> If the CEO of an open-source company said, "Software isn't our product, software is our marketing", would you feel the same way?
Yes, absolutely, 100%! By all means care about image and marketing, but if your actual product isn't good (software, music, or anything else), then who cares?
For software, actual usefulness. A word processor that's trending on Twitter but which makes it hard to actually edit documents is not useful. A graphics driver that actually renders my screen is useful even if nobody's ever heard of it (although I grant that below a certain size maintenance issues might come up).
For music... I'm far less qualified to say, but there's still a notion of (subjective) quality, which is only loosely correlated to marketing. I suppose we could say "good" music is that music which people enjoy, which still is driven by the actual music and not marketing.
My (implicit) point was that software can be optimized for one factor, which is usefulness. By evaluating usefulness, we can evaluate whether a piece of software is "good" or not.
However, use-value makes for a very poor metric for evaluating music (I am ignoring arguments about the "usefulness" of music when it's played on a factory floor, making workers slightly more productive). So what must we use to evaluate its "good"-ness? Anything you propose here will have certain people agree, and others disagree, e.g. one set of people say good music must make you dance, and another set will say good music must make you feel relaxed (the opposite of dance).
When aggregated, these individual differences will cancel out. So the average most marketable music must, therefore, be as bland as water.
I am speaking of the usefulness of software in a much more general sense - does it work?
Adobe Photoshop and GIMP are two competing softwares for a specific task - editing photos. Not everyone needs to edit photos. Those who do are voting with their money, because Adobe Photoshop worked faster and more easily at the same task than GIMP. (This can change, with the recent subscription-model only Adobe is pushing that is irritating its core users, but that's a different topic)
Let's return to dance music. Let's say we create a category on spotify and call it - "does it make you dance"? Well, for a swath of youth, EDM makes them dance. But for a swath of Latin-speaking users, bachata music makes them dance. Let's say they are of equal number in the population. You can see here that even averaging based on the category of "does it make you dance" will produce some rather unfortunate results (a combination of half-EDM-half-bachata that will fail to tug at the heartstrings of anyone at all, but might be vaguely palatable vaguely dance-y playing in the background of a grocery store). However, this kind of music - the one that can vaguely capture both audiences - is exactly the one that will get the most people to listen to it, and based on Spotify's model of quantity over quality, it is the one that will get most rewarded.
Yes, absolutely, 100%! By all means care about image and marketing, but if your actual product isn't good (software, music, or anything else), then who cares?