There is no doubt that Zoidberg and facebook are pissing a lot of people off lately with some unceremoniously weird user interface and feed changes, to pile on top of their previous privacy sketchiness, but this should not distract us from the rise of Spotify, which has already been several years underway in Europe.

I just started using it today, and so this coverage will not be a technical tour de force so much as a socio-technical commentary leading to an opinionated prediction. So far, this service seems like a more organized, quality-controlled version of Grooveshark, with bitrate quality control (160 kpbs minimum, to 320kbps ogg,) a desktop client (no web client) and a mobile app.

For the very large and important "streaming apathy" market, which demands a top notch recommendation service--that which has been the primary strength both traditional and internet radio as well as pseudo-genomically-driven Pandora--Spotify "brandishes" a high-level "Radio" feature which is a complete turd encased in a bag made of recycled turds, which were, in the interest of full allergen disclosure, processed in a facility which also processes similarly worthless turds. In other words, generic as shit. It has only 25 genres an is about as granular as that giant orb thing that almost rolled over Indiana Jones, or that art thing that they blew up in Fight Club, which resulted in the manboob guy getting shot.

However, there is a slightly more useful feature that is too hard to access called "Artist Radio," which is the ability to kick off a "similar artists" playlist by clicking on the eponymous tab while viewing an artist's profile. This is not available for more obscure artists, and the "recommendation technology" driving it is unknown to me, but I do know three things: firstly, it's not driven by the musical genome project, because it is not Pandora, although it could never suck as bad as Pandora even if bands were picked by ramming a dowsing rod into a corkboard while wearing a blindfold because Spotify has 15 million songs to Pandora's 800,000--almost nineteen times as much music; secondly, it seems not that bad or socially irrelevant, given my own knowledge of my favorite genres; and thirdly, let's not be naive--that shit is driven by payola.

As of this writing, Spotify's biggest failing is that their mobile app does not give you access to their "Artist Radio" feature, and driving and commuting time is one of the primary slots the average person uses to seed their brains with new music. There are sites where you can generate "artist radio" type playlists and import them into the mobile app, but you can't just, for instance, create "[Obscure Local Band] Radio" in the mobile app on the fly, and then have it pull up all sorts of similar artists at that particular level of granularity. Of course, as anyone who has ever had something more cultish like "Infected Mushroom Radio" bring up a Chemical Brothers song on Pandora, sometimes NO recommendation is better. I mean that kinda shit borders on "musicologically embarrassing."

Anyway, Spotify premium is available for $10/month. I am not going to talk about anything other than premium, because premium features two important things: access to the mobile client, and an absence of ads. I don't judge people who eschew mobility, but anyone who can contentedly and silently tolerate advertisements in ANY media format is not generally someone I give a flying shit about. So as far as this coverage is concerned, premium is the only thing that exists. If you can't afford $10/month for access to more than 15 million songs, perhaps you need to spend less time clutching your chest and crooning along with your vicarious muses and spend more time securing the skills needed to obtain a reasonable income.

Premium also allegedly gives you access to 320 kbit/s bit rates, which is the highest I've ever seen for streaming. As long as the buffering is not assy, and your data plan is not metered, this should work well, while also pissing off some net neutrality people, as you gobble disproportionate amounts of bandwidth with impunity. I'm kind of skeptical of its ability to work smoothly on top of the layer of fibrous, mushy, cylindrical elephant dung that is 3G networking.

Pandora is infamous to independent artists for its archaic and banal "print CD UPC" requirement, which states that they must be able to import a print CD UPC from Amazon.com as evidence that your 1990s era drink coasters (CDs) are being actively sold on Amazon, in order for them to be willing to import your stuff into their anemic library. This has changed in recent months, but it may be too late for this company which has apparently never encountered the "long tail theory." This weird policy was no doubt in no small part due to their probably hidden and embarrassing need to establish a music submissions rate threshold, in order to compensate for their complete inability to feverishly and manually crunch music into mathemical matrices at ultra slow, chin-scratching, lab-coat speeds.

If it seems like I don't like the Panda-dora [sic] model, you're not hallucinating. Who in the world ever thought that such an anal and left-brained, nerdy method of cultural music analysis such as "attribute" genomics should have ever gone beyond a scientific curiousity, to become a dominant source of cultural phenomena/suggestion data in the first place? This not only creates monstrous amounts of analysis and processing overhead in the technical sense (they hire analysts who must classify each song manually) butit completely denies the fact that social factors combine with musical factors in order to create the subgenre clouds which comprise the DNA of the popular music idiom.

Bands can be "one degree" away from each other (friends), who have played gigs with each other several times, who are associated with each other pretty notoriously, and yet they would never be classified as "similar" by the musical genome project due to very different musical styles. And yet, plenty of fans of one will tend to seek the other out without excessive worry about stylistic homogeneity. This is because band associations are largely social, and the process of finding and relating one band to another tells a story of social evolution, social network development, friendships and tribes, not just a tale of SELECT * FROM INDUSTRIAL WHERE {"four on the floor beats AND melodramatic string arrangements OR wistful female vocals"} = 1. My SQL sucks, but I think my point does not.

Spotify has been endorsed by Zoidberg himself. facebook has a gigantic userbase. Spotify's value-added boasting functionality (boast-tech) integrates well with facebook's own native boast-tech, allowing facebook users to easily indulge the primary purpose of the internet, which is to optimize downtime by bragging about one's prowess, capability, and [sub]cultural and aesthetic refinement as often and as effortlessly as possible, so as to maximize self-esteem, social status, sexual opportunity. Much like porn, anything which allows people to either look cool, or to fantasize about themselves reaping the benefits of doing so, is poised to splam into massive popularity.

I can't cover Spotify because they are privately held, so instead I willshort the balls off Pandora. This overly academic and deductive song analysis model will collapse upon itself as soon as Spotify's facebook app userbase tips to the point where basically the same associative subgenre data can be gleaned from 4 million+ Spotify users' shared playlists and Likes, all of which the music genomicists were attempting to map out with small parts isolated and destroyed. I give this about 12 months, at the end of which time, I expect to see P worth about half its current value, or about $6/share, as lazy, conservative late-adopting listeners continue to falsely bolster its price.

Partnered services with full libraries like Spotify will eventually master drainage, drink Pandora's milkshake, and leave it for dead like a used, unnamed canary in a data mine which has been wracked by the ass-gas of collated self-interest. Content is king, and a lack of content only reigns when one is protected by a "lack of competition castle," which is about as useful in the real world as a "jump to conclusions mat."

Pandora will never sign an agreement with facebook, nor will they sign one with Google, even if the Google+ userbase tips, because their library is woefully incomplete, and they lack the resources to compete in that area. Why do you think they IPO'd in the first place? Because they know that this is a time-critical race, due to the catch 22 that you need the userbase in order to mine the data with which to outshine their panda-rific scientific recommendation system. Pandora will languish in obscurity as soon as Spotify or a similar Google+ partner's userbase tips, because then they will have all the facebook musical "Like" data to mine all of the "band names off all the bookbags" to power their radio recommendation functionality. A quality recommendation system for mobile will be the impetus which ironically steals niche-negligent Pandora's "radio" niche, and thus shines a bright light on their incomplete library.

To continue grounding and pounding Pandora, in addition to "the long tail" concept which apparently whipped right over their head, which is forgivable due to their legacy structure, it appears that the concept of "crowdsourcing" also passed them by, in an effort to probably try and maintain an untenable level of product integrity in a hostile market. At least in any meaningful way, above and beyond the questionable utility of "thumbs up/down." The longer Pandora exists, the more songs they have analyzed, and the more thumb votes they have received, but the less accurate their recommendation system seems to be, and that's probably because money, not the science-shtick, is now driving those recommendations.

This is why facebook should BUY both Spotify and Pandora, and put Spotify in charge of integrating Pandora's attribute algorithms, so that Spotify can cannibalize and use the music genome project. Spotify should then build a song classification interface into their software, which opens up genomic analysis to the general public (as with the wiki concept) since the bottleneck is the time-intensive genomic attribute analysis, and then the actual users of social networks can determine all that data for them. Reward the crowd with some sort of publicly viewable ratings rank page, allowing top song classifiers to successfully leverage some brag-tech themselves, and you're done. Pandora and Spotify, combined into one.

Spotify will be an important IPO, contingent upon two very important things: firstly, that Zoidberg's megalomania does not upgrade/sidegrade the entire world right on over to Google Plus, and secondly, that Spotify quits fucking around and implements real "Artist Radio" functionality in their mobile app, asap. I mean, that dumb fleeting iLike shit was able to post "look at me" updates to facebook too, but it did not have boss-assed granular radio recommendations for both mobile and desktop.

People want to find new, well-recorded and well-encoded music while they are driving or fucking around on the subway. Most working people who aren't minors and/grounded do not feel like wasting their time constantly building new playlists, and even then, no matter how large or diverse their playlists, there will always be days where the averagelistener is beholden to complete whimsy and the element of surprise, and this is what has sustained the concepts of "radio" and "channelization" for almost a century. It has also sustained Pandora, but not for long.