, seeing patterns and creating control points (parameters) can do this better than any automated system.
The automated result of fase 2 should save a lot of time, attention and nerves compared to manual inspection and actions. A machine parsing symbols, applying well-defined rules and crunching numbers can do this much faster than any human.
So, proper application of IT is using the power of people combined with the power of computers. The human should analyse the problem and define the template and parameters. The machine then carries out calculations at super-human speed. When the program cannot handle a configuration, we go back and exclude (or maybe include) that exception.
Sometimes the problem at hand requires common sense and practical innovation that solves only the core issue.
When there is time, it pays off to look at the problem sideways, and deliver a set of templates for future solutions. Here are a couple of examples from my experience with Fase 2:
Generating XHTML + CSS + SVG + patterns/backgrounds, then finetuning
Fast set up for the web, freeing time for finetuning design. And you get to play with Scalable Vector Graphics, minimizing transfers of binary images (but watch out for browser-specific SVG quirks).
Next step: a styling validator for checking golden rule proportions, colour matching and readability of a web page.
Document Generation through XSLT
I built a structured set of XSL Transformation that received XML input (from a financial business framework) and created PDF output. The document templates + database containing variable data + transformation scripts can produce output almost instantly. The only weak points can be the XML structure (checks on the structure can be automated) and incorrect data (checks on the data can be automated to a degree).
Custom Automation within CRM
Working with the combination of Microsoft CRM, SQL Server (+ Reporting Services) and Sharepoint, one often has to choose to implement a manual fix to a problem, or to set up a generic solution. For instance, in CRM it is possible for users to create workflows that automate checks, actions, and warnings. Customizations of entities can be exported to XML, either in order to extract the definition of the system, or to change definitions of fields and filters in semi-automated way. As soon as we have the system definitions in XML, we can parse and change it automatically.
Controlling 3D Generation through Parameters
I wrote a VRML generator in C++ that non-IT users could direct via parameters. The goal was to create urban environments from scratch that were geometricaly valid and acceptable to the human eye (as realistic). Ultimately, my aim was to replace the waterfall-like normal 3D modelling process for non-specific content.
The modellers would create or acquire specific content such as a statue, and then complement this by choosing out of a wide array of generated 3D. Directing this generation would be a matter of trying out different sets of parameters and choosing results based on visual inspection, a much more predictable process than depending on creativity while under time and budget constraints.
The reality of IT is often working under the contraints of frameworks, time, budget, and agreements. These constraints often exclude any real generic solutions, and so efficiency is reduced to human manual action speed. Which can mean that the net effect of an IT system is only 10% higher efficiency, when it could be 200%.
In order to achieve more generic solutions and align business and IT it is beneficial to keep asking: how can we harness the main strength of humans and the main strength of computers?
