Because from what I understand, the whole idea of having individuals sort their trash is stupid, and done either beacuse of cargo-culting environmentalism, or just to make people feel like they're making some difference.
Centralized trash sorting seems like a much more efficient solution (due to the usual reasons centralized solutions are much more efficient than distributed ones). In fact, many times I've seen the segragated bins on streets being dumped into one pile on a single trash truck. Besides it being more efficient this way, it's also safer - you generally can't trust regular people not to screw up such a task, so some centralized sorting before processing is required anyway.
If segregated bins are ultimately dumped into one big mixed waste stream, it's stupid cargo culting to have people separate things. But that's not how I have seen it work in the cities I have lived in. Some centralized checking/sorting before processing is still required even if people are supposed to separate their own, but much less than if you just give people one giant bin to hold everything they want to get rid of.
If you had a centralized waste sorting machine that could separate as well as human labor, but was much cheaper, that would be even better. (Though it wouldn't completely remove the need for household level separation: some perfectly recyclable clean paper becomes un-recyclable if people indiscriminately mingle it with food scraps or other dirty things before collection.)
How is this supposed to work? Sorting trash into many containers takes me a little more time and effort and pretty much exactly as much energy as putting it all in one place. It also keeps the materials from mixing (that is, becoming dirty). How would it help to first mix it all, compress it together (which is pretty much required to keep transport efficient) and then sort it again?
Even if you want to use a sorting machine, would it not be better to have presorted ("richer", in mining lingo) input material streams?
a) can't trust that you'll do your sorting perfectly, and won't occasionally throw e.g. glass into metals bin out of hurry, carelessness, spite, or while drunk. And even if you are perfect, your neighbour isn't.
b) might want to change the segregation rules to better optimize recycling process; you don't want to force an entire city to update their bins and/or habits in lockstep with whatever the waste processor does.
Point a) means they have to sort themselves anyway, at which point individuals doing the sorting is just wasted effort.
Sure, the processor will want to seperate, say, aluminum from iron, and remove any glass from either stream, but I am fairly sure that these tasks become significantly easier if the stream does not also include diapers, orange peels and wet cardboard
My point is, they can't even trust the stream won't include those. General population can't be trusted like that.
He's a story about a New York facility that does a lot of what you ask, note that they import the machinery from Europe:
https://arstechnica.com/science/2015/01/inside-new-york-city...
As a suggestion to all the robotics / computer vision / AI geeks out there: Build an automated trash sorter with a very high degree of accuracy.
If you want the incentive, here it is. Everybody has 3 separate bins. By having 3 separate bins, not only do we force lazy people to think about how they classify their waste into those 3 separate bins, but we also end up with 3 different waste trucks driven by 3 different drivers back and forth between 3 different waste depots. And we still get it wrong, misclassifying compostable things as recyclable, recyclable as trash, etc., forcing us to pay for processing costs to sort out the 40+% of stuff that gets misclassified.
This adds up. Waste management is a $75B industry [0]. Let's break down how money is spent, and how your system will help:
1) 55% is spent on collection. Your sorter can eliminate up to 2/3 of that, but due to variant efficiency factors maybe it is only 1/2. This is about $21B in potential savings.
2) Treatment and disposal account for 20%. This includes remedial sortation (fixing the problems created by people who misclassified their waste), type-based sortation (things like separating glass from plastic, as well as separating the different types of plastic). A good automated sortation system will save at least 20% of these costs, resulting in another $3B in potential savings.
3) There are unaccounted for missed opportunities. Sortation costs force us to make compromises; instead of the more expensive sorting of thermoplastics from thermosets, which would result in higher resale value, we sometimes chop them all up and sell them as a lower valued aggregate for things like road asphalt. I have no idea how much is missed, but I do know that recycled thermoplastics have a resale value that is a good 10x higher than what they get out of that chopped aggregate.
4) Additionally, we have massive landfills that have accumulated billions of tons of trash that, due to oxygen starvation in landfills, isn't biodegrading any time soon. If your system can extract valuable recyclable materials from trash and resell them, you could meaningfully create a second-hand mining industry with basically free land rights. I have trouble seeing anywhere less than 10s of billions of dollars in annual market from this alone.
So there you have it. You can make everyone's lives easier, save us time, save us money, save the environment, and you'll have the waste management industry groveling at your feet at the chance to save them $24B a year, possibly much much more. Do it already.
[0] https://www.gridwaste.com/news/2014/8/20/10l92d52vzaw1zdsf0z...