Petabyte Age negates scientific method? Rubbish
This is the kind of cockamamy claim that infuriates me. The author of this article in Wired Magazine is its editor, Chris Anderson, who also coined the term "long tail" as a contemporary business model. Wired has often published provocative articles, but I think it has long tailed out.
Chris' claim is that with petabyte storage (yes, that's lots), "we can stop looking for models. We can analyze the data without hypotheses about what it might show. We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science cannot." The patterns found are the emergent model, so this is just rubbish.
Chris clearly has not actually done academic research. If he had, then he would know that it is the research question that defines the methodology, not the subject or the capacity of your computer. There are already many approaches, especially in social studies where meaning is sought, which lead to qualitative methods of data collection (ie. they need human recorders) from which patterns are discerned. Grounded theory is a well-established methodology that generates explanations through a structured inductive process from focussed qualitative or quantitative data. Other research questions are precisely designed to validate theory, and thus lead to hypothesis-driven, experimental research designs that may only rely on quantitative data.
Unfortunately, many outside academia (and a few inside those walls) think only the latter is "proper research".
via Harold Jarche, who is editing OLDaily while Stephen Downes takes a long break.
Chris' claim is that with petabyte storage (yes, that's lots), "we can stop looking for models. We can analyze the data without hypotheses about what it might show. We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science cannot." The patterns found are the emergent model, so this is just rubbish.
Chris clearly has not actually done academic research. If he had, then he would know that it is the research question that defines the methodology, not the subject or the capacity of your computer. There are already many approaches, especially in social studies where meaning is sought, which lead to qualitative methods of data collection (ie. they need human recorders) from which patterns are discerned. Grounded theory is a well-established methodology that generates explanations through a structured inductive process from focussed qualitative or quantitative data. Other research questions are precisely designed to validate theory, and thus lead to hypothesis-driven, experimental research designs that may only rely on quantitative data.
Unfortunately, many outside academia (and a few inside those walls) think only the latter is "proper research".
via Harold Jarche, who is editing OLDaily while Stephen Downes takes a long break.
Post a Comment