Introduction

The acceptance and adoption of free and open source software (FOSS) is widespread and expanding exponentially across many industries. The use of FOSS is an attractive option today for many organisations and enterprises including for development, deployment, operations and strategy. Most organisations that use SW/I.T. today are certainly using FOSS in some form or the other, though the manner and extent to which FOSS gets used would vary greatly across organisations. There is no way of objectively assessing and stating the difference in the ‘FOSS Maturity’ or ‘FOSS Friendliness’ of organisations in a quantified and objectively verifiable manner. The present work has come up with a “FOSS Adoption Index (FAI) Model” that objectively computes a number using data gathered through online questionnaire.

The FAI Model and Approach

Our Model uses multiple ‘levels’ depending on the type and extent of FOSS –use data that an organisation may have with it. In the simplest case of the model, there may be just one level where a series of questions would be answered by the organisation which is used to compute a single index. A 2-level model would however collect more refined or fine-grain primary data from the organisation at level-2, whose computation would produce secondary data which would be used for computation at level-1 (which is above level-2) to produce the single index. If the organisation has data that can be further drilled down, then a 3-level model could be used where the primary data would be used at the 3rd and the lowest level. Etc. At each level, values of the indices are computed by assigning weights to the different criteria at that level, the criteria themselves being coarsest at level-1, and becoming finer as one goes down to the levels below.

The following definitions are used in building the model –

Criteria (Level-1) used in defining the FAI are indicated by the subscript ‘ i’, i = 1 to N(l) The set of N(l) level-1 criteria capture the most important factors or attributes that impact on the value of the FOSS Adoption Index (FAI) in an organisation belonging to the class ‘l’, and their selection is a vital part of the model building exercise. For each level-1 criteria ‘i’ for a class ‘l’ of organisation, the ‘level-1 criteria scores’ (Si(l) are calculated using the data collected from organisations of that class through a survey, and a weighted sum of these scores (Si(l)) defines the FAI(l), as explained below.

Criteria (level-2) indicated by the subscript ‘j’, j = 1 to Mi, are the finer elements to which each level-1 criterion ‘i’ is broken down for greater clarity and precision in data gathering. Selection of appropriate level-2 criteria for a given level-1 criterion is an important part of model design that impacts on the eventual usefulness of the model. ‘Level-2 criteria scores’ (sij(l)) are obtained from the data collected, whose weighted sum gives Si(l) as explained below.

Model Parametersαi(l) , βij(l) for a given class ‘l’, are the weights defined that eventually impacts on the contribution of a particular criterion on the FAI value of the organisaiton. The FAI, weights αi(l)βij(l) , and scores for level-1 criteria Si(l) are defined through the following operations (shown for two levels) :

(3.1)

(3.2)

With these definitions, the 2-level model looks as follows:
Data from online survey of the organization

In this study, the scores sij(l) obtained through survey questionnaire have been assigned a range of 0 to 10, which makes Si(l) as well as FAI also to have the same range; a low value of FAI(l) (closer to zero) implying very little adoption of FOSS in the organisation belonging to class ‘l’, and a high value (closer to ten) meaning high levels of FOSS adoption and FOSS friendliness in that organisation.

The key aspect of this modelling exercise however is the proper choice of the criteria at multiple levels that would cover all the areas of an organisation. The other key aspect is correct assignments of values to the model parameters αi(l), βij(l) and are discussed below.

Choice of criteria at multiple levels

A criterion captures the factors or attributes that impact on the value of the FOSS Adoption Index (FAI) in an organisation. Those factors that most significantly affect the adoption of FOSS in an organisation should be selected as the ‘level -1 criteria’ and those factors that further helps in measuring a ‘level-1 criteria’ in more details should be selected as ‘level-2 criteria’. In large and complex organisations, it may be necessary to drill down to still finer scales of granularity by defining ‘level-3 criterion’ etc. Our model is open and extendable in this sense, with the definition of the Index for a 3-level model given as below:

(3.3)

Where are the weights associated with level-3 of organisation 'l' and sijk(l) are the scores obtained through the survey. What ultimately decides the number of levels to which one might want to drill down is the ability to gather meaningful data at finer and finer levels from within the organisation.

Choice of weights at multiple levels

Choice of weights or model parameters αi(l), βij(l)etc associated with each level is another critical decision needed for working with our model. The logic of such choice that works at one level also works at other levels of the model, and we may illustrate it with a level-1 situation. Taking the example of the criterion “FOSS Policies and Guideline” once again, the value of its score contributes to the final Adoption Index value through a weighted summation – the score Si(l)appears as multiplied with its weight αi(l), and not by itself. In other words, the extent of contribution that this particular score makes to the final Index value is also determined by the associated weight value – a low weight value reducing its contribution and a high weight value enhancing the same.

Calculation of level-2 scores ‘sij(l)

For a level-2 model, the calculation of level-2 score sij(l) is the most crucial exercise and has to be carefully done. For measuring the usage and adoption of FOSS at level-2 a series of questions are framed that cover the particular criteria. Each question is assigned a score and a summation of the scores for all the questions pertaining to a particular level-2 criterion gives the score sij(l)’. This score along with the weight assigned βij(l)are used for calculating the Si(l) and FAI(l) and can be seen in equation 3.1 and 3.2.