By Elena Tiis
October 21, 2009 Felix Meritis, Amsterdam
In the crowded top floor room of the Felix Meritis, the moderator asks what the composition of the crowd is. Raise your hands. They go up at music labels representatives, managers, producers, artists, IT people, some students of this or that. Possibly the single most unifying aspect of this melange is the frequent ownership of an iPhone… Music & Bits was a pre-emptive kick-off to the Amsterdam Dance Event unfolding from the 22nd to 25th of October 2009. It consisted of a Hack Track in the morning for the technically inclined, and of a conference in the afternoon. In the following, I will loosely sketch out the course of the conference.
The talks begin by an exchange between the Berlin-based SoundCloud’s Eric Wahlforss and the DJ Speedy J. They strike quite a relaxed pose; no powerpoints and no notes, just an exchange to which the audience was invited to participate quite heavily. First, Wahlforss is describing how he’s been involved with music and “web stuff” for 15 years to date. But these have largely been parallel tracks for him. Frustrated with the limiting landscape of Yousendit and MySpace on the web for the purposes of sending and being exposed to new music, and the rote of sending/downloading/emailing in order to participate in it, his musicianship was beginning to move into the direction collaborative software creation. Music had become links that you pass around on the web since 2005, and SoundCloud began two years later. Wahlforss notes that there’s a big value for instant exchange and the new opportunities that technology provides. At this point, a manager of some kind from the audience prompts him about the legality of the matter – “how is this allowed?” In response, Wahlforss acknowledges that there is a grey area of content on the web, and there are certainly mashups of pieces that haven’t been cleared legally, but these form only a fraction of what SoundCloud contains.
Speedy J notes that it’s rather surprising that the first audience question should gauge the legality of grey use only. Importantly, a lot of the musician’s craft is in abusing technologies, and pushing these to uses not set up for them. Stretching these to their limits does the difference, it makes music interesting. At the beginning of his career, about 10 to 15 years ago, he had to often wrap his head around non-musician systems which meant often trying to enter the head of the instrument producer in order to create music. To this respect, SoundCloud is distinguishable by its musician-friendliness: it’s exploiting its possibilities to the fullest.
As concerns the audience question about the limited publicising capacity of SoundCloud, Wahlforss notes that the platform is not really oriented towards it: for the moment it mostly caters for people who already have a network. “We are before viral spreading yet, there is no “viral button” you can push.” Its use is as a tool with which to solve specific problems. SoundCloud’s business model is rather like that of Flickr – one can buy a pro-account in various forms, from single use to labels for their back catalogues but what mostly constitutes it is free use. A live application might be coming out soon, but at the moment it is not too mobile, computer-based, perhaps soon one can use the iPhone as a field mike…
Muxtape’s Justin Ouellette is next with a history of his application. He traces the beginnings of Muxtape in about 2003 when he hosted a college radio and wanted to have log of the tracks he played online for other people. Further, even before that the concept of sharing and mixtapes was inspired by the late 90s culture of turntablism, DJs and remix culture. Muxtape launched in early 2008 as a site hosting mixes of other people’s music but was shut down in late 2008 by the RIAA. In 2009 it resurfaced again as a platform for bands. Ouellette expands upon this aspect. As well as, importantly, on his preoccupation on keeping it stylistically simple.
Five things about design that dealing with Muxtape has taught him:
1. Focused design is all around.
It is centered around experience and engages the everyday. To this purpose, he devotes some time to what seems like an ode to the Galanz microwave in his NY apartment. It fits the space and the controls are simple and present a good user interface. This fascination with simplicity translates into Muxtape’s outlook.
2. Clean does not equal simple.
3. Event or the context are important for the experience.
For instance, in most car stereos there are too many buttons. Using it to listen to music therefore becomes a kind of exercise in frustration; and when an application frustrates you already before you start listening, there’s something off with the design.
4. Don’t throw away old models.
Muxtape itself is modelled after the analogue cassette tape. There is charm to old forms.
5. Limitations can be deceiving.
Mixtapes are 90 minutes long, which makes it easy to understand its limits. One packs concentrated effort into creating a mix in stead of trying out a few off the cuff like with a digital playlist now. The maximisation of choice, like in huge supermarket, is not necessarily the best; the charm of mixtapes was always about how people collaborate to delimit choice.
In response to audience question of whether Muxtape was singled out as a poster boy for the copyrights crowd, Oubliette notes that there might be something in that because Muxtape always was so self-evidently about music and did not couple it with blogging or other types of uses which would have meshed with easy categories. For the moment, he is concentrating on developing the second generation Muxtape as a promotion site for bands. The first generation Muxtape cannot come back because it seems that the industry is not yet ready for it!
Next, Brian Whitman of the Echo Nest gives “a short personal history of computers listening to music, 1999-2009”. He begins with tracing out his own beginnings as a producer of IDM in the New York scene, and his eventual dissatisfaction with the dynamic of “guys sitting in front of laptops looking so serious”. Software making is something that has made him a better musician. He took a PhD at MIT in information retrieval, in an environment which was looking at music as a file and at audio like a text. Algorithms do not understand music, however, to which respect he is very much concerned about figuring out how to get music into music analysis. To illustrate, or rather to soundbite some of these concerns, he played some automatically generated holiday music based on a statistical reproduction of 1,000 Christmas songs. It’s a strange disjunction; I for one can’t hear the holidayness in the machine’s rendering of the 1,000 such songs. He argues that it is necessary to understand language and audio at the same time.
As for Echo Nest itself, it grew out of the understanding that the best music experience is still manual. Data is actually hard: collaborative filtering (recommends other things that people have clicked on) is a bad recommendation system – it “destroys music”. If only collaborative filtering is the only system of online recommendation, the popular acts will eat the minor ones. Music forums like “I Love Music” capture some of the excitement of people about music, and it forms a much better palette of recommendations.
What the EchoNest does is trying to know everything about music and its listeners by processing data, which they sell to social networks, labels, video games etc. It has produced a range of products such as Fanalytics (a toolset for artists), and maintains an open source remixing community and code base which has produced, among other things, Morecowbell.dj for adding cowbells to songs… In the end, Echo Nest aspires to be something like a Google Earth of music.
Andie Nordgren and Martin Roth of RjDj present the concept of reactive music, and the possibility of having a sound studio in one’s pocket. Music not necessarily linear, they claim: encoding music can be done so that it is different every time. They begin with something of a timeline: in 1998 the sound studio becomes software, in 2008 the sound studio is found in the personal music player. This changes the consumption, distribution and production of music.
To make this audible, they showcase recent project: Kids on DSP’s reactive minimal techno. While demoing it live, Nordgren is talking/blowing/clicking fingers into the iPhone microphone which feeds it back into the music in distorted form. Someone with an iPhone tweets their fascination about this to their Twitter account right in front of me.
The RjDj ecosystem is composed of the reactive music player, music scenes which can be uploaded to the iPhone and composer tools. There is an online recording and scene community where recordings can be uploaded and which contains a search catalog. RjDj is not composition software of itself, neither is it attempting to build this up. It is rather about tapping into a “sweet spot” for a listener, to modifying sound in reaction to the environment that a person is moving in. … explains that they were taking technological concepts from, for instance, sound installations for the creation of a mobile experience. RjDj needs artists construing a base scene that the iPhone user can download to be reactively played. So it is a way of making interesting the value of someone else’s recordings. In audience prompt about the nature of their web presence, Nordgren notes that the site does not contain networking aspects yet but it might be on the roadmap. It is an API for musicians, not developers and the app is just a transport medium.
Next Last.fm’s Matthew Ogle climbs up to talk about the online music ecosystem. He stresses that he does not want to give a historical presentation about what has been happening with the application during the 7 years since it was first inaugurated. Human years are like dog years in the Internet world, so Last.fm – as a big dog – is over 50 about now.
In 2002 what would become the Last.fm of today started as two projects: a personal online radio which learns over time and an audioscrobbler which is a desktop media player plugin that tracks what you listen to. In combining the two, the developers got a feedback loop for crowdsourced music recommendation. 2009 was tough in the ad-supported music space because Last.fm’s music licensing and revenue model was constrained to the US, UK and Germany at the same time as the radio service was truly world wide. Making radio listening by subscription only in non-ad supported countries, caused some controversy and a fair degree of hate mail but the application has been weathering this quite well subsequently.
Music is not a product and not a service, but it exists within a “shifting ecosystem of discovery and use”. Last.fm in this sense is guided music delivery, the “connective tissue for your online musical life”. Ogle notes that it is true that Last.fm must be self-critical at certain points: they must acknowledge their own ecosystem – community of users and influence – and communicate better with it. This last point was engaged better in the case of the inactivity bear sign – a bear pops out when the user has been listening for a long time without doing anything. This actually managed to engage a lot of playful response in the form of people submitting various different versions of the inactivity bear. Currently the team is working on combo radios, multiscrobbles and party radio intersections (although everyone’s average music is not necessarily the best of the people involved) as a way of making the online radio listening a more fluid concept. There is also a project for Xbox with Microsoft.
Hypem or The Hypemachine (Anthony Volodkin and Last.fm’s artistic director Hannah Donovan) talk of the inevitable effects of style on music websites. Hypem aggregates what music people talk, or blog, about. In their talk about style, they start with noting that visual designers often last looked at when it comes to developing new applications. Good art doesn’t match your sofa and there are certain principles than need courting in the creation of a good website: user experience design (user needs), limitations, interaction design (user feel), content and visual design (looks). All of these have to be considered together, they cannot be separately developed and fused together later and their grouping like this should not necessarily reflect the order of importance although they are executed like that quite often.
When it comes to a evaluating a site, start with the “ooh” metric. However, it is hard to impress with style alone when this is not paralleled by content. The applications that take or track your data must create an “atmosphere of trust” with people. To this extent, Volodkin and Donovan flick through examples of sites that overstyle and detract from the point of a music site – that is, the music – to sites that are afraid of styling at all. Also, certain examples tap into a very restricted audience - there is an indie feel to the redness of Last.fm to which extent they’ve added the option of “paint it black”. Donovan argues that MySpace can actually manage to be communicative, connecting the public of a band with the music with the help of customisable aesthetics, for instance the page of the band Beirut. One can style without stylizing, which the route that Last.fm has taken.
The look of a social networking site has to contain visual shorthand for: “this site has social stuff on it”. Design culture can grow pocket-like, geographically specifically on the web, for instance Muxtape and tumblr tap into a certain New York geek look to Japanese sites disliking white space, to the extent that they are striving for being as crowded as possible. In short, sites do not exist in a vacuum, and music is messy so there is no reason why it can’t be presented in messy ways online too. Also, a site has to be conscious of what it wants to be when it grows up: does it want a mainstream audience or is it merely wanting to capture a certain audience. All in all, style does not necessarily mean stylisation.
As a last minute surprise revelation, Henrik Berggren from Citysounds.fm offers the audience the invitation to test the iPhone application to be launched next day. Citysounds.fm adds cities and location – pictures from Flickr – to music – from SoundCloud – so that cities have musical landscapes. The popularity ratings come from Twitter.