> From: Burton Coggles
>
> Surely this study is only as good as the dataset used. Reading the supplementary information, the "million song dataset" is slightly misleading in this study as more than half of it was discarded due to lack of date information or duplication.

You're obviously a glass-half-empty person. They still managed to find 464,411 individual pieces they could use, which is surely enough to be statistically significant in this context?

> From: catspaw49
>
> Great research under the "Who Gives a Shit?" classification and obviously there are some folks with a sincere interest in shit! The dumbass lines in the article and on this thread just keep on coming!

Just because you don't consider something relevant (1) or even just interesting to you, doesn't make it shit.

(1)        You may not even realise its relevance. The article proves that the Loudness War is actually happening. This is something that *should* concern you if it doesn't already, because the louder on average music becomes, the less overhead there is to accommodate peaks. IIRC, when I borrowed the CD from the local library, Lene Marlin's "Sitting Down Here" actually clips and distorts as early as about 8s in and constantly thereafter - given that it's an otherwise approachable track with more variety than many in both lyrics and use of acoustic instruments, that's rather depressing.