Tuesday, June 16, 2020

The Ragascape of Hindustani Music: VI

Methodology and the data source

Research design for this study had first to determine the set of Raga-s to be considered.  The list of “Candidate” Raga-s was compiled by merging the under-graduate and post-graduate syllabi of the major teaching universities and examining bodies granting certifications/ degrees/ diplomas in Hindustani music. This list was submitted for suggestions/ approval to a panel of musicians known for their vast Raga repertoire and academic orientation. Through this two-tier process, a list of 97 Raga-s was finalized, for which audience engagement measures were to be computed.

For compiling the data, the YouTube platform was searched for each Raga. The coverage of “significant” musicians performing each raga was attempted as a census (100% coverage), and not as a sample.  The assessment of “significance” utilized my own knowledge of musicianship over the last century on the national as well as regional levels. Considering the global scope of the enquiry and the diversity of material, my judgement of “significance” could have been less than 100% defensible. Care was taken to include significant musicians representing (i) vocal music and all significant instruments (ii) Dhrupad, Khayal, and Thumree genres (iii) contemporary/ fusion genres where recordings explicitly claimed and performed Raga-based music (iv) musicianship originating in Pakistan, Bangladesh, and Afghanistan performing Raga-based music (v) musicianship originating in the US, Europe with credible performing attainments (vi) musicianship originating from the Carnatic tradition but performing Hindustani/ shared Raga-s in Hindustani style.

Media researchers have found that YouTube on-screen data is singularly unhelpful for systematic research. This would seem intentional because unambiguous data can be used for competitive purposes (can help some musicians at the cost of others). This can jeopardize YouTube’s impartial status as a media owner/ platform provider, and damage its commercial interests. Having conducted a major study in 2015 based on YouTube data, (Refer Chapters 15 through 20, in The Musician and His Art – Essays on Hindustani music: DK Printworld, New Delhi 2019) I had developed a method of usable data acquisition from YouTube. This method obliged me to personally create every one of the 8100 records for analysis. Despite utmost diligence in database creation, I am unable to vouch 100% for the soundness of my results. YouTube’s evasive purpose is fulfilled.  

A notional limit of 100 data-points was envisaged for each Raga. The data-base compilation was completed in the last week of March 2020 and the first week of April 2020.

The Audience Engagement Indicator has been computed as follows:

 Limitations of YouTube data

YouTube as data source is far from ideal for systematic research, and that I am using it in full awareness of its limitations. Despite this, it is now being used by researchers even in the more advanced research environments. Presumably because it has usable value, its use will not stop; and because of its known limitations, the controversy over its value will not end.

The perception of limitations will obviously be guided by the researcher’s objectives – what he is looking for, and what he may instead be measuring. My concerns are focused on the following infirmities.

YouTube reports “views” for each upload. The basic question is: what is a view? What minimum duration of exposure qualifies a “visit” as a “view”? The information displayed on the screen does not distinguish between a 10% viewing of a video and a 40% viewing of it. The difference matters to me.

The duration aspect of viewership is also connected to a change in YouTube policy some years ago. There was once a duration limit on uploads. This affected Hindustani music very significantly, as most performances exceeded the limit, and had to be split into 2, 3, and sometimes 4 parts for upload. After the duration limit was lifted, mostly complete performances were uploaded. Looking at that data today, we implicitly equate the engagement measure of a concert split into 3 parts with that of a complete performance. This seems unreasonable.  

YouTube viewership data is cumulative from the date of uploading. By using that data, I am implicitly accepting one view of 2010 on par with one view of 2018.  Intuitively, we know the assumption is flawed. During this period, YouTube content has changed substantially, viewership has grown exponentially, and audience profile for every kind of content has almost certainly changed radically. No correction factor can help.

Another problem with cumulative data is that it effectively equates 20,000 views accumulated over 20 months with 40,000 views accumulated over 40 months. Intuitively, this equation does not look reasonable. If the propensity of a recording to accumulate viewers is important, 40,000 over 4 years is more valuable. And, if the speed of audience accumulation is considered important, 20,000 over 20 months is more valuable. YouTube data is unhelpful in this respect.

The YouTube audience is global, and so is the audience for Hindustani music. But, we have no data on the geographical spread. By implication, we are accepting that foreign audiences of Hindustani music and foreign musicians —across all nationalities and cultures -- have the same relationship with the music, as Indians have. This is unrealistic. The opacity of the geographical spread can easily mislead us – as it has the possibility of doing in this study.

YouTube viewership – by whatever criterion registered – is a partially manipulated expression of audience engagement. This is because a sophisticated program guides the viewer into “viewing” content beyond his purpose, and spending much more time on YouTube than was necessary or planned. As a result, the numbers we see include “incidental” viewership. A separate reporting of the primary (search word) and incidental viewership would be very helpful.   

Although YouTube is a video medium, the nature of the content is itself not uniform. In fact, even the notion of “viewership” may be irrelevant to a lot of the content. Do 100 people listening to an audio recording with just a photograph of the musician on the screen represent the same level of audience engagement as 100 people watching him or another musician in action on film? If not, how much can we depend on a standard measure of audience engagement across different content formats?

With specific reference to Hindustani music, YouTube neither offers a uniform media experience to its audience, nor publicly provides a rigorous measurement of audience engagement. What, then, does YouTube data provide? It cannot be said to provided “statistics”, but can suggest “orders of magnitude”. Inferences can be drawn judiciously from its analysis, with every inference reflecting the analyst’s awareness of data limitations.

YouTube has invested heavily in generating analytics for user management and advertising value assessment. At some stage, it will have to start understanding itself as a cultural force. It will then benefit by cultivating communities of media researchers through a more transparent stance with respect to its data assets.

Until this happens, the Indian musicologist should be content with indicative inferences. Is this better than the “Delphi method” of polling 10 veteran connoisseurs and observers of the music scene? I believe so because, firstly, surveying this population globally is an almost impossible task, and secondly, present-day oracles – whether of Delphi or anywhere else -- are susceptible to "personal" preferences, biases and prejudices; impersonally generated numbers are not.