As a developer of a feed reader that supports both RSS/Atom and Twitter/Facebook/Google+/Instagram/VK (https://bazqux.com) I say that making client for well defined API hosted by one provider (silo) was much more simple than support RSS/Atom feeds.
It requires a lot of work to make a feed reader that supports all possible feed generators from all possible sites. And it's more related to multiple buggy producers than to multiple feed formats (for example repeating item IDs, different item IDs for the same items, missing IDs -- no matter what format you use you get same issues).
Another complexity is feed updating itself. Silo hosts content and knows what was updated. Feed readers don't. So they need to poll and update a lot of content even if they have just a few users (but with thousand feeds each). There is a PubSubHubbub, but again, lots of providers and implementations could be buggy (for most feeds I just fetch them directly when receiving a push instead of use push data itself since it's more reliable).
So the open nature of web syndication itself dampens its usage by making developing and running feed reader a costly task. It's really simpler (and maybe more profitable if you're lucky) to write a silo.
And there are more problems that are not related to feed format at all: relative complexity of subscribing, not everybody need syndication (many people prefer occasional visiting of the site instead of reading each article), new forms of "syndication" (related videos, algorithmic feeds), less blogs (maybe it's good -- let people post cat pics in social media), less popularity (not a part of big sites, browsers), less education (social media and messaging apps is the only Internet some people know) and so on.
But there are people who want to follow every article from interesting sources in software optimized for this task. And feed readers are the best thing here. For example, some people use my feed reader for Facebook/Twitter feeds only (no RSS feeds at all) just because it's more convenient to read them here.
In the end, I think, process of reading subscriptions is more important than underlying format. If RSS will die (quite unlikely) another methods of getting articles will appear.
Here is an incomplete list of various problems you'll have if you want to write a feed reader http://inessential.com/2013/03/18/brians_stupid_feed_tricks
It requires a lot of work to make a feed reader that supports all possible feed generators from all possible sites. And it's more related to multiple buggy producers than to multiple feed formats (for example repeating item IDs, different item IDs for the same items, missing IDs -- no matter what format you use you get same issues).
Another complexity is feed updating itself. Silo hosts content and knows what was updated. Feed readers don't. So they need to poll and update a lot of content even if they have just a few users (but with thousand feeds each). There is a PubSubHubbub, but again, lots of providers and implementations could be buggy (for most feeds I just fetch them directly when receiving a push instead of use push data itself since it's more reliable).
So the open nature of web syndication itself dampens its usage by making developing and running feed reader a costly task. It's really simpler (and maybe more profitable if you're lucky) to write a silo.
And there are more problems that are not related to feed format at all: relative complexity of subscribing, not everybody need syndication (many people prefer occasional visiting of the site instead of reading each article), new forms of "syndication" (related videos, algorithmic feeds), less blogs (maybe it's good -- let people post cat pics in social media), less popularity (not a part of big sites, browsers), less education (social media and messaging apps is the only Internet some people know) and so on.
But there are people who want to follow every article from interesting sources in software optimized for this task. And feed readers are the best thing here. For example, some people use my feed reader for Facebook/Twitter feeds only (no RSS feeds at all) just because it's more convenient to read them here.
In the end, I think, process of reading subscriptions is more important than underlying format. If RSS will die (quite unlikely) another methods of getting articles will appear.