Hacker News Reader

Comment by iconjack | Hacker News Reader

"The jury was not legally allowed to hear the original recordings of "Stairway to Heaven" and "Taurus" in determining their verdict."

View on HN

This is good. I don't think a jury could accurately identify that one song is "copied" from another just through hearing the songs. They could only identify similarity and then confuse themselves conflating similarity with copying.

There are other means to demonstrate copying that are less prone to errors in human pattern-detection.

How do you mean? If I write a song that happens to have the same words, melody and chords as All About That Bass, should it be assumed I wrote it myself unless somebody catches me red-handed transcribing the original?

During the trial, they did bring in a pianist (defense) and guitarist (plaintiff) to compare and contrast the two songs. With a test like that, if you copied lyrics, melody, and chord progressions from another song, it'd become obvious.

That test only works because it conflates similarity with copying.

50% similarity: likely not copying. 90% similarity would be much more arguable, and you'd have to rely on other evidence to make a decision. With 100% similarity (as in the scenario you originally outlined), it's extraordinarily unlikely that you didn't copy the original song.

Even if % similarity and % copied don't have a direct mapping, there's going to be a correlation between them. It's the correlation that's important, not some mistaken idea that "similar = copied".

You and I seem to agree that at some point similarity is good evidence of copying.

I was just interested to know what the "other means to demonstrate copying that are less prone to errors in human pattern-detection" are, that were proposed as an alternative above.

Reply on news.ycombinator.com