Hacker News Reader

Maybe it's wishful thinking, but I don't believe this will happen, and unless I'm mistaken, there's absolutely 0 evidence for it. People are usually attached to the artist when it comes to music. You buy a Bowie album because it has his name on it. You learn about them, look forward to their future work, maybe go see them live, etc.

Will computers replace EDM and elevator music? I don't know, maybe, who cares?

The next generations will follow pseudo-Bowie, because David Bowie is for old people.

edit: I'm imagining what they would say. I'm not here to offend Bowie fans.

> because David Bowie is for old people

until it's new again, because young people of today are going to become old too and their kids are going to think tik tok hits are old, but will remember when their grandparents took them for a walk in the park listening to David Bowie.

Honestly Bowie already seemed like an "old person" act when I became a fan in the '80s. When Live Aid happened in 1985, a popular take was that it was a bunch of old has-beens trying to revive their careers. Eg. Bowie, McCartney, Led Zeppelin, Queen, Elton John.

Counterpoint: Hatsune Miku

Hatsune Miku was deliberately gamified as participatory marketing for creators, and is also representative of a specific culture.

You can't generalise to pop in general, because being a pop headliner is as much about glamour, lifestyle, fashion, and branding as it is about music.

No one wants an imaginary Bowie album, because it's a glamour-free proposition - a curiosity at best, most likely irrelevant.

People have been trying to make an Idoru for decades for now, and they always fail. You can only sell fantasy and identification when they're implicit. When you try to attach them to an artificial marketing concept it becomes impossible to suspend disbelief and make the fantasy stick.

> No one wants an imaginary Bowie album, because it's a glamour-free proposition - a curiosity at best, most likely irrelevant.

They're all going mad for an imaginary Abba concert

Even when it comes to vocaloids like Miku, there's still attachment to the producer behind the song and not just Miku herself (and several producers have gone on from their vocaloid careers to launch "real" music careers).

Nobody under 40 is buying any albums, at least not enough to matter. Physical music media is skewed towards the pre-Napster demographic. Bandcamp is nice but no one's getting rich off of it.

Physical outsells Digital downloads. The switch happened two years ago, largely due to the growth of vinyl and streaming replacing downloads.

Adele sells so much vinyl they had to build a factory line just for her releases

I read an article recently that claimed that for major genres most of the songs suggested by Spotify are from "fake artists". There are people making music for Spotify for a fee which allows Spotify to keep the cut that would otherwise go to a non-Spotify entry.

It seems like the formula for an acceptable song is well understood. With the necessary push provided by Spotify these songs get streams in the millions easily. Perhaps it's cheap enough to have this music produced that they don't even have to move on to machine generation.

So it’s not conincidence that 70% of the US music market is old music. People might be not so stupid as thought.

My god everything about this eventuality just sounds absolutely awful. Just more and more recycling of the same meme-tastic trash (which isn't that far off from the current state of pop music), writ even more large and just plain worse.

I'd like to think the future is the opposite -- artists/bands managing themselves as disparate entities, forming around small labels or even their own labels, and having someone in charge to bring the resources they need to bear for touring or performance related stuff so that they can focus on their music.

There is no eventuality. Relax. Punks are punks because they can’t not exist. You just can’t comprehend what they’re doing next. Neither can I if it makes you feel any better.

More than 20 years ago MP3.com tried to educate and support the independent artist. Some would get popular on the platform and eventually get signed and remove their content from the platform - a contract with a label was and is a huge draw, even if it doesn't provide the support and compensation they expect.

On the flip side, Eve Selis is an example of an artist who broke on the scene during that time and maintained control over her art https://www.eveselis.com/ I agree that it would be great if more artists followed that path.

In the future, artists and bands will be obsolete, there will just be music. To some extent this already happens, where tracks get copied, passed around and re-mixed so much that no one knows who did them.

The recording industry is notorious for pay for play schemes.

The Internet leveled the playing field for many years.

The industry has found that social media's dominance allows them the opportunity to distract independent artists from releasing their own quality music and also distract the public from hearing indie music through the illusion that social media helps them to be seen more.

The real question that anyone who works on these platforms needs to ask themself is "Is the platform really helping me?"

The only way they help people now is if you choose promoted ads, and it takes a ton of money for an unaffiliated indie label to break into their main feed on each single post, and that's true for any of the platforms that facilitate sponsored ads.

> We don't blink an eye at efforts to do things like complete one of Beethoven's symphonies. But that is going to be increased treading a thin line when it's something like create a new album by pseudo-David Bowie. And arguably over that line when it's extending the hit debut album of a living artist who just isn't recapturing that magic.

As long as it's generating everything on past data it won't. New stuff is interesting because it doesn't sound like old stuff, and usually that takes over and further mutates. I'm not a fan of trap, but it wouldn't have existed if everything was generated by the current gen of software. It moves fast towards an optimum of the current generation, but we're not at actual AI, so I don't think what you describe will happen (at least yet).

> When the software world can convincingly create new songs that sound like the hit songs people like, but sufficiently different to dodge copyright law, then it's game over.

Bands have been creating music like that since forever. Most of small bands are like this.

What worries me is that the copyright industry, being wealthy and politically connected, is not going to go down without a fight.

I'd imagine trying to ever-broaden the concept of "derivative works", so your "completely new that sounds like Bowie" falls within their catchment.

Or aggressive license mandates on the data set used to train your ML model, so you aren't legally allowed to use it to produce competitive products in the first place. (In that scenario, I can imagine a run on used-CD stores as people rush to get unencumbered source material for that purpose).

On the other hand, it definitely feels like chasing local maxima. Yeah, they can try to fence in "the best song David Bowie never wrote", but that will just drive experimentation to unowned corners of the market. The AI-driven music industry might be a pivot on the scale of going from, say, Mozart to EDM in a matter of years.

It will be possible for computers to generate the content but will they be able to generate the culture that surrounds content.

I like your analogy of the tidal wave but without computers being able to generate the culture I don't see it coming.

Take football [or soccer]. Whole sports seasons can be simulated and are in games like Championship Manager. But these have never changed the collective & cultural shared experiences of real world sports.

> When the software world can convincingly create new songs that sound like the hit songs people like, but sufficiently different to dodge copyright law, then it's game over.

In what way?

This is already a thing in shoes, clothes, handbags, watches... Yet the original brands thrive. I'd counter that listening to "cheap knockoff" music will have a similar appeal (or lack thereof).

When software can convincingly create new songs the risk will be so low that it will open up the scene for more experimentation and diversity. People tend to risk averse, AI doesn't care.

Perhaps, but presumably people will still be the ones choosing whether to release or promote a given work generated by an AI, and by your reasoning they are the ones who are risk averse.

I don’t imagine many people subscribing to an AI who can generate and release countless songs on it’s own without some sort of filter to discern desirable content from undesirable.

Barring some large development, having a human to curate the generated material will be a requirement for success, or else drown your audience in a torrent of unchecked content.