hearsay

A little friendly back and forth today over email about the new Genius function in ITunes. Good luck making it to the end!

On Thu, Nov 20, 2008 at 11:47 AM, freed wrote:

this is a pretty nice mix. care of genius itunes.

On Nov 20, 2008, at 11:49 AM, topomodesto wrote:

in terms of resurrecting forgotten tracks, i don’t really understand what this has to offer that party shuffle does not.

On Thu, Nov 20, 2008 at 12:06 PM, freed wrote:

there is no intelligence with party mix, it selects at random, with some basic preferences. so you may get, as i just did, a Buck 65 track next to Deer Tick. genius algorithmically selects related tracks. that may or may not be successful, in the case of the list i just sent you i think it was successful. its kind of like searching on google, they algorithmically predetermine what comes to the top of the list. the below explanation is from computerworld, but i like the attached image’s explanation better.

Genius is designed to update its analyzing capabilities not just by the anonymous data it collects from your iTunes library but by the data collected from all users. Theoretically, this should mean matches for everyone get better as time goes on. Privacy advocates will probably question just how anonymous this information actually is, particularly when the feature also requires an iTunes Store account/Apple ID to function. Even without that stipulation, the question of privacy could easily be raised based on a computer’s network identifying information.

What isn’t completely clear yet is just what information Apple is using to generate Genius results and how that information is being analyzed. Though the results in my initial testing are more or less spot-on, I have noticed a couple of interesting tidbits.

Genius can’t create playlists for a handful songs because it “Genius is unavailable” for them. Ironically enough, some of these songs will show results in the Genius sidebar. In a handful of other cases, the Genius Sidebar reports I’m missing songs related to an artist even though those songs exist (albeit not fully tagged) in my library. To further deepen the mystery some of the songs that show up in Genius playlists sound great next to the song used to generate the playlist (showing Genius is working pretty well) despite having misspelled artist and track names and no additional tags.

Genius definitely seems to be pulling information from more than just the expected artist, album, and genre tags as well as the more obscure beat per minute tag (which isn’t populated for anything in my library except for iTunes Store purchases). Given that each iTunes library stores a wealth of additional information about music (play counts, skip counts, ratings, playlist inclusion, and the dates a track was added and last played to name a few), Apple certainly has a great deal of information available for Genius to use – most of which seems to be encapsulated in a new file “Genius.itdb” in the folder containing an iTunes library. No doubt this file will be synced to new crop of iPods allowing them to generate Genius playlists on the fly without a computer

Whatever the magic is, it definitely seems to work pretty well (though it might prove helpful to be able to tweak some of the settings, or even know what they are) and it will be interesting to watch in the coming weeks to see the changes that occur in Genius results as Apple begins to analyze more data.

On Nov 20, 2008, at 12:09 PM, topomodesto wrote:

but what if i want those random occurrences? what if i want to be the one who decides what goes with what, not a google-like survey of internet idiots?

On Thu, Nov 20, 2008 at 12:19 PM, freed wrote:

then make your own playlists and refuse the possibility that a growing collective intelligence may have something insightful to suggest about your 200g of music, a volume so large that you cannot rationally conceptualize it for the purposes for making a mix. i am not advocating the wholesale acceptance of these playlists, except perhaps for convenience. they certainly contain unexpected connections between disparate genres, connections i would not have made. the key is in the design of the algorithm, which is to say, understanding and selecting for elements of your own taste. how many times do you think when making a mix, “next i want this kind of track”. genius/algorithm gives you 50 choices to fill that slot where you may only able to come up with 5 in a reasonable amount of time.

On Nov 20, 2008, at 12:24 PM, topomodesto wrote:

sounds like you are replacing your brain with an algorithm. happenstance with science. eclectic diversity with adherent monotony. the dj with a calculator.

awesome! sign me up!

and yes, you are advocating them. wholesale.

On Thu, Nov 20, 2008 at 12:51 PM, freed wrote:

actually, i am advocating a ‘both and’. so to get at that let tell you this. all my hearsay mixes were generated by an algorithm (they weren’t). would that change whether or not you listen to them? no, because its about ends, not the means. that is because mixes are not indexical, they do not reflect a sign of their making. that is why i like doing the cd covers.

so dispensing with a means based justification for the status of mixes i realize that i am making argument for jack fm. which means that i actually would support jack fm and similar corporate music institutions if they did not generally promote shitty music, sponsored by payoffs, incentives and advertising, preventing the possibility of connecting with authentic and creative music.

of course i don’t want to replace my brain with an algorithm. but i also recognize that i am not an autistic music selecting genius, whose one talent and ability is to compile incredible mixes. my interest in making a good mix is limited by my music knowledge and is tempered by competing interests, responsibilities, and commitments. so why not turn to an algorithm in my critical selection of tracks for a mix, especially given limited time and resources. it will inevitably lead to a better, newer mix, filled with more variation, that challenges and extends my music knowledge rather then reifying the same old selections, relationships, patterns, and emotions.

On Nov 20, 2008, at 1:14 PM, topomodesto wrote:

i strongly disagree with your jump from algorithm to “inevitably better mix.”

and i think this is about more than the quality of single mixes. it is about saying that an understanding of music and emotion is beyond our capabilities: that we cede authority of our taste to a foreign entity. the fact that it is an inhuman set of equations determining how music affects us is a disconcerting joke straight out of hitchikers guide to the galaxy. that participating in it creates a database of music under the control of a commercially driven corporation is just a way of re-branding commercial radio: a new frontier for profiteering to influence art.

so yes, enjoy. i like the individualism of music taste. the nuance of college radio dj’s selections. the encyclopedias of music that we all are. my answer to the will sheff concern that we have too much music to listen to is to put it on random. or get thoughtful recommendations from people. not equations.

On Nov 20, 2008, at 1:46 PM, topomodesto wrote:

i also notice that the selection of tracks is rather skewed to the “hits.” perhaps this is your collection – i don’t know, but “laid” is probably the only james track one would consider. “wolf like me” is THE tvotr hit. “fake empire” THE national hit. “mushaboom” is the biggest pre-itunes feist track. “young folks,” “smile,” “the funeral”… and i still think “two silver trees” might be the worst song i’ve heard all year, i don’t care how many plays on kcrw it got. the rest of the album is way better.

On Thu, Nov 20, 2008 at 3:08 PM, freed wrote:

i’m not laying down a ideological polemic stating that computers and corporations are better then individual taste. what i am arguing for is the possibility that an algorithm can introduce more variation and make more choices then we can, thereby empowering us to better fulfill one of our primary goals, so seek out and recognize the best music out there that is not adequately appreciated. algorithm as a tool. my answer to sheff’s concern is that the excess of music empowers the critics over the musicians. the systems of selection, what blogs you read, what music writers you trust, essentially the tastemakers are eclipsing the artists. i would hate to see that happen. some dj’s do it an artful way, creating something new from a kit of parts, but far too many rest on the quality music of the artists themselves without injecting anything new or creative.

so what i think we need is transparency in this algorithmic process of selection, something that does not exist with ‘genius’. what are these rules that they are using to match songs to each other? more importantly, i want to make my own rules. so to bring this back to making a mix, here are some rules i want to test out for my next downtempo mix:

20% songs from the top of college lists in the 90′s that are also in the genre shoegazer
30% songs from the top of college lists in the 00′s that are also in the genre shoegazer
20% songs released in the past year similar to The New Year (they’re one of my favorite bands)
10% songs similar to Slint’s Spiderland
10% songs top 10 blues songs from the Billboard Charts in the 50′s for flavor
10% songs top 10 postrock of all time with at least 1 minute of politically disenchanted mumbling

and i want “genius” to produce 25 mixes that each can fit onto 1 cd…

here’s the thing, and this is a slippery slope, but i would never trust the algorithm to develop the concept for the mix. given the time, i would never take a single mix from an algorithm and publish it as my own. i would never give over the ordering of the tracks. i think algorithms are incredible powerful as filtering devices that limit, in a designed way, what the available options are.

the conclusion here i think has to do with the definition of ‘best mix’. the algorithm is going to give you a better set choices. if the choices are not good, its because you didn’t set effective parameters. so i think in the end it has to do with personal meaning and shared experience. it has to do with the way the you want to relate to the mix, and how the intended audience should relate to the mix. the ‘best mix’ is perhaps the ‘imperfect mix’.

all this though, really does tend to commoditize the song/artist no? perhaps that is where ‘genius’ is ingenious. it generates a mix based on one song.

On Nov 20, 2008, at 3:23 PM, topomodesto wrote:

sheff’s argument was more about filesharing; that you download so much you couldn’t possibly give it a fair listen, and it becomes more about collecting a disposable item than understanding the item itself.

i’m just skeptical about what factors into the algorithm’s filter. when i took a second look at your mix it seemed that it might be filtering out the really intriguing stuff for the status quo popular stuff. i don’t want to hear a mix of hits. i want a mix that surprises me. maybe there is a track by the national that i skip over and actually is a hidden gem. maybe if you go outside of your itunes library there is some 70s rock track that has stood the test of time but is unfairly forgotten. mixes are a dime a dozen, but one with care, taste, a pace, and emotion is worth at least the cd it is burned on.

and now we’re coming back to the means rather than the end, aren’t we? for someone who was arguing “mixes are not indexical, they do not reflect a sign of their making,” why is the process of the mix so important all of a sudden? it all comes down to our basic disagreement when we started hearsay mixes. i don’t want rules imposed. i trust my ear/mind as the filter. i like my mixes. i think they have depth, surprise and range. i don’t need a scientific filter limiting me.

On Thu, Nov 20, 2008 at 4:02 PM, freed wrote:

perhaps that is the confidence of a music ‘genius’? i am not so decisive. my mixes in the end are always aspirations, rarely do i settle on series of songs that i feel are complete. but this whole process of selection and organization is only important to the one who makes the mix. perhaps my feeling on that is best summed up by david foster wallace when he said, “…a writer has to understand that his primary allegiance is to the reader.”

and this is the most troubling thing to me about blogging. entertain me, inform me, but don’t just tell me about what you like, i care not for your preference. and this is where i think your mixes are great, they are always engaged in a deeper conversation about what a song is and what it is trying to say, rhetorically and musically. you engage shared experience and personal meaning.

but i do think that ‘genius’, since it selects from your own music library based a certain song, can inform on the selection of what goes into a mix.