Starting research

Setting out a research agenda is hard. I should know. It's possible (probable) that if there's the freedom to spend time doing any research at all -- if the effort it's not results-driven -- then the initial directions are hard to define. That can lead to difficulties in making the final stages of the work yield value for the numerous stakeholders. These difficulties are endemic to Masters and Doctoral theses but are common in "higher", team-oriented, research efforts too.

Some work and proposals flowing near me have crystallised some views and criteria, all of them pretty obvious and well-known:

• In the beginning, seek to define the terminology and knowledge landscape under study. This establishes a beginning place.
• Have a working hypothesis (Proposition A), and work for or against it (transforming it into Truth A). This is an initial direction. Even though it may be a stalking-horse, it should be conceivable that demonstrating the hypothesis could be a goal of the research.
• Tools (e.g. Tool A) you acquire to deal with knowledge and statements shouldn't be an end in themselves.
  
• Be prepared to abandon the hypothesis if it looks inappropriate to what can be found. Don't proceed without forming a new one (Proposition/Truth B).
  
• At a later stage, you might be fed up with hypotheses (too many of them, self-evident by now, etc) and may wish to progress towards demonstrating a usable product (Tool B). That's fine, but again, don't abandon the hypothesis/es until you can define B's characteristics.
• Keep defining the targets of the proofs and products in terms that the audience of the research will value (this does not necessarily mean that ready Applications are always required).

What's most important? If you come away with a Tool B, then that's something in itself. Then, less vitally, Truth B, then Truth A. Tool-A's are incidental (but creditworthy). The whole thing is creditworthy if the above paradigm is followed adeptly.

This article has turned into Marking Research, which is as it should be, since Specification must always have its own Evaluation in mind.