| Our Services | Resources | News | About us | Search | Contact

The Macer View

Comment and analysis from Tim Macer
Number 1, September 2003
Research magazine

Understanding the meaning of 'open' software

Open software is a good thing, and fortunately, it is increasingly prevalent in research. Companies that have made the effort to prize open the back door to their interviewing systems or tab products, announce it with pride. But beware, for the term basks in the twin qualities of being both optimistic and vague. ‘Open’ may mean no more than the inevitable consequence of developing with industry standard tools such as Microsoft SQL Server. Given your own copy of SQL Server, it is not hard to peer into the murky depths of someone else’s database software and fish out something that might be useful for another in-house system to bite on: accounting data, project management, reporting and so on.

Of course, this should be the virtue of open system architectures - a place where program shall speak peace unto program and the old monolithic world of software banished. These good intentions need to be backed up with the discipline of adhering to standards, and where standards do not exist, rigorous consultation and co-operation.

Developers must be committed to change these open parts rarely, if at all. Incompatibilities introduced with each release mean misery for the user, as applications exploiting this openness grind to a halt and demand costly and unplanned interventions to fix. A vicious cycle can develop where customers, wary of new releases, stick limpet-like to the one they have; developers find themselves supporting multiple previous versions and only new customers get the privilege of ironing out the bugs in the latest software.

There is a fear that developers may use openness and the ability to customise as an excuse for bringing out incomplete software. We are witnessing a move towards the toolbox release, as developers create application programming interfaces (APIs) -libraries of programs you use to build your own programs. FIRM (see review) has one as does SPSS with its Dimensions platform.

Programmers love APIs as they can use them to whip up a new program in an afternoon. For companies with in-house development resources, this is real progress. Yet some SPSS customers fear their supplier may opt out of building heavyweight applications, and supply lightweight ones, expecting customers to ‘finish them off’ using the API. SPSS continue to say this is not their intention.

Many firms approach my office seeking advice on the ‘best’ software for them. Invariably, a complete off-the-peg product is the goal. A few alterations may be OK if they don’t take too long or cost too much. What research firms and research departments don’t want is a bag of wool and a knitting pattern.

It begs the question of just how many customers there are for this new generation of middleware. Open software is good to a point, but not when it is so wide open that you can no longer see the application.

Published in Research, the magazine of MRS (The Market Research Society), September 2003 , Issue 448.
© Copyright Meaning Ltd 2003. All rights reserved. Reproduced with permission.