3 Method

The approach of this thesis is qualitative, a methodical stance focusing on the more intangible qualities of the research topic. The alternative would be a method focusing on quantification of research results by using for example extensive empirical studies and statistical methods. There are also other differences between qualitative and quantitative methods [Starrin 94]. In this context, two main issues regarding the method demand attention: precision of measurement and objectivity of the results.

3.1 Precision of measurement

Using a qualitative approach mostly means that results are not easily measurable. This is also true for this thesis, since the aim is to find the more or less abstract qualities of client-side proxies under different conditions.

It might be possible to measure some of the results with acceptable precision, for example by providing statistics regarding impact on network efficiency when using proxies, accuracy of filtering proxies, the number of users of different applications, etc. If it were the main goal to answer these or other quantifiable questions, a quantitative approach would be preferable.

However, the goal is to answer questions that are more general, such as when the examined approach is preferred over other solutions. Because of this, a quantitative approach would not suffice. Inevitably, using a qualitative approach means that the results will not be thoroughly validated or invalidated with empirical or statistical methods, but this is not uncommon for this kind of research.

3.2 Objectivity

Since the results are not easily measured, there must also be doubts regarding their objectivity. It is true that answers to questions about the relative qualities of a specific technique are rarely objective. The whole field of software design mostly depends on the notion of good practices, rather than on fixed truths and objective evidence. Only in low-level areas of research, such as determining the efficiency of specific algorithms for particular tasks, is it possible to obtain truly objective results.

Obviously, this work does not deal with this kind of low-level research, so there can be no claim that the presented results will be truly objective. Where some kind of conformity with the current ìtruthsî is desirable, the evaluation will be coloured by the generally accepted good practices of software design. However, a large part of the work will be dependent on rather subjective interpretations of the design and performance of the examined systems.

3.3 Method in action

The main part of the work is an open-minded evaluation of existing client-side proxies aimed at finding the qualities of the technique. As mentioned above, this evaluation will be based partly on what is accepted as good software design but mainly on more subjective perceptions of existing solutions and hypotheses regarding the potential qualities of client-side proxy applications.

Through a comparison of proxy and non-proxy solutions, the goal of the first phase is to identify the circumstances when a proxy solution is better and which factors speak in favour of using them. Based on the issues raised in the background survey and the results of the first evaluation round, the second part of the evaluation will focus exclusively on existing client-side proxies. The aim is to investigate to what degree they realise the potential of the approach and to find ways to improve them.

Together, these two phases will give some possible answers to the introductory questions, some of which will be visualised in the Blueberry module. In total, this will provide results that admittedly are not final, but should be a useful starting-point for further evaluation and serve as a set of guidelines for those interested in the approach.