In discussing methods to assure the quality of VGI, Goodchild states that; “The degree to which such triage can be automated varies; in some cases it might be fully automatic, but in other cases it might require significant human intervention.” In VGI, the source of the data is human (as opposed to a scraping algorithm in SDM, for example), but the verification of data quality would definitely benefit from automation to deal with the large scale of geographic data that is produced everyday. He goes on to say that “Some degree of generalization is inevitable, of course, since it is impractical to check every item of data”, but by using the data analysis tools that have been developed to deal with large datasets, researchers can strive for a more complete assessment of accuracy.
To reintroduce the concept of positivism in GIS, Goodchild states that ” Our use of the terms truth and fact suggest an orientation towards VGI that is objective and replicable, and for which quality can be addressed using the language of accuracy. Thus our approach is less likely to be applicable for VGI that consists of opinion….or properties that are vaguely defined” This position seems to indicate that only quantitative or objectively measured geographic phenomena are capable of being tested for accuracy/uncertainty. I find this a flawed position because of the strong explanatory power of qualitative GIS and alternate ways of measuring attribute data. In suggesting it is not possible to apply the same rigorous standards of accuracy to these methods, the implication is that they are less scientific and worthy of merit. Even if this is not the intention, I would have appreciated some suggestions or potential methods by which to ascertain the accuracy of VGI when applied to qualitative GIS data.
The three definitions of crowd-sourcing provided by Goodchild describe its different applications, from “solving a problem”, to “catching errors made by an individual”, to “approaching a truth”. This progression appears traces the familiar role of GIS as a tool, tool-making, or science. It is interesting to note that the third definition does not converge onto a truth as observations approach infinity, but rather that after 13 contributors, there is no observable increase in accuracy for a position contributed to Open Street Map. This suggests that unlike a mathematical proof or principle which will always be proven true given the correct assumptions, the VGI phenomenon is messier and has to account for human factors like “tagging wars” born out of disagreement about geographic principles, or the level of “trust” which may discourage someone from correcting a contribution from a reputed contributor.
The social approach tries to minimize the human errors mentioned above by quantifying variables like “commitment” and “reliability” and allowing for social relations amongst contributors to act as correction mechanisms.
-FutureSpock