In post 49, hitogoroshi wrote:I've won with Andrew Bird's Bowl of Fire (folk-swing), Hudson Taylor (new folk, maybe folk-pop), The Ditty Bops (bubblegum pop), Beltaine (neoceltic), and Micheal Guy Bowman (electronic). Throw in my second and third placers, we get Less Than Jake (ska), Humanwine (dark cabaret), The Cat Empire (ska/funk), The Planets (orchestral), Matt Dusk (swing), Tokyo Ska Paradise Orchestra (ska), The Format (indie), Toh Kay (hard to classify on his own: a dude with an acoustic guitar), Codeine Velvet Club (alt. rock), Reverend Glasseye and His Wooden Legs (dark cabaret).
A pretty eclectic mix of stuff gets to the top. There are certainly some trends: songs relying solely on the lyrics like traditional folk tend to do much worse, more immediate songs tend to do better, technically impressive stuff often gets very highly rated independently of what CDB would call the "Ipod test". But overall I think SC is only "easy to win" if you happen to have a really great song, which is sorta the idea and a good state of things. I mean, Ebb And Flow won, but I wouldn't say it's because you figured out the secret code of SC or something - you just found an incredible song, y'know?
On that note - I'm going to randomly link this as an example of a song that I love but wouldn't enter in the SC. This isn't being said in a spirit of "YOU PHILISTINES!" - it's just not conducive to the environment of the SC.
I actually disagree with the idea that the SC isn't easy to win - I have methods that I've used in the past to determine whether a song will be popular that have provable success. I would market test it through scumchat (which at the time contained a significant-ish portion of the Song Contest entrant base), use feedback to alter my entries to things that tested better, then went with it.
I used this method from contest 12 through contest 20, during which time I won twice, came 2nd twice, came 3rd twice, and had 3 top 5s (which were my lowest finishes). I then broke it when I entered Janelle Monae (which was entered because hey, free wins) and more importantly Michael Franti in contest 21. Since then, I've continued to have success based off of the knowledge gained through said method of market testing. I feel that if I want top 5 in any given contest, I can get it at the snap of my fingers. Winning takes a bit more effort but I can shop it through similar channels to see if it'll work.
I would highly recommend that people do not use this method in the future - it's really not worth the effort now that we have 60+ contests worth of data to look at if you really want to know what tends to win. In addition, there's no real pooling of standard Song Contest entrants (there is more of a rotating cast as opposed to the same people every time) like there was then, so it's not reliable. There's a very, very distinctive pattern and I'd kind of like to do more data testing (this is where I really wish Pandora's music genome project thing was open-source - if I could look up every winner I think we'd see a pattern of "tags" strewn throughout) on if I ever have the time.