If you have not tried MOG’s new streaming music service, then take a few minutes and watch the demonstration video. MOG’s streaming music service features a fader that enables music fans to simply adjust the flow of new (relatively unknown) music that’s inserted into any MOG music stream. I believe Echonest powers this feature. Enabling music fans to completely control their music experiences is no longer a pipe dream, it’s now a must-have feature that will appear everywhere over the next twenty-four months.
Quality faders will change the promotion game.
I think it’s relatively easy to enable consumers to control just about any part of their musical journey, but what about quality? Quality is subjective (or maybe it’s not?), however with artists creating over a million songs a year, the absence of a quality fader (filter) reduces the flow of new music to a trickle (the new music fader stays pinned to the left), as no music consumer wants to be burdened with the need to sift through a truckload of poorly written or poorly produced songs. (Note: I believe Echonest is already (somewhat) filtering for quality (hotness)?)
In my opinion, a quality filter-fader that artists can trust - changes (ends) the promotion game for everyone. When we get to a point where quality, combined with other attributes, can be faded in and out, the entire industry will terminate the marketing department and hire a gaggle of people that can improve quality (subjective or not, it will be measurable). Promotion will become something you (possibly) do after you measure “quality”, not before.
As I don’t want to be labeled a “futurist”, I am going to qualify my next sentence as “food for thought”: Imagine any promoter (promoting anything to do with music), simply adjusting her quality fader to show all rock bands, with quality scores above an eight, available on August 8th, with more than 1,000 certified fans, and willing to travel to Lancaster, Massachusetts. Note: if the quality filter-faders work (are trusted), then the 1,000 certified fans are probably going to be acquired based upon…pre-measured song and/or entertainment quality and not much else.
Will the quality faders work?
Do spam filters work? I believe we will rapidly reach a point where consumers are so sick of artists engaged in self-promotion, that it won’t matter if the quality filters are perfect or not. Consumers will use whatever is available, and if switching costs (between music services) are low, the consumer that really cares whether or not the latest song from any-random-artist has been “unfairly” filtered out, will switch to the services that have the (perceived) highest-quality quality faders (which may be filter-faders that can be “personalized” and “taught” by users).
Another side note: I’m not one of those people that believe that there is an audience (somewhere) for just about anything; if only that audience could just be found (on this planet). There’s a (narrow?) window for quality (not rubish) that 98% of the population will more or less agree upon, simply adjust the fader (left-less or right-more) to enlarge or narrow the opening.
Which quality filter will you trust?
Once refined, would you trust any one of the following methods: enough to withdraw or improve a song (for example) until its’ quality score improves?
Method One: Using “crowds” of music fans to measure quality.
Sites like OurStage enable regular fans (anyone) to essentially filter and funnel music. Due to the complexity of song adoption (see the song adoption formula), I believe music fans are not going to be the best filters. 10,000 music fans, given 1,000 decently produced songs, would probably filter in (categorize, tag and approve) 800 different songs. I am not a fan of solely using crowds of consumers to filter music; unpaid consumers will be easily overwhelmed by the task of screening a 500,000++ songs a year, and (IMHO) most consumers lack the skills and the reference points to do the job adequately (where artists would trust this filter alone). See this related Washing Post article.
Method Two: Using market trending and statistical information as a quality filter.
It’s now possible to gather up just about every piece of information pertaining to an artist or a song (tags, plays, mentions, shares, downloads, purchases, attendance, etc), and then use this information to filter (in and out) artists and songs. Although usable (as a filter), this method will exclude (to some extent) those artists that either don’t want to, or suck at, touring, promoting or actively acquiring proof-of-popularity data (yuk). This method also suffers from the cold-start problem where new artists and new songs (that may be great) have yet to accumulate a single bit of market traction information to be filtered by. (Preparing for this post, I took a peak at We Are Hunted’s Compete graph. Draw your own conclusions.)
Method Three: Using computers and algorithms to filter and funnel music.
Last year, I spent money and months researching and trying different systems that use software algorithms to measure hit potential and quality. Here’s my serious, unscientific opinion: neither artists nor passionate music fans are ready to make serious actionable (economic, career, music adjusting, etc) decisions based purely upon the analysis of a computer. These systems are just good enough to be fun or interesting, but not yet good enough to make course-changing decisions by. I believe this technology will improve over time, however the error rate may remain too high for artists to fully trust. See this related (unscientific) post and comments for more thoughts on machine learning.
Method Four: Using collaborative funnels that are staffed with music experts.
Another method to measure quality would be to motivate (hire and pay) five hundred (for example) music (and/or industry) experts (those that know the difference between a great song and a good party), and then have them work (blindly) within a collaborative funnel (software that moves tagged information down the funnel) to measure quality. It seems to me that this is somewhat how Pandora works. Revenue models aside, I would trust this method over exclusively using methods 1, 2, or 3, as I believe that most people that have been working in the music industry for the last ten years (and even super committed music fans; once verified) are far more suited to measure quality than the rest of humanity.
Method Five: Using a smart combination of all of the above.
Each of the methods described above have some value. Smartly sequencing (stages of the funnel) and combining all of the above (with music recommendation technology) is probably the best way to filter-funnel music for quality. Over time, operators of funnels should be able to tweak and adjust their funnels for accuracy, and end-users should be able to teach and personalize their faders.
The bottom line: the world needs quality funnels (spam filters) to separate the wheat from the chaff. Like it or not, they are coming. Will these quality control mechanisms change the way artists navigate the music industry? My bet is yes they will. Stuff won’t get very far until it has been rated (so promotion will come second). The operating margins within the entire music industry ecosystem are just too thin to continually subject consumers to unwanted artists and songs (which causes people to navigate away), and/or to keep paying people to find (or not find) quality, as doing so should be (needs to be) hyper-efficient.
If you read my last post, doing this (building or finding a trusted funnel system), comes before that. I am targeting mid-summer of 2010.
about Bruce Warila