In my last post I looked at the first part of Kwan and West’s paper, which concerns the phases of technology adoption. Now I’m turning to the second part, which concerns the adoption model.
Kwan and West propose the following model for open source adoption:
The model has two decision points: the first where the choice set of applicable solutions is identified, and the second where a selection is made from the set available.
What I find interesting here is how strongly they position software policy in the first decision point. They I think rightly identify policy as being at least partly reactive; that policy is determined to some extent not by the intentions of the organisation so much as its past history (the feedback loop in the diagram). There is also the contextual influence of the culture of the firm and the industry it sits within.
Within UK higher education we have a reasonable data set for how open source is represented in software policies from 10 years of OSS Watch surveys:
To some extent “policy” is a little bit misleading in Kwan and West’s paper as they elaborate this as not just formal policies but also the “pattern of attitudes and decisions that constitute a de facto policy”, so could be said to encompass institutional cultural attitudes.
(In practice I’ve found that in higher education, policy is set at an organisational level, but cultural attitudes towards, and awareness of open source can vary by department or even by group within a department.)
The second factor influencing the choice set is the application context, which is where we refer back to the model of phases of strategic importance to determine the relative weighting of features, risks, and costs.
Finally, there is the set of available products that can be considered.
These three factors determine the choice set, which are then compared in the second decision point for selection.
In my experience the first decision point, of determining the choice set, is where many open source candidate solutions tend to be excluded; for quite a few different reasons:
- The process of procurement is primarily passive in nature, expecting suppliers to respond to RFPs at their own expense, a route that tends to favour suppliers of proprietary systems
- The organisation has low awareness of open source options, and does not look beyond its set of usual suppliers when identifying available products
- The procuring organisation has a set view on the composition of the solution, requiring a single product from a single supplier, for example, rather than considering an integration of different components to perform the same functions.
When we looked at the OSS Watch survey results for 2013, we found that a perception that “there isn’t a solution that meets our needs” was the overall number one supply-side reason given for deciding against open source, while interoperability and migration issues were the number one demand-side reason.
For me this indicates that there is perhaps something missing from Kwan and West’s model, or that perhaps the “policy” component is trying to cover too many different factors. For example, would a passive-RFP approach rather than an active discovery approach fit within “policy” or is it an influence on “available products”? Also, where do constraints such as contractual obligations and interoperability issues (particularly lock-in) reside in the model?
So, thats the first decision point. The second decision point is I think fairly standard, though here the “selection metrics” factor has some special considerations for open source that can be overlooked.
When I talk to organisations about procurement (with my OSS Watch hat on) I tend to talk a fair amount about sustainability, and in particular how to evaluate open source projects for their likely future viability. This isn’t necessarily a criteria used in traditional selection metrics, or if it is, it isn’t one which is easy to evaluate or have good data for.
Another aspect of selection is pre-sales activities such as demonstrations and pilot systems; again there is sometimes a problem here with procuring organisation’s expectations being conditioned to how closed-source companies tend to work, and this isn’t included anywhere in the model. I guess generally Kwan and West have operated from a presumption of general good procurement practice by the organisation, whereas in the context I’m interested in (UK HE) I’m very much aware of the existence of gaps between policy, process and practice, and the impact this has, and so can’t really exclude them from the model.
Finally, I think the model is suited to procurement of systems at all phases of strategic alignment apart from strategic systems; in these cases I think its more likely that a solution does not already exist, but has to be created in partnership with suppliers as a “pre-procurement” activity. However, as I’ve already pretty much concluded that there is no strategic IT in HE, its not something I need to dwell on.
I think Kwan and West have made a good stab at a useful conceptual model here, and in particular I think the strategic alignment phases concept elaborated in the first half of the chapter is a very useful one. The adoption model has some good features, but seems to miss out or have insufficient emphasis on factors that I know do affect selection practices, so it would require some further tweaking before I could make good use of it. The authors have done a fair amount of follow up work which I’ve also put on my reading list, and hopefully they revisited the model.