It is with pleasure that I received the news that my long paper proposal for the Digital Humanities conference in Sydney during summer 2015 has been accepted. Fierce competiton apparently – five reviewers for each long paper proposal; 75 such papers accepted with an overall acceptance rate of 52 percent of submissions. Anyway, nice to be on board – my proposal entitled “Record Label Set Up” runs as follows:
Within a short space of time, ways of listening to new (and older) music have rapidly changed. If visions of a celestial jukebox two decades ago promised music in the cloud for everyone—this is now a reality, with Sweden’s Spotify as one of the foremost examples of the music industry’s technological transformation. The digital download model, however, still remains a key revenue stream. Yet subscription and ad-supported streaming services have grown from ten to almost 30 percent of digital revenues in the last five years. As stated in the latest IFPI Digital Music Report 2014, ”it is now clear that music streaming and subscription is a mainstream model for our business. In 2011, there were eight million paying subscribers to subscription services—today there are 28 million.” In the report, especially the Scandinavian countries are said to be a showcase of music industry revival—where Spotify has a total dominance—demonstrating the ”regenerating potential of the streaming model.” Rediscovering older music is a key concept for Spotify, and the company has worked fiercly with constantly expanding and upgrading its catalogue. Spotify thus poses a number of challenges to traditional ways of handling music heritage—especially with regards to the ALM-sector (archives, libraries and museums). Seamless access to music has simply led to alternative usage patterns, new forms of engagement and the popularisation of older musical forms in ways that traditional cultural heritage institutions can never replicate. In short—when streaming heritage becomes a fact, what is the (new) cultural role of heritage institutions?
My presentation takes the form of a first report of a new research project, ”Streaming Heritage. Following Files In Digital Music Distribution”. The project is financed by the Swedish Research Council—with approximately one million euro—and involves five researchers working in an interdisciplinary research team. The project is situated at HUMlab (Umeå university) and will run between 2014 and 2018. On a general level, the purpose of the project is to study emerging streaming media cultures, and the music service Spotify in particular, with a bearing on the digital challenges posed by direct access to musical heritage for memory institutions. However, the purpose is also both web historical and contemporary. On the one hand, this so called”Spotify-project” will track the development of online music cultures, from file sharing at Napster and The Pirate Bay to legal streaming services as Spotify. On the other hand, it will follow files in digital music distribution by way of digital ethnographic methods. Research is, in short, conducted and based on the creation of a non-profit record label in order to study unexpected file ’behavior’. Programmers from HUMlab are involved in the research process. The record company acts as an innovative research tool, with the aim of following—or simply pursuing—digital music files throughout the intra-digital distribution process: from creation, via aggregation, and through to playback. Using various digital methods the ambition is to observe the files’ journey through the digital ecosystem—streaming media culture’s black box—normally not accessible to traditional media researchers. The basic idea is that digital (or digitized) media objects has changed the way in which they should be conceptualised, analysed and understood—not the least from a heritage perspective. That is, a movement from studying static music artifacts to an increased scientific focus on dynamically active files with a kind of inherent information about factors such as broadband infrastructure, file distribution and aggregation, user practices, click frequency, social playlists, sharing and repetition. During autumn 2014 we have initiated and started our record label, and via the aggregator RouteNote—”a free way to get your music onto the world’s leading stores and distributors”—begun to track ”our music files”. Research results will follow.
Den doktorandkurs på Umeå och Lunds universitet, som jag ger i samarbete med Patrik Lundell och Johan Jarlbrink, har satt igång. De inledande passen har handlat om kulturhistorisk medieforskning, om mediearkeologi och om vetenskapens mediehistoria. De tre första föreläsningarna (i lite olika form) finns att ladda ned som PDF här: 1. kulturhistorisk medieforskning 2. mediearkeologi 3. vetenskapens mediehistoria
Joachim Sundell på SVT Kulturnyheter har skrivit ett reportage om The Pirate Bay, apropå att sajten blir allt egendomligare: “Piratskeppet går mot mörkare vatten“. Jag och flera kollegor uttalar sig. “Pirate Bay återuppstår – men idealismen är död. När fildelningssajten återvänder är det med större vinstintresse och mindre ideologi, menar flera experter.”
Tillsammans med min kollega Johan Jarlbrink arbetar jag för närvarande på en artikel inom ramen för forskningsprojektet Digitala lägg. Om pressens gränssnitt 1800. Arbetstiteln är “Maskinläsning” och vi tänker oss en möjlig publikation framöver i Nordicoms medietidskrift. Artikeln tar i nuvarande skick sin början så här:
Hösten 2014 samlades ett sjuttiotal bibliotekarier, forskare, byråkrater och representanter för tidningsbranschen på British Library i London för en konferens om digitalisering av äldre dagspress inom ramen för kulturarvsportalen Europeana. Projektet Europeana Newspapers har haft ambitionen att digitalisera 18 miljoner tidningssidor, varav drygt hälften kommer att vara sökbara i fulltext genom så kallad ”optisk teckenigenkänning” när projektet är avslutat. Just OCR, ”Optical Character Recognition”, var ett av konferensen mest omdiskuterade teman. Det är inte förvånande eftersom OCR på flera sätt är den teknik som gör digitaliserad dagspress till en helt ny typ av forskningsmaterial. OCR konverterar dagspressen som medium – från tidning till maskintext – genom en närmast cirkulär textuell migreringsprocess där tidningssida blir inskannad bild, vilken därefter omkodas till maskinläsbar text. Ur den senare kan dels genereras inherent data, extrapolerad ur det inskannade materialet, dels möjliggör den maskinkodade texten storskaliga kvantitativa analyser av exempelvis ords samförekomster.
Samtidigt är OCR-tekniken fortsatt behäftad med problem, vilket London-konferensen gav besked om. Har vi nått ”peak OCR” var det flera deltagare som undrade. Särskilt tröstlöst tycktes läget vara i länder där frakturstilen dominerat. 1930 publicerades exempelvis fortfarande omkring 60 procent av alla trycksaker i Tyskland i fraktur, och frakturstilen var även vanlig i Sverige fram till 1800-talets slut. Men även den latinska tryckstilen, antikva, orsakar fortsatt igenkänningsproblem menade andra. Det marknadsledande företaget ProQuest, och deras inskanning av 472 brittiska tidskrifter mellan 1681 och sent 1930-tal, ger exempelvis fantastisk access till ett omfattande källmaterial, och samtidigt har forskare påpekat att till och med ProQuest sofistikerade OCR-algoritmer fortsatt inte riktigt går att lita på. I en genomgång av viktorianska influenser på Kuba i det inskannade brittiska tidskriftsmaterialet har Albert D. Pionke bland annat visat på betydande svårigheter för Proquests algoritmer att urskilja ordet”Cuba” från ord som: ”cash”, ”Cheap”, ”Cheapest”, ”Colours”, ”cube”, ”Curacies”, ”curator” eller ”cure” (Pionke, 2014: 391). Det säger sig självt att sådana ”bad matches” kan leda till felaktiga slutsatser och forskningsresultat.
Den här artikeln handlar om digitaliserad (och maskinläsbar) äldre svensk dagspress och de möjligheter som nya digitala metoder ger den mediehistoriska forskningen. Utifrån en mediehistorisk infallsvinkel uppmärksammar artikeln den digitala teknikens möjligheter att på ett generellt plan analysera stora textmängder på jakt efter lingvistiska mönster, bland annat genom så kallad ”distant reading”, en sorts data-metodologisk distansläsning av tiotusentals inskannade böcker som utvecklats av litteraturvetaren Franco Moretti med fokus på 1800-talets skönlitteratur (Moretti, 2013). Samtidigt pläderar artikeln för en mediehistorisk specifik analys, där man som forskare inte bör reducera gamla tidningar till enbart en samling text: bilder, layout, annonsers placering på sidan etcetera är samtliga också delar av det som vi kallar en tidning.
Ingen forskare kan emellertid läsa hela 1800-talets svenska press – enbart datorer. Digitala metoders forskningspotential är därför betydande, om så i form av ”topic modeling”, ”macroanalyses”, ”pattern recognition”, ”cultural analytics” eller andra former av algoritmiska användningsområden för analys av stora kulturella dataset som den digitaliserad dagspressen utgör. Vår artikel väjer dock inte för de problem (och ofta bristfälliga resultat) som den mediehistoriska forskningen än så länge uppvisar på området. Med andra ord handlar det om att uppmärksamma en i allra högsta grad aktuell forskningsdialektik där digitala metoder dels innebär betydande möjligheter för medieforskningen, dels att förhålla sig kritiskt till resultat de genererar.
På rj.se uppmärksammar landets näst största forskningsfinansiär vår infobok: “en både vacker och mycket viktig bok i dessa dagar är Information som problem. Medieanalytiska texter från medeltid till framtid. För alla som är intresserade av information och hur den hanteras, kontrolleras och bevaras är detta ett hett julklappstips.” Det är också roligt att bokens rika illustrationsmaterial omnämns – mer info finns här.
I am currently writing an article – long overdue – with the working title, “More Music is Better Music”. It will hopefully appear in a collection next year edited by Patrik Wikström and Robert DeFillippi. The article departs from the fact that the core motives underlying the disruption process within the music industry during the last decade, has been the establishment of vast online music archives. Streaming services (like Spotify) wouldn’t have been able to gain global and rapid popularity if it hadn’t been for the swelling back catalogues of music providers. The number of tracks available at popular music streaming sites have, as a consequence, during the last years been constantly promoted as a market advantage: Xbox Music claims 30 million songs, Grooveshark 15, Deezer and Rdio 20, Soundcloud 16, Last.fm 12 (at least according to the latest figures). Spotify has some twenty million songs on the service, with twenty thousand new ones added every day. Getting users hooked on the service and continue to listen to more music (than they need) is, hence, perceived as key to (potential) success – even if most services are still far from making a profit.
Anyway, the introduction to the article – which is now at last and finally taking shape – runs as follows:
In March 2014 the funk band Vulfpeck released the conceptual album, ”Sleepify”, containg five minutes and 16 seconds of pure silence. The purpose was to crowdfund an upcoming world tour, and songs were specifically prompted to be available on the Swedish music streaming service Spotify—hence the title of the album. In a video posted at the same time on YouTube, band leader Jack Stratton stated that, when he had sat down with his band to talk about potentially touring during the fall of 2014, ”they said that they would do it under one condition: that all the shows would be free.” Jokingly, he replied: ”That’s not a problem—Yeah!”
In the video Stratton continued by explaining ”how it works”: Vulfpeck releases ”Sleepify” on Spotify, an album that ”is different from our previous albums. This album is much quieter. In fact, we believed it is the most silent album ever recorded.” Essentially, what Stratton was asking fans to do was to play and stream the silent album put on repeat while sleeping—”make your sleep productive”—all in order to exponentially multiply royalties from Spotify. Since the latter are only disbursed once a song is registered as a play, which happens after 30 seconds, all songs on “Sleepify”—ingeniously entitled ”Z” to “Zzzzzzzzzz”—were 31 or 32 seconds long. According to Stratton, 800 streams would roughly generate four dollars in royalties to the band. ”If you stream ’Sleepify’ on repeat while you sleep every night, we will be able to tour without charging admission”, he concluded, all the while vividly exclaiming that if someone was unaware of what Spotify is—it’s a service that’s ”gonna through in the entire history of recorded music” (Sleepify 2014).
Vulfpeck’s recent prank is illustrative of the fundamental changes the recording industry has gone through during the last 15 years. When music is treated as data and handled in binary form, it literally looses its meaning (at least from a computational perspective). Noice reduction and signal processing (or the lack of it)—that is, silence—instead becomes distinctive features. In fact, at the very origin of communication theory (or high modernism for that matter) lies a mute meaninglessness. After all, all semantic aspects of communication were once deemed ”irrelevant to the engineering problem” by Shannon in 1948—only four years prior to the well know ”four minutes thirty-three seconds of silence”. Hence, Spotify’s spokesman Graham James seem spot on when commenting on ”Sleepify”: ”This is a clever stunt, but we prefer Vulpeck’s earlier albums. Sleepify’ seems derivative of John Cage’s work.” (Shannon 1948; Ramirez 2014).
Then again, when music is treated as data, visions of completness also occurs, and it becomes possible to claim (or at least envision) that one is going to ”through in” the whole history of recorded music, as Jack Stratton admiringly put it. The notion of ’the celestial jukebox’ has, no doubt, been the most prevalent one within this archival discourse. It was the first dominant vision of a networked database of consumable on-demand music proposed already two decades ago (Goldstein 1994). Digital purchase of individual songs then became a widely promoted solution, spearheaded by the iPod, iTunes and Apple’s palette of gorgeous mobile devices (Snickars & Vonderau, 2012). Today, ’disruptions’ have occured again, with the game changer towards cloud-based streaming services of which Spotify is perhaps the most salient one—or as the New Yorker recently stated: ”Spotify is a force for good in the world of music, is almost Swedenborgian: salvation in the form of a fully licensed streaming-music service where you can find every record ever made” (Seabrook, 2014).
Since the establishment of YouTube in 2005 streaming media has been a general phenomena (Snickars & Vonderau, 2010). Still, novel forms of access to streaming music also needs to be understood and treated in media specific ways. In fact, as a media form suited and unusually fit for the digital transition, popular music has in more than one way been situated at the vanguard of media industry developments during the last two decades—from the ’devastating’ file sharing á la Napster to the promoted ’solutionism’ brought forward by Spotify and similar services (with the irony being that both rely on the same communication protocols).
In this article, I will make a claim that throughout these digital shifts and changes within the music industry, more music has been a recurring lead metaphor, as well as marketing gimmick for digital music consumption—even if the promoted uniqueness has rapidly become totally habitual. The underlying idea of the article is quite simple, albeit important I will argue. On the one hand, digital production and distribution of music has during the last decade led to a situation where particular songs are also (and always) part of the ’whole history’ of recorded music. The lure of streaming services are, after all, that they offer (almost) everything recorded. But on the other hand this sort of archival mode of online media—all in the form of a giant and (more or less) inflated database—also runs the risk, or (depending on perspective) has a technological and inherent ability to go berserk, and potentially burst—and thus, completely undermining classical notions of archives and/or collections as trusted and secured repositories of material and/or cultural content. Spam or automated content generation are obvious examples, as well as the bot culture currently in vogue online. The latter, I will argue, is not only underestimated and poorly understood, it is also way more ubiquous than regularly and publicly apprehended. Estimation vary, but it is often said that around a third of all web traffic is nowadays non-human. More specifically, some 20 million ’users’ on Twitter are bots, more than a quarter of all changes on Wikipedia (the world’s fifth most popular site) are done by machines etcetera. Hence, machines pose a big threat; on burnerbrothers.com, for example, for a dollar each scripts or bots will ”get you plays & listeners on Spotify”. Then again, humans can also subvert music catalogues as ’the artist’ Matt Farley, who apparently has earned more than $20 000 from his music since he has released over 14 000 songs. There are, as a consequence, different ways where aggregating musical content bot wise or à la web 2.0 runs the risk of technological back-fire, damaging the very notion of what a musical archive is, should or could be. More music does not necessarily mean better music.
Är idag med och “undertecknar” ett akademiker-upprop mot nedskärningen av kulturtidskrifternas ekonomiska bidrag i Aftonbladet: Stoppa attacken mot kulturtidskrifter. Upprop: Akademiker tar ställning mot alliansens åtstramningar. Det hela har raskt och imponerannde organiserats av Daniel Strand, doktorand i forskarskolan i kulturhistoriska studier vid Stockholms universitet. Det hela är så dumt att man tar sig för pannan; “som akademiker kräver vi därför att Alliansen drar tillbaka förslaget och låter kulturtidskriftsstödet vara kvar i sin nuvarande form.” Tänk om – tänk rätt.
Inom ramen för vår grundutbildning i medie- och kommunikationsvetenskap ger jag imorgon en föreläsning om “Digitalisering & mediepolitik”. Det kommer dels att handla om behovet av att reformera kultur- och mediepolitiken i ljuset av den omställningsprocess som digitaliseringen innebär, dels om behovet av att ta ett större begrepp på dessa frågor – och inte stirra sig blind på ständigt återkommande frågor kring nyhetsmedier och demokrati. Jag har tidigare använt Harry Scheins 40-år gamla uppmaning till utbildningsminister Ingvar Carlsson, och den förtjänar att citeras igen: “Det finns en filmutredning, en litteraturutredning, en massmediautredning, presstödsutredning, etc. Dessutom håller jag på med kabel-tv. Är det riktigt att på detta sätt isolera frågorna från varandra? Krävs det inte en sammanhängande informations- och kommunikationspolitik?”
För den intresserade kan föreläsnigen laddas ned som pdf här: mkv_digitala_medier_snickars.
In two weeks time I will give a lecture in Norway at the seminar of aesthetics at Oslo university, From Documents to Data – Following Files within the Digital Humanities. Not quite sure yet exactly what I will talk about – but the relation between heritage object and various metadata attached to these and the inherent data that can be deduced from heritage materials in digitized form … well that is something I have been thinking about lately.
Det är så denna vecka dags för den sista föreläsningen i doktorandkursen i medieteori – och då handlar det om medialiseringsteori. Medialisering betraktas ofta som en sorts lång moderniseringsprocess där forskare försöker få korn på hur medier med tilltagande intensitet påverkar samhälle och kultur – detta i form av en historisk och vågliknande process (som enligt vissa) pågått under mycket lång tid (tänk: alfabetisering). I grova drag är det ungefär så som medialiseringsteori i en medie- och kommunikationsvetenskaplig kontext ter sig. Det är ett forskningsområde som snabbt vuxit i omfång och attraherat en mängd medieforskare i framför allt Nordeuropa. Å ena sidan finns här en betydande forskningspotential, å den andra sidan är medialiseringsteorin synnerligen historiskt anspråksfull (och därigenom ganska problematisk).
För den intresserade kan föreläsningen laddas ned som pdf här: mkv_medieteori_8.