Prophesying on the edge: The future of collaborative software development

The CSCW ’12 conference, held in Seattle, WA, 74 miles apart from Victoria, BC, was a blast. Or, so I heard. I didn’t attend. I couldn’t attend. I belong to a different cultural species, at a level where preponderant cultures impose visa restrictions. Nevertheless, I had a paper accepted at the Future of Collaborative Software Development (FoCSD) workshop, whereby experts were invited to talk about the future. My paper, titled “Embracing Distributed Work: Why Distance Shall Matter (Even) Less,” (PDF here) coauthored with Adrian Schröter and Daniela Damian, talked about the importance of distributed software development in terms of its inherent benefits vs. the benefits of collocated development. In brief, FoCSD poster for the benefits and my prophecy for the future: The benefits of distributed work comprise a whole larger than the sum of its parts; thus, it should be embraced without any local hesitation. Here’s Adrian’s post on this and another FoCSD paper where I appear as a coauthor.

But the speculation per se is not the real issue when it comes to academic venues. What really concerns me here is the actuation of such workshops, whereby experts talk about the future using scientific principles, such as dialectics, statistical inference, etc. The entire process of predicting the future is not only erroneous in its intent, but also reduces to intellectual poverty.

The field of CSCW entails social and cultural research because, after all, studied actors include human beings and their ecologies. One major argument for talking about future trends in social sciences is exactly the presupposed abundance of historical data. That is, social scientists intending to predict the future assume a historicistic approach with respect to a social and cultural process, such as distributed software development. Yet, as Karl Popper has argued in his book “The Poverty of Historicism,” such attempts are doomed to philosophical, intellectual, and actual failure, most certainly. This is what I meant by intellectual poverty above.

Furthermore, in less philosophical, but more realistic, terms, Michael Shermer has nicely articulated in a Huffington Post article, why we, the expert prophets, are merely wrong. According to Shermer, no matter how many existing studies one quotes to talk about the future, one has confirmation bias, “which our brains employ to reinforce what we already believe while ignoring disconfirming data.” If the rebuttal here is that such academic venues as CSCW are highly specialized and focused on such research, then this indeed “increases confidence, but it also blurs dissenting views until they are no longer visible, thereby transforming data collection into bias confirmation and morphing self-deception into self-assurance.”

While I refrain from considering my CSCW FoCSD paper as a prediction to future trends in distributed software development, I believe the Popperian criticism of historicism and Shermer’s nice recollection about the futility of predictions do apply in my case. It is this wit which may have probably affected some FoCSD organizer, who decided to not make the FoCSD proceedings available to the public.

Wise move to save oneself :).

Open to debate: Social research and formal methods

Here’s an open-ended question regarding social research and the researchers’ reluctance to firmly resort to theoretical strategies beyond merely empirical ones.

McGrath states that social psychology research has had a limitation in its history—it hasn’t given full consideration to theoretical bases. Also, most of empirical research we read nowadays uses certain empirical methods without a particular guideline as to which to select and without any generic reference to theoretical strategies (formal methods).

The question that arises naturally seems dichotomous: Is such reluctance a result of the lack of substantial knowledge and/or expertise among social scientists, or is it because formal methods are too abstract to apply on or devise from human systems, including the CSCW context of such systems?

I believe that formal methods, while not generally sufficient on their own account, are fully necessary for social and empirical research. Your thoughts…?

Methodological considerations

McGrath’s classical treatment of the social research process elicits several intrinsic facts about conducting empirical research, such as the individual strengths and drawbacks of each available/new method, or the collective strength of at least two methods employed conjunctively to tackle some research question.

It is important, however, to distinguish that empirical methods in software engineering and the social sciences that deal with human systems, for that matter, are susceptible to converging towards dismality unless at least one formal method (theoretical strategy) is also employed. McGrath alludes to a similar perspective on p. 159, wherein he argues that the methods of Quadrant IV are crucial for philosophical reasons (“research is scientific”) and empirical ones (“theories are built from such observations”). It is a relief to read such an affirmation because, in turn, it does affirm the fact that science is only meritocratic in its entirety. This view becomes a little too misty when jumping to social science research, wherein conclusions along different studies using different methods for the same problem may not consistently converge.

If the overall objective of social research is to elicit truths about human systems, then Russell’s “recommandation for future generations” resonates in my mind: “If something is true, you should believe, otherwise you shouldn’t. If you can’t find out its truth, you should suspend judgment.” Can social researchers afford to suspend judgment when the methods they have employed have yielded inconsistent results or conclusions for whatever problem they’ve tackled?

(One note of correction, given that we’re philosophizing on science: When describing what computer simulations are on p. 159, McGrath mentions that the designed model is “complete and logically closed.” Such a model, alas, cannot exist because co-existence of completeness and consistency is precluded by Gödel’s Incompleteness Theorems…)