AUG.25 RESPONSE TO JOSHUA BLUMENSTOCK’S ARTICLE
Ziyue Feng
I think Professor Joshua Blumenstock’s warning is practically meaningful, instead of being catastrophizing. There is overstating about the power uses of data science as many people believe big data is so reliable and representing that everything can be achieved by only a stream of algorithms. Joshua points out a series of limitations and problems of up-to-date big data analysis, most of which sounds right to me. And that’s why I agree on his idea: data science should be humbler than the deified image presented to the public.
In my opinion, what he means by being ‘humble’ includes two aspects. First, as I mentioned above, people are supposed to realize the limitation of data science and the truth that it is not a mature technology to tackle development problems. For instance, Joshua believes that data collection may be biased since its approaches like through mobile phone or social media platforms have already excluded segments of population. This makes data become no more representative for a census. It turns out that solutions for development problems will ignore even sacrifice the benefits and willing of people who do not have a mobile phone or a Facebook account.
Second, a humbler data science also means it is necessary to make this technology down-to-earth. In other words, it is necessary to consider the context and local situations before using hi-tech big data. This helps these advanced approaches to avoid their defects. When you know there are millions of people having no mobile phones or social media accounts in a country or region, you probably understand it is not an appropriate way to collect data by these two ways. Traditional means like interview or survey seems to be better in this case, and these progressive methods of data science can be as assistances.
But we should not underestimate the power and potential of data science and these advanced approaches. It is an evolution from conventional data. It generally provides the world with more accurate, more efficient and more accessible approaches to analyze International development nowadays. All we have to pay attention to is that there are flaws and side effects like lack of regulation to privacy offense. There are rooms to improve for advanced approaches and old, conventional data is still irreplaceable.