Tutorial: Reconstructing the Genealogy of a TV-Clip

Jasmijn van Gorp, Utrecht University

Tutorial description, case and objectives

‘A television program’ is often reduced to one particular broadcast in time. Usually the date is formulated in terms of ‘year’ and set on the year the program was initially broadcast. However, television can be characterized by re-runs and repetition, or as Kompare (2005: xi) puts it, by “repetition geared towards the constant recirculation of recorded, already-seen events”. The VCR and the DVD, as Kompare also argues, historically played a pivotal role in the culture of repetition. More recently, the advent of user-generated content platforms such as YouTube, as well as the proliferation of VoD systems such as Netflix and AmazonPrime, have resulted in a widespread ‘re-run’ and ‘re-use’ of television.

In this tutorial I provide some steps for critically questioning and interrogating the tendency to use initial broadcasting dates of television programs as the one and only identifying date. I will focus on the different transformations of a particular program -in this case a television clip- to illustrate its different timely appearances in digital archives. I will not only focus on rebroadcasts and reuse, but also on archival transformations of the record. The latter is important because, as archival scholar Eric Ketelaar (2001) reminds us, each record is transformed throughout the archival process, which leaves traces behind. By looking at the different metadata transformations of the same item, we aim to -again following Ketelaar (2001)- reconstruct ‘the semantic genealogy of the record’.

By following the steps in this tutorial, the reader will gain a better understanding of the archival processes affecting the record and, therefore, obtain some basic ‘video forensic’ skills that can help in assessing a video’s origins. Within media and culture studies, such skills can be situated in the method of ‘source criticism’ (media history) or ‘data criticism’ (critical data studies). This tutorial specifically focuses on criticism of the time/date allocation of a video.

In this tutorial you will learn how to use the CLARIAH Media Suite and other digital archives such as Delpher and the WayBack Machine to reconstruct the genealogy of a TV-clip. Upon completing the tutorial you will:

• develop a basic skill of digital video forensics

• understand the concept of ‘provenance’ of sources in relation to television materials

• know how the Collection Inspector Tool of the Media Suite works

• know how the Date Fields of the Sound and Vision Archive metadata can be interpreted

• know how these date fields can be deconstructed by opposing them to date fields of other archival search systems

• know the basics of working with Delpher and the Wayback Machine.

Types and levels of teaching and research

This tutorial is aimed at students at an advanced level in Media, Data and Journalism Studies. The tutorial aligns with a critical data studies perspective and is useful in developing critical thinking and reflection. It is highly recommended to use this tutorial in a supervised setting. To run through the full tutorial takes about 90 minutes.

Recommended reading:

• Aasman, S. (2019). Finding Traces in YouTube’s Living Archive: Exploring Informal Archival Practices. TMG Journal for Media History, 22(1), 35–55. https://doi.org/10.18146/tmg.435

• Ketelaar, E. (2001). Tacit narratives: The meanings of archives. Archival Science, 1(2), 131–141. https://doi.org/10.1007/BF02435644 .

• Kompare, D. (2005). Rerun Television: How Repeats Invented American Television. London & New York: Routledge.

• Noordegraaf, J. (2010). Who Knows Television? Online Access and the Gatekeepers of Knowledge. Critical Studies in Television, 5(2), 1–19. https://doi.org/10.7227/CST.5.2.3 .

For lecturers teaching this tutorial

• I recommend consulting my article on Interstitial Data, from which this tutorial is derived: Van Gorp, J. (2022) Interstitial Data: Tracing Metadata in Search Systems. In Verhoeff, N. & van Es, K (eds.) Situating Data: Inquiries in Algorithmic Culture . Amsterdam: Amsterdam University Press. Forthcoming.

• I also created a more basic and shorter version of this tutorial. In this set-up, I have divided the classroom in two in which one group worked with the 1991 broadcast and one group with the 2006 broadcast. In the discussion we compared the results of both groups. Handouts for this set-up can be downloaded here (1991 clip) and here (2006 clip). The Lecturer’s manual for both handouts can be found here . The duration of this short version is about 30 minutes for the practical part and 15 minutes for the discussion (45 minutes in total).

Prerequisites

Use of this tutorial requires that students are already trained in using the Media Suite for basic search functionalities. A basic knowledge of archives is required.

This tutorial may be used for a workshop series with or as part of a tutorial package in combination with the following tutorials:

• Logging in, Workspace and Creating a User Project

• Searching and Bookmarking for Television History

• Basic tutorial Viewing Rates Basic tutorial Viewing Rates (forthcoming)

Acknowledgements

This tutorial was made in the context of the courses Television History Online and Data as Media at Utrecht University and for workshops within the CLICK-NL project RE-FRAME .

Steps

Step 1. Preparation

• Login with your Dutch university credentials

• Make sure you have set up a project within your Media Suite workspace cf. Screenshot above). Consult the tutorial on setting up a User Project if you are in need of a step-by-step guide.

• Secondly, take a few minutes to prepare your own research journal. As we will make use of different search systems in this tutorial, it is necessary to keep track of other systems than those of the Media Suite in the process. Moreover, by keeping a research journal you will also be able to consult your own thoughts when continuing your search in subsequent days, and when formulating your research findings. I prefer to use a GoogleDocs or Microsoft SharePoint Excel sheet as it is shareable, but a regular Word-document will work equally fine. I wrote down many different aspects in my own research journal, but I recommend making notes on at least the following five topics

  • date of your search

  • your search terms

  • search engine used

  • URL of the record

  • personal comments (remarks, analysis, theoretical concepts etc)

• You are now all set to start your research.

Step 2. Select a Clip

• Select a clip that you want to trace. I recommend taking a clip of which you know it is widely re-used or even ‘iconic’. In addition, the clip should be available for view in the CLARIAH Media Suite.

  

• If you do not have a clip you can use in mind, you may use my example clip. In my search for Interstitials (see my dedicated curated playlist here) I found a fascinating clip: a fictional trailer for a film. In this TV-clip of the comedy show Jiskefet , a male voice-over announces the non-existing film La Pluie Sans Droit of the non-existing French filmmaker Jean-Philippe Le Grain, which will be (not) broadcast the upcoming Saturday on VPRO Cinema . I select this clip for my endeavor to reconstruct its genealogy. To find this clip, go to Search Tool, select the Sound and Vision collection of The Netherlands Institute for Sound and Vision, search for ‘Jiskefet film noir’, and bookmark the second result: JISKEFET. Jiskefet 1991 (see Screenshot above).

• Click on the Jiskefet item and look at the mentioned date. It reads: 1991-09-22.

Step 3. Newspaper’s Broadcast Schedules

• As mentioned in the introduction, our prime objective is to understand the program’s date. In the previous step, we learnt that the original broadcast date is 22 September 1991. To know the day in the week, a simple Google search tells us this is a Sunday. So we learnt that Jiskefet was programmed on a popular time slot of a Sunday at first. But at what time? Was it broadcast during prime time?

• To get the answer, we have to consult a broadcast schedule. In the Netherlands broadcast schedules are published in broadcast magazines. However, these are not broadly available. The KB holds broadcast magazines until 1961 . The Netherlands Institute for Sound and Vision has the more recent digitized broadcast magazines (from 1962 onwards), but these are not available outside of the building due to copyright reasons. If you want to search recent broadcast magazines, you can make an appointment with the Customer Service of Sound and Vision to use their in-house search system DAAN.

• A good alternative to arrive at broadcast schedules are newspapers. Newspapers until the 1990s can be found in Delpher , the newspaper database of the Royal Library (KB) with for example the archives of De Volkskrant (until 1995) or NRC (until 1994). After this date, it is advisable to search in the commercial database NexisUni at your university library (e.g. here the one of Utrecht University). However, the challenge of using NexisUni for historical research is that it removes the lay-out and photos, and only shows the plain text.

  

• For the 1991 broadcast schedule, I used Delpher, search for Jiskefet, and select 20.9.1991 to 21.9.1991 (as then you are sure that you have the broadcast schedule of the Sunday, as most newspapers publish their schedules on Saturday for the weekend). I selected the sixth result, the one of the NRC Handelsblad . I found the schedule in the NRC of 21.9.1991 on page 29 (cf. Screenshot above).

• Look into the broadcast schedule and note down: broadcast time, title, description. Also write down what programmes were broadcast before and after.

• Did you get to this page? The quality of the scan is not very good, but I could see that the broadcast time of the original broadcast is 20:10, the channel is Nederland 2, and the broadcaster is VPRO. The description reads: “Series played and written by Herman Koch, Kees Prins and Michiel Romeyn”. It was scheduled after the news and before a Swedish drama series. It was broadcast in prime time.

• In this step, we learnt about the broadcast times as they were scheduled in advance. Let’s now turn to a method to know if the exact broadcast aired at the same time as it was originally scheduled: the digitized viewing rate reports. Step 4. Broadcast schedules in digitized Viewing Rate Reports

Step 4. Broadcast schedules in digitized Viewing Rate Reports

• As explained in the Viewing Rates tutorial , the digitized Viewing Rate Reports as part of the Media Suite contain the exact broadcast time, the viewing density and the viewers’ appreciation index.

• It is only possible to find viewing rates until 1994, so it should be possible to retrieve the viewing rates for the Jiskefet broadcast of 1991. The reports are compiled in weekly reports so I search broadly for “Jiskefet” in a 10-day timeframe (10/09/1991 to 30/09/1991) in order to be sure I find the correct week in which the program is scheduled. Open the first result in your result list, which is indicated with the date ‘16-09-1991’.

• If everything went well, you automatically land at the right page: you have arrived at page 175 (p. 175 in the page browser and page 7.6 on the digitized page) to consult the rates for Jiskefet of 22 September 1991 (cf. Screenshot above).

• The viewing rates are difficult to read. I decided to only use the totals of 13+, Men 13+ and Women 13+ (feel free to also use 6+) and copy-paste them in a separate table (cf. Table above). KdH refers to Viewing Density and Wrd to viewers’ appreciation. What do we learn from these ratings?

• We notice that the news was broadcast at the exact time: the news of 8 o’ clock starts at 8 sharp, while Jiskefet 3 minutes later and Age Unknown just one minute later compared to the schedule we found in the newspaper. Can you think of any reasons to explain the differences?

Step 5. Analyze date fields with the Collection Inspector Tool

• As the title of my clip already states (‘Jiskefet. Jiskefet 1991’) it looks as if the clip was broadcast for the first time in 1991. To unpack the date field, we now need to take a detour via another tool: the Collection Inspector Tool . This tool is designed to inspect the completeness of metadata fields and indicates the percentage of missing information per metadata field per year. We are going to use this tool to explore what kinds of date fields are available in the Sound and Vision Collection.

• Go to Tools, and open Inspect Tool. Add collection ‘Sound and Vision Archive’ (see Screenshot above). Now click the collection title to open the collection analysis window. Click on the ‘Select Fields to Analyse’ button and type in ‘date’ without pushing the enter button. You will now get a list of 120 different metadata fields for date.

• Browse through the list of 120 metadata fields for date to explore the different date fields. Is it easy or rather difficult to understand the meaning of the different date fields? Click on some date fields in order to inspect the completeness. Go for instance to “Date of Recording“ and “Date of Editing”.

• Then, select the default metadata field, which is ‘date, sorting PREFERRED’. Read the explanation. It says that sort date is an automated field that calculates the most relevant date that could be the broadcast date. What does this mean? In most cases, the first broadcast date is known. If the first broadcast date is unknown, the ‘date, sorting PREFERRED’ metadatafield is based on other date fields that are considered to be relevant to estimate the date, such as Date of Recording or Date of Production. Discuss with your peers what we know and especially what we don’t know when using the metadatafield ‘date, sorting PREFERRED’.

Step 6. Analyze date fields with Unix Time Converter

• Go back to your bookmarked item from Step 2 and open it. The Resource Viewer opens and shows the clip with two panels on the left: one panel with metadata and content annotations, and one panel with My Annotations. Close the My Annotations panel.

• Scroll down in the Metadata and Content Annotations panel and open Full Metadata (cf. Screenshot above). The Full Metadata opens in a new window.

• The Full Metadata document is a large file. Search for “date” (ctrl F). As you notice, there are multiple date fields, which we already explored when working with the Collection Inspector Tool in Step 5.

• Search for ‘date’, ‘date, sorting PREFERRED’ or ‘curatedDate’ - these are referring to the same and are all equally fine to use (cf. Screenshot above). In my case date, sort date and curated date are “1991-09-22”. Interestingly, the metadata naming already reveals that the plain metadata field ‘date’ is the broadcast date, confirming our point of departure that dates are generally set on the first broadcast date.

• Now we are going to dig into other dates in the full metadata. This is the most difficult step in this tutorial as metadata is shown at different levels in a ‘tree format’, so it is easy to make mistakes (if you would like to know more about the ‘tree format’, look at page 26 here). Therefore, you really have to make sure that you select the right fields within the upper branch (marked with an indentation or in Dutch ‘inspringing’). How do you do that? Within the Full Metadata search first for the so-called upper branch “AssetItems”. You will notice two dates within this branch: “date_created” and “date_last_updated”. In my case these are the data: “date_created”: 1207949409000 & “date_last_updated”: 1564458885345.

• Take the date_created, which is the time that the digital file is entered in the system and, therefore, can be (admittedly roughly) interpreted as the digitization date. What does this date look like?

• As you notice, the date_created is not readable as it is a Unix Time Stamp date. Therefore, we have to go to the unix time converter here . Enter the numbers, and the exact dates and times are revealed. To check whether it is correct, look at the nearest ‘date’ field 9 lines above the “date_created”. Now try to read this date, and unravel what the T and the Z stand for when comparing it with the date_created you unlocked via the unix time converter.

• Repeat the same steps for the last_updated date, which is the date on which final changes to the metadata were made. In our case, it probably refers to the moment in which the metadata was converted from the old database system iMMix to the new database system DAAN. As I explain in my article (Van Gorp, 2022), it is difficult to pinpoint exactly for each date field what it stands for as the metadata are made in early years and have been converted multiple times.

• At this point, we learnt that the program with the clip was broadcast on 22 September 1991 (date/sort date/curated date), digitized or streamed into the system on Friday 11 april 2008 23:30:09 Amsterdam time (asset date / asset item date created), and the file updated for the last time on Tuesday 30 July 2019 05:54:45 Amsterdam time (asset item date last updated).

• We now reconstructed a small part of the broadcast’s physical and digital trajectory within the archive. Take a moment to reflect. What does this entail in terms of provenance? Can you think of any other parts of the trajectory that are still unknown?

• If you are intrigued by this question, I invite you to search for the metadata field ‘Annotated’, and find out more!

• Let’s now look for other versions of the broadcast.

Step 7. Finding Re-Broadcasts in the Media Suite

• Clips are often re-used in different contexts. In this step we will first look at re-broadcasts. This is a bit more cumbersome step as there are multiple options to find re-broadcasts within the Media Suite depending on your selected clip. In my case I have a specific clip, so I search again for “Jiskefet Film Noir” and find that it is re-broadcast in 2006 in a compilation of the ‘best ofs’ of Jiskefet: Jiskefet De Afgelopen Jaren (cf. Screenshot above).

• For your own clip, you can use the Search Tool and query for the title of the clip in the metadata field description by a search in ALL Fields. You can also conduct a web search to try to figure out re-broadcasts and then search for the exact broadcast date in the Media Suite. Also think of talk shows in which your program is discussed. Especially when you have a heavily canonized clip, you can expect to find up to 10 re-broadcasts of your clip. Choose the ones that spark your interest!

• Repeat steps 3, 4 and 6 for your re-broadcasts

  • Go to Step 3 and search for the broadcast schedule in the newspaper. In my case, the re-broadcast is of 2006, which means it will quite likely not be in Delpher. I used NexisUni via my university library of Utrecht and found the broadcast schedule in De Volkskrant of 4 October 2006 on page 16 (see screenshot above).

  • To arrive there: Go to Publication > type in search term Jiskefet > Type De Volkskrant in [Title Publication] and select De Volkskrant from the drop-down menu. Select in [Tijdlijn] the date 4 October 2006

  • Note down: broadcast time, title, description. Also write down what programmes were broadcast before and after.

• Go to Step 4 which is in my case not possible as the Viewing Rates are only available until 1994 while my rebroadcast is from 2006.

• Go to Step 6. Let’s again dive into the date fields. Open the Full Metadata, and collect the same dates as for the original broadcast, meaning those with “AssetItems”. Use again the Unix converter to convert the dates. For my AssetItem it is: “date_created”: 1207949409000 and “date_last_updated”: 1569773762194. So, we now know that it was re-broadcast on 4 October 2006 (date/sort date/curated date), digitized on Friday 11 april 2008 23:30:09 Amsterdam time (asset item date created), and the file updated on Sunday 29 September 2019 18:16:02 Amsterdam time (asset item date last updated).

• Compare your findings with the ones of the original broadcast. Based on the newspaper schedule analysis of Step 3, you notice that De Afgelopen Jaren has moved to the channel Nederland 3 and to a Tuesday at 22:45, which is very late. It is squeezed in between a film and a talk show. The description is brief: “Selection” and it has teletext subtitles for the hearing impaired (TT). If we compare the results of Step 6 of the original broadcast with the results of the rebroadcast, we can conclude that both broadcasts have been entered in the system at the same day and time, while the files are updated on a different moment.

Step 8. Finding Re-broadcasts on YouTube and the WayBack Machine

• To find re-broadcasts, we also have to take into account videos on the web.

• Let’s first search “Jiskefet Film Noir” on YouTube . Let’s now investigate whether the video has been on YouTube all the time since 2008. When videos are removed from YouTube and thus no longer accessible via that route, you can try to find screenshots of previous uploads on the WayBack Machine : a service of the Internet Archive that enables users to retrieve archived web pages. Importantly, these are based on screenshots of the webpages, but do not enable playing out the videos. Aasman (2019: 14) discusses the potential of the Wayback Machine for historical research on YouTube: “The history of this single snapshot makes us aware of the fact that, even if the video is still there, the context changes continuously, making the post a living document. It reflects the dynamics of YouTube as a platform, as the sequence of the snapshots of the same video document the many changes.”

  

• To do this, copy-paste the exact URL of the Jiskefet YouTube channel. That is: https://www.youtube.com/channel/UCi1LpRIlG1tDY5Z54VTel2w . Use this URL for the WaybackMachine.

• We notice 32 captures of the page in 2022. Go to the archived web page of 22 February 2016 (cf. Screenshot above) and analyze the Jiskefet Film Noir upload. What is the title? When is it uploaded? How many views does it have compared to others? Investigate the context of the page: Where is it situated on the webpage? What does the banner look like? Consult the ‘about this caption’ page and write down: who has made this capture? where? when? why?

• Now skip through all captions and stop at the archived web page of 28 August 2019 . What happened with the Jiskefet Film Noir clip? Do you know why? Then try to skip further. What happens? What does this mean?

• Did you find the answer? Conduct a web search to learn what happened in 2019. If you would like to know what happened with the video content, click here and here . It seems that, while the video was initially deleted, it was made visible again on YouTube in 2023. In addition, the Jiskefet platform seems to have changed its policy and installed a paid membership-system in which members (1) could watch videos without commercial breaks and (2) receive special content. On their YouTube channel they encourage viewers to watch the full episodes on their own platform.

• You can repeat this Step 8 also for other YouTube URLs of your own clip. In my case, there are no other YouTube URLs available for the clip as I couldn’t find any fan uploads. I assume the reason for this lack of fan uploads is that (1) Jiskefet has its own platform and youtube channel and (2) my clip is not the most ‘classic’ or ‘iconic’ Jiskefet clip (‘iconic’ Jiskefet is rather Debiteuren/Crediteuren and The Lullo’s for instance).

• Now you have reached the end of the tutorial. You learnt how to conduct data criticism on the metadata field ‘date’ and gained insight in the various forms of repetition and re-use of television programs.