The past. At least pre-Spotify, before that particular rot essentially took the business model of piracy and added a veneer of legitimacy which eroded the already tenuous economics of recorded music into afterthoughts in the common case and fractional microtransactions in the best.
Or since copyright & the arts tend to make tech discourse stupid, we could turn our attention to the vampiric nature of food delivery and rideshares where we can also observe the tendency to not merely build platforms and work on experience but also look for places where labor/service costs can be opaquely externalized and investment can be used to subsidize anticompetitive pricing and lower wages.
Once that die was cast, of course it was inevitable that most platforms were going to go digital vampire on creators.
The other place we could go is law & policy -- really, the idea that training is "fair use" is wrong. High scale automated training can't be fair use, fair use was created well before HSAT was even conceptualized let alone possible. The law should be that every training use should require explicit use-specific opt-in, with consideration. Every other training use of a copyright claimed work should be an infringement.
Without that, there is no place to go that doesn't surrender everything to digital vampirism, and there never will be.
> Or since copyright & the arts tend to make tech discourse stupid
Oh, you can say that again. Software engineers are often amazingly small-minded, and as a software engineer, it took me quite awhile to realize that.
> The other place we could go is law & policy -- really, the idea that training is "fair use" is wrong.
Especially since that idea that "AI" training is fair use is little more than a vigorous assertion by interested parties that hope to profit from it, and rests on self-serving false equivalencies.
Or since copyright & the arts tend to make tech discourse stupid, we could turn our attention to the vampiric nature of food delivery and rideshares where we can also observe the tendency to not merely build platforms and work on experience but also look for places where labor/service costs can be opaquely externalized and investment can be used to subsidize anticompetitive pricing and lower wages.
Once that die was cast, of course it was inevitable that most platforms were going to go digital vampire on creators.
The other place we could go is law & policy -- really, the idea that training is "fair use" is wrong. High scale automated training can't be fair use, fair use was created well before HSAT was even conceptualized let alone possible. The law should be that every training use should require explicit use-specific opt-in, with consideration. Every other training use of a copyright claimed work should be an infringement.
Without that, there is no place to go that doesn't surrender everything to digital vampirism, and there never will be.