The Shazam Effect (2014)
theatlantic.com
theatlantic.com
Great article regardless. Glad to see more peoples take on it.
And because they're terrible at this type of optimization problem, they're getting stuck in a local minimum (the space is anything but convex), which happens to be repetitive, awful, and targeted at the largest common denominator.
Treating music as an optimization problem may not be a bad thing in itself, the issue is being unable to solve the problem properly.
Iterating endlessly on a single recipe is just not how you build a sustainable artistic business. It makes money in the near term, but it is doomed in the long term. The music labels and Hollywood will have to realize this at some point.
I presented and DJ'd on a community radio show for a few years, and had a bit of a cult following around here, however my main focus was local talent, and was decidedly not the mainstream nor even the mainstream-indie like Triple J plays. My audience was limited by that.
People don't really want to hear new things on the whole, and if they do, they want the new thing to sound like the old one.
Example: http://www.forbes.com/sites/parmyolson/2014/08/20/shazam-hit...
Article makes it seem like Shazam has metrics like a social network, but my experience, and everyone I've ever seen is that it's rarely used -- and simply to quickly identify a song.
This latest one in The Atlantic implies people use it far more frequently then I think they do, and I'm again, very skeptical. And even if they do, they don't shazam hits they know, or songs about to be hits, but they shazam that random song they can't identify -- not sure that's an indicator of hit'ness.
From user numbers, to capital raised, to stories like this, each time I read about Shazam, I say to myself "Gosh, I really love the app, but can it really be this big and lucrative?" I feel like I'm missing something.
I hope I'm wrong.
One of the Shazam folks gave a 10 minute presentation at O'Reilly conference.[1] She gave an example of how early Shazam datapoints correctly predicted that Katy Perry's song would beat Lady Gaga's song. (The singles were released at the same time.)
Perhaps she was cherry picking clear correlations but it doesn't seem too far fetched that the early trends it sees from users is a reliable prediction of a future Billboard chart position and sales.
[1]https://www.youtube.com/watch?v=mcTPvxo8SXY
deep link at 3:32 that compares KP and LG:
I personally rarely use Shazam, but when I do, it's more often then not a song I don't know at all rather than trying to remember a song's name.
So someone like me hears a song and wants to identify it. Then another person, and another. So you have three people in a specific location at the same time looking to identify a song. That would suggest that those people like what they hear. You know what the song is so you know interest has piqued. If you cross reference that with radio plays at that time you have a new metric that you couldn't measure previously - reaction to a radio play. It's incredibly interesting because of the type of data it collects, which to be honest I think the article explains explicitly and very clearly!
Perhaps this was obvious to everyone else.
Personally I've never used it.
they have 2min ads on feature films showing music news.
i bet this is but paid PR. they are desperate.
the music id business is so dead. Google, apple, yahoo... everyone does it!
<troll>but agreeing to apple terms, show that a company is already in bad position to negotiate. :) </troll>
now let the downvotes begin.
That doesn't sound right. On average one identification for every 16 downloads? I'm guessing that second million should be a billion.
I'm currently in the planning stage for a similar project and this might end up being really helpful for me.
EDIT: Thanks guys.
That's probably too obvious though I'd imagine there's something else going on as well.
[1]: http://www.ee.columbia.edu/~dpwe/papers/Wang03-shazam.pdf
[2]: http://www.royvanrijn.com/blog/2010/06/creating-shazam-in-ja...
I cringe and then laugh inside whenever I see tech writing like this.
If one writes a program to analyze Shakespeare text, he might describe it as "converted the text into a multi-dimensional vector space" or "mapped ngrams into a point cloud" -- and that would be acceptable lexicon for a Dr Dobbs Programming Journal or possibly ArsTechnica. However, for The Atlantic, it seems reasonble for a journalist/editor to write that as "convert the plays into data."
Yes, Shakespeare the text itself was already "data" so in a pedantic analysis, saying "converting data to data" seems pointless but that's not the level of indirection the general readership is at.
I'm on the cusp of deleting Shazam as useful as it is it's getting more spammy as the days go by too many notifications of junk.
I LOATHE NOTIFICATIONS that are junk! Far too many apps do this. I turn them all off now. This is why we can't have nice things - appdevs cramming too much stuff into stupid notifications.
Years ago there was a GNKSA - Good Net Keeping Seal of Approval. This was a set of criteria for Usenet clients to promote sensible usability and formatting standards.
There's probably a niche for something similar for apps.
https://en.wikipedia.org/wiki/Good_Netkeeping_Seal_of_Approv...
Context? What did they do?
They get around it by making up a username or pay someone to submit.
Physorg is another terribly spammy site often with misleading headlines.
Both phsyorg and the Atlantic have been banned from reddit at least once if not multiple times I'm not sure about on HN.