RSS
  • http : / / www . mog . com / spaceling

    13. Jul. 2006, 1:50

    From here on out, my music blogging will continue on mog.com.

    Meanwhile, I will continue to enjoy the charting and statistical resources that good ole audioscrobbler has long provided. Long live audioscrobbler and audiophile based social networking!
  • Mog and Audioscrobbler

    24. Jun. 2006, 2:57

    If you were reading boingboing.net this past week, you might have been surprised to learn that the new Mog.com is the first social network site with the idea of connecting audiophiles based on their listening habits. When David Pescovitz asked David Hyman where he got the idea for Mog from, Hyman didn't reply that his idea had already been implemented by others with similar concepts. Rather, Hyman explained:

    My friends and i always turn each other on to music. i wanted to automate the process of sharing what was in our collections and what we are listening to without having to do the work of typing it in.

    Could it be that Hyman wasn't aware of the a handful of free (or pay) sites that already do this? (See Gnoosic, Echocloud, Audioscrobbler, last.fm, musicbrainz, musicplasma, etc...). It would have been insightful to read why existing music-based social networking sites such as audioscrobbler/last.fm were too limiting in comparison with his vision. After all, there is certainly quite a bit of room for improvement among these sites.

    From my few days experience with Mog, there are some significant differences, large and small, between it and other audio based social networking sites:

    1. Mog offers an option of "undeclared" for users who prefer not to identify as male or female. Compare this with the last.fm option of "unknown." Neither site offers much of a choice for gender queers, but at least "undeclared" offers a less offensive option for those who know perfectly well that male and female don't make sense for who they are.

    2. Like last.fm and musicbrainz, Mog seems to be having trouble with the data quality of its established database of artist and album names. (Perhaps Mog will experiment with user's working together to clean up this data? Audioscrobbler tried this with mixed, but I think mostly positive, results.) This is my biggest complaint so far with Mog.

    3. Unlike audioscrobbler, Mog will scan your music collection up front (and post it to your user page). This could theoretically be a boon for filesharers seeking ultra-rare tracks... or for media companies looking for quasi-anonymous users to litigate against.

    4. Mog offers uses the ability to customise the look of their page. Last.fm maintains a single template for all user sites, which drew the ire of many users of the old audiscrobbler site. Last.fm could learn a few things from Mog.

    If one has the ambition and audiactiy to create yet another of these sites they should aspire to a greater goal, and more pioneering goal than Hyman described. How about creating a site which celebrates less hipster exhibitionism, and promotes more creative and/or collaborative behavior. I'd like to see some cooperative tools that aren't forums. The closest we've come to seeing anything like a collaborative project on these sites is cleaning up database cruft (see below). The most valuable collaborative product of these social networking sites is the aggregate statistical data of popular and niche music. But what good is this data other than extremely valuable fodder for sale to media companies?


    I'll be experimenting with Mog and comparing my experience with other free sites over the coming months. I'll be very interested to see how Mog helps to enable communities to form that aren't the more typical bastions of trolling pedants you'll find at musicbrainz and elsewhere.</p><p/> Best of luck to Mog. It's never a bad thing to reference the ideas (both the mistakes and successes) of past inventors. Here's to David Hyman and the rest of his team learning from them.
  • New Audioscrobbler Age/Gender Distribution Chart available

    29. Okt. 2005, 23:24



    Nice colorful new chart! Yum.


    But I must ask, once again, why women appear to not be as interested in this resource as men? The only age group where the usage is roughly equal is the 1-15 cohort. Additionally, the number of femal scrobblers as a segment of the total appears to have fallen since the last.fm/scrobbler transition, from 26% to 9%. That is an incredible drop. Why??? And what do the developers think about this?

    Also, what about the binary gender definitions here at last.fm/Audioscrobbler. Russ had promised that they'd do something about it last April 2005. Do we need to fight this out in the forums again?
  • Grapevine is back, as are feeds; FOAF support on its way

    2. Okt. 2005, 22:50

    For personal reference. I picked up most of this information from http://audioscrobbler.net . I'll be adding more links to this as I find them in the forums. FOAF support, a kind of rss feed which supports the aggregation of personal feed data spread over multiple blogs is on its way (see this post in the forum: http://www.last.fm/forum/21717/_/47009).


    Grapevine
    http://www.last.fm/user/spaceling/grapevine/

    Information about the profile (eg: Name, Gender) http://ws.audioscrobbler.com/1.0/user/spaceling/profile.xml

    Top Artists
    50 most played artists from a music profile http://ws.audioscrobbler.com/1.0/user/spaceling/topartists.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/topartists.xml

    Top Albums
    50 most played albums from a music profile http://ws.audioscrobbler.com/1.0/user/spaceling/topalbums.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/topalbums.xml

    Top Tracks
    50 most played tracks from a music profile http://ws.audioscrobbler.com/1.0/user/spaceling/toptracks.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/toptracks.xml

    Top Tags
    Most used tags by a music profile http://ws.audioscrobbler.com/1.0/user/spaceling/tags.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/tags.xml

    Friends List
    Friends added to this profile http://ws.audioscrobbler.com/1.0/user/spaceling/friends.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/friends.xml

    Neighbours
    People with similar taste to this profile http://ws.audioscrobbler.com/1.0/user/spaceling/neighbours.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/neighbours.xml

    Recent Tracks
    10 recently played tracks for this profile http://ws.audioscrobbler.com/1.0/user/spaceling/recenttracks.txt
    http://ws.audioscrobbler.com/1.0/user/spaceling/recenttracks.xml http://ws.audioscrobbler.com/1.0/user/spaceling/recenttracks.rss

    Recent Journals
    Recent journal entries for this profile http://ws.audioscrobbler.com/1.0/user/spaceling/journals.rss

    Weekly Chart List
    List of dates of available weekly charts for this profile http://ws.audioscrobbler.com/1.0/user/spaceling/weeklychartlist.xml

    Recent Weekly Artist Chart
    Most recent weekly artist chart http://ws.audioscrobbler.com/1.0/user/spaceling/weeklyartistchart.xml

    Recent Weekly Album Chart
    Most recent weekly album chart http://ws.audioscrobbler.com/1.0/user/spaceling/weeklyalbumchart.xml

    Recent Weekly Track Chart
    Most recent weekly track chart http://ws.audioscrobbler.com/1.0/user/spaceling/weeklytrackchart.xml

    Previous Weekly Artist Chart
    Select a weekly artist chart by date http://ws.audioscrobbler.com/1.0/user/spaceling/weeklyartistchart.xml?from=1114965332&amp;to=1115570132

    Previous Weekly Album Chart
    Select a weekly album chart by date http://ws.audioscrobbler.com/1.0/user/spaceling/weeklyalbumchart.xml?from=1114965332&amp;to=1115570132

    Previous Weekly Track Chart
    Select a weekly track chart by date http://ws.audioscrobbler.com/1.0/user/spaceling/weeklytrackchart.xml?from=1114965332&amp;to=1115570132
  • Site Statistics and Demographics

    16. Aug. 2005, 18:01

    Russ has already indicated that site statistics will be coming back . No word yet on my request to integrate user stats (age, sex ratio, country) into users Neighbour lists, and Groups. Until then, here is an old URL which apparently still works:
    http://labels.last.fm/admin/demographic/

    Audioscrobblers:


    Considering that female's accounted for more than 25% only a few months ago, the current statistic (10%) is an extraordinary change. Perhaps these stat charts are unreliable?
  • Tagging Music

    10. Aug. 2005, 19:01

    Tagging, familiar to Flickr, Del.icio.us, and other online social networks based on the collection and aggregation of user submitted data, has finally come to Audioscrobbler. There are many caveats to tags as has been written at length about by tech bloggers and journalists. See here, here, and here.

    So does manual tagging of music files make sense? Unlike photos in Flickr, music files already have a number of pieces of information useful to easily locating them. And these pieces of information are located in the id3 tags already. (At least, Audioscrobbler is dependant on users having correctly tagged their files). The question, of course, rests on the function of the metadata, i.e. how the metadata is used to provide some helpful means of organization, classification, and aggregation for a single user or community of users. Looked at from this light, much of the information stored in our id3 tags is Taxonomic data, data determined by the recording artist themselves (artist, album, and title name, year of recording, copyright and so on). Data appearing in these id3 tag fields isn't up to much argument -- that is the whole basis for services like Musicbrainz and freedb -- to help correctly tag music files. The basis of the idea that this data is correct, is that this is the data which the creative author has agreed to in producing the file and disseminating it.

    But there is no such agreement for Folksonomic systems. In a folksonomy, like at Flickr or del.icio.us, users create huge lists of descriptive, often entirely subjective tags. They have special meaning and provide helpful search functions for that user. Folksonmic tags can also be used for searching by a community when the tag keywords chosen are popular descriptors. Searches can then be performed in both taxonomic and folksonomic realms of classification.

    (Initially, the possibilities of a folksonomy intrigued me as a way the community could determine a useful taxonomy for artist genres. I haven't given up this hope and I'll be looking at Audioscrobbler closely to see if the popular tags chosen reflect ignorant or thoughtful and informed sensibilites. My prediction is that popular tags will reflect trypical mediocraties.)

    Within id3 tags there is room for folksonomic classification. Even in id3v.1 tags, the genre field is ripe for disagreement, and all sorts of information can be written in the comment field. So why doesn't Audioscrobbler simply begin indexing genre or comment fields. There are also, extended fields: tempo, situation, style, preference, available in many id3 tagging applications like the Godfather, Musicmatch, Dapyx, and others.

    For a service already dependant and using id3 tag data, why can't it simply begin collecting taste, style, and genre information included in id3 tags extended fields? This would have the benefit of leveraging audioscrobblers existing tagging experience while removing a service which would may provide a strain on last.fm servers. From what I can tell, the Profile Editor has no Export Data function. There is also no function for exporting Audioscrobbler chart data, but for the effort I'd put in to create my own folksonomy, I'd at least like the priveledge of taking that work with me if and when I leave Audioscrobbler.

    My dark intutition is that by not relying on id3tags for folksonomic classification, last.fm is betting that the service will help keep users using (and dependant) on last.fm. I love the service Audioscrobbler/last.fm provide but I appreciate my freedom more and that means keeping my metadata in my files.

    What are your thought on folksonomic tagging in Audioscrobbler? There is an open thread in the forums on this topic here.
  • What is Audioscrobbler?

    9. Aug. 2005, 21:06

    Here we are with a brand new audioscrobbler site. Props to Russ and crew for their hard work. Now that we have journal space to work with, I'd like to write down some thought I have concerning social networking and music.

    The promise of sites like Audioscrobbler and Echocloud (now totally defunct) was that they could leverage our listening behavior to tease out genre and taste relationships between various artists and bands, and give music lovers a stastically generated idea of what music we might like which we'd nver heard before. Echocloud did this farely well using the gnutella and soulseek filesharing networks as its sources of data, before it suddenly died mid-2004. Echocloud generated a list of similar music artists for any queried artist. The innovation of Audioscrobbler was to collect data for users and provide a relevant list of tailored recommendations based on their listening habits.

    My own interest in discovering network relationships between music stemmed from my experience as a DJ. I was dissatisfied with genre definitions and market classifications as a helpful way for me to understand both music history and find new music I'd like to listen to and play. Getting recommendations from friends always consisted of sharing a list of artists I liked in conversation and then relying on their similar taste to help fill in the cracks. Meanwhile, I'd study up on the music history of the bands and artists I was interested in to hopefully find other obscure gems.

    The possibility of a music map being generated to reflect historical relationships was deeply appealing so four years ago I was excited to discover a project called aMG Music Map which generated flash music maps based on relationships derived from the All Music Guide database. AMG's music maps are no longer accessible, but they did establish that relationships could be derived from AMG's nicely researched artist histories and genre classifications.

    But the idea of deriving and displaying relationships based on historical links lost out to an interest in social networks. The principal had long been established by Amazon, showing buyers the popular purchases of other similar buyers. Echocloud was the first to show how by populating a database with list of shared mp3s on the gnutella network, one could query an artist and find artists of potential similar intertest.

    Representing these similarities as a musicmap like MusiCompass had done was a natural leap. Musicplasma and Gnoosic soon followed up with interesting maps showing relationships between artists. The former relied on an unrevealed source of data, possibly Amazon, while the latter populated its database with user input and created similar maps for books and films.

    After Echocloud died, Audioscrobbler took the idea in a different direction than musicmaps. The appeal of finding related music could also be used as a tool of finding related friends and establishing music communities -- actually creating new social networks based on listening habits -- besides the obvious utility of promoting new music based on personal interest. Although not integrated into the Audioscrobbler/last.fm site, music maps based on audioscrobbler relationships can be viewed at Audioscrobbler Browser (powered by Touchgraph). Another music map implementation based on Audioscrobbler, flash based and currently defunct is MusiCompass (see RNR's comment for more info).

    This is all well and good, but we still need a distributed model for describing historical relationships between artists. Only then can we truly say that an artist which a number of "music neighbors" share in popularity is indeed "similar." What is called for is a distributed knowledge project akin to the (in)famous moderated audioscrobbler database cleanup project, where artist names were corrected by human eyes. The cooperation exhibited in that project was similar to Musicbrainz and attracted a zeal for correction familar to Wikipedia authors and editors. Database cleanup needs to be a constant feature open to the audioscrobbler community, just as it is to freedb and musicbrainz members.

    Meanwhile, the social network provided by audisocrobbler could also be improved by helping users better understand who their neighbors are. Is the music I listen to of shared interest to ther people my age, sex, and nationality? What "groups" at audiscrobbler does my interest in music recommend me to join?

    The possibilities for feautre improvement at Audisocrobbler are certainly exciting and the continued development of the site and interest of the site authors gives me some hope that future features will continue to be created helping music listeners to understand their music better, and find each other more easily.