In one of my favorite articles we’ve seen so far this semester, “Trending: The Promises and the Challenges of Big Social Data,” Lev Manovich proposes a new type of humanities student and scholar: the kind that can both think and analyze like an English major, but also research and construct digital environments in which to host and process their work like a computer scientist. During this whole class, I have wondered about digital media as a study of English and literature, especially when considering what kind of (albeit “stupid, little”) digital object I, and the rest of the class, would create. I’ll assume we all have the capability to dream up digital objects that crunch numbers, move wildly about the screen, or aggregate all instances of certain themes on the world wide web, but…are we capable of actually creating those objects? Manovich says that, “if each data-intensive project done in humanities would have to be supported by a research grant which would allow such collaboration, our progress will be very slow,” indicating that we (as humanities students) may not currently possess the ability to program or write code and algorithms necessary to do the type of “big data” research we would like to, and we’d better start enrolling in IT and computer literacy classes in additional to contemporary lit and cultural studies classes.
I definitely think Manovich is right, that the humanities (and particularly the college major course requirements for humanities) could use an infusion of computer science. That said, I think most courses of study could benefit from this infusion. Not only can computers help us to parse big data useful for humanities research, they and (knowledge of/about them) can help tackle all sorts of hurdles more easily accomplished by an algorithm than “by hand.” I work as an online sales manager for a small business, and I totally understand what Manovich means when saying that you sometimes need to have specific computer knowledge in order to collect the types of data you want. If I want to organize inventory in a specific way or track trends in sales that are not “pre-supported” in the algorithms that the program automatically offers, I have to create myself a new Data Import file or a new Data Export file, that tells the program how I want it to read the information that I will upload into it as en excel or text file. This is not something I was trained to do or previously had knowledge of, and as a result has caused me to seek out a lot of computer skills knowledge that I didn’t already have. Gaining this knowledge and ability to manipulate inventory and sales data through the computer has not just benefit my understanding of the company’s fiscal position, but has allowed me to more thoroughly analyze trends and make adjustments to the way we do business as a result.
Maybe this is because I don’t know too much about how programming works, but the one question I did keeping asking myself throughout reading the article (especially when Manovich is talking about reducing the “data landscape” to a useable size) was: What are the computer algorthims for videos, photos, and non-text datas based on? How would you ask the computer to put constraints on the data set? Are these constraints based mainly on the “formal” aspects of the data, i.e. time, date, length, size, color, original tags or descriptions associated? How would you organize the data by themes, if all you had was length of video and file size? For that matter, how would one organize the data based on any content with physically watching all 1 million videos and tagging them all with relevant terms? For that matter, wouldn’t doing something like that result in a fairly subjective idea of what the themes or content of each video is?
So my original plan was to post about Manovich’s article with a very similar type of conclusion, so upon reading through your post, I decided to take my response and post a comment instead. Thus while I most certainly do agree with your assessment of the article, I hope to help provide a new perspective.
In the article, he alludes to some of the already known and established fields of quantitative social sciences that take large pieces of information about lots of people and come to conclusions and predictions based on what is called “surface data.” One of these fields happens to be my degree of choices: Economics (with an English minor). Thus as an econ major I can confidently say that one of the main struggles and divisions between many economists today is between the field of macro and micro, for the exact reasons that Manovich is presenting in his article. “Is it really true that ‘We no longer have to choose between data size and data depth’ as I stated? Yes and no.” In that section he makes the argument that there are different types of data being analyzed and it is very difficult, at least at the current time, to be able to combine them into one all-seeing analysis. This is exactly what economists have problem understanding at times. While macroeconomics takes data sets across multiple markets and fields of behavior, microeconomics reminds that at its core, each individual human interaction is guided by a completely subjective motive, behavior, and value system.
Thus what I see as the potential in big data sets in analyzing economics is taking already multiple social interactions to a further level (namely through personal public data such as Facebook, Twitter etc.) and making better predictions about markets based on that so called “size” and “depth” he is alluding to. Where your conclusion comes into play brings us back to a deep level: how is it possible to convert these truly subjective ideas and create objective algorithms to analyze them? As We Feel Fine has showed us, attempting to take a massive amount of subjective surface data often results in more subjectivity. While it is interesting to look at, like Manovich, application and analysis to make predictions still feels less optimistic to me.