There have been a few Spotify-clone studies that showed 70-80% of musical success was attributable to "quality" but the ranking of the top 10-20% was essentially random ("luck" to use the author's word). Now consider Dall-E and other art-making tools. If it becomes easier to make quality art, then the luck factor gets much more important, because the baseline quality is higher. So one can ask whether e.g. the Mona Lisa got famous because it was one of the few quality works of its time. I expect that if someone made a similar-quality painting today they would probably have to sell it on the street. The trend is that art's value goes down but at the same time quality art becomes much more prevalent. Meanwhile economic success becomes even more random.
This phenomenon you mention is interesting in other disciplines and topics. The gradient of power law distributions across many measurable spaces looks the same.
There is a steady, near linear association with “quality” (no matter how you abstract this definition), and then the more exponential gains are typically exceptional instances with more unique circumstances for how they were measured along this portion of the curve.
Another widely measured example is income. Most people have jobs with increasing pay commensurate with market demand, but the top have exceptional combinatorial factors involved: e.g. they are BOTH highly skilled AND own a business or have some obscure high risk job, were an inventor of something, receive substantial trust fund income, etc.
The more boring way of stating this is… exceptional results are by definition exceptional.
One thing that interests me about Dall-E and other art-making, is, I wonder if it will eventually lead to individuals making their own high-quality animated feature films and "triple-a" games. Will these tools get to the point that an individual can make a unique triple-a game in their bedroom?
Well, there are already high-quality films / games by small teams with no external funding, like https://www.youtube.com/watch?v=LsGZ_2RuJ2A, Braid, and Celeste. But for long form stuff like these, the tools don't matter as much, it is more like writing a novel where the key is to get something done every day. The main issue is perseverance - a tool dropping the workload from 100 days to 10 days is nice, but it doesn't change the fact that most people will get bored and give up in 10 minutes.
I'm pessimistic. I see DALL-E and other generative algorithms as akin to the camera (or video camera): they are powerful new tools for creating art, and their full impact on the art world remains hard to foresee. But what is clear is that AI-assisted art will become a new medium, and "triple-a" games may just shift to being produced by teams of professional experts at using the new tools.
In fact, the Mona Lisa didn't become an iconic painting to the general public for a very long time.
But, in general, while there's a lot of variation in musical styles/art styles/etc. that a given person likes, I find that there's fairly broad agreement that an expert list of, say, the 50 top classic rock songs are pretty good--among people who like classic rock even if they might disagree on the order.
Yeah but the "classic rock" phase came at a particularly unique time in history when hegemonic record labels and focused their efforts to popularize a very small cohort of artists. And cheap broadcast technology and syndicates made uniform radio the cheapest form of entertainment in human history. I can't say it enough. Mid 20th century America and Western Europe are one of the mlst unique media landscapes in human history.
Monopolistic mass media which reaches hundreds of millions of people is weird. Like most of the mid 20th century we should be cautious about using it as typical of anything.
I could make the same statement about classical music, opera, ballet, film, "oldies," folk for at least some subcategory, etc. and I think it would still be generally true.
Our perception and knowledge of most of these things has been filtered through a fairly narrow, ubiquitous and recent cultural criticism.
What's more, the canonization of these thinga has caused our modern media to refrence and reflect them. Meaning we likely have a predisposition towards them as their structure and tropes are familiar to us. Your popular measure of quality is terribly tainted.
How much time have you personally spent digging into obscure classical music or have you mostly listened to the cannon? I can say I certainly have stuck to the big names. But the big names of today weren't universally heralded in their time. In fact the reputation of art changes with the time.
It got famous after it was stolen and “returned.” I know a few art nerds who are convinced the one we know is a fake and the original is lost/destroyed.
Precisely there are some old studies linking popularity in music (as a metric for success) to blog posting about the given songs in that field/niche. It turns out that luck could be just money, media press or just a good network.
