I eventually went back to RSS, picked feeds from twenty or so minor news sources focused on my interests, and in general I'm much more satisfied. I eventually started using Pocket as well to retain a small trickle of "best of", which has the best ratio of interesting headlines (to me) I've encountered in an aggregator.
It works because it services majority tastes specifically at the expense of minority ones.
Put another way, imagine you’re a PM trying to make a chart go up and to the right. Here’s a magic algorithm that gives you best users more of what they want, and ones for whom the product doesn’t work no reason to stick around so they leave.
An increasing numerator with a decreasing denominator. That’s a hyperbolic upwards curve! Promote that PM! You let the prisoners run the asylum in this case, choosing which metrics they will be evaluated on, that totally runs counter to the whole point of building platforms.
As long as your inbound user count exceeds churn due to booting out people with minority tastes—which a PM EXTREMELY RARELY had anything to do with—the PM’s career takes off.
No dude, discoverability as it is defined by every tech company is killing culture. And it’s Product Management that’s holding the knife.
Everybody has some minority tastes. Everyone's got their own long tail of stuff they like. Obviously there's stuff that's more popular than other stuff - otherwise they wouldn't be majority tastes - but it's not like someone who watches Jimmy Kimmel on Youtube therefore doesn't also enjoy, like, woodworking channels.
The difference is, I guess, what you're willing to settle for - if the algorithm discovers you like a thing that's easy to find more of, it will really start shoving that in your face, because it's easy, and if you're not persistent enough to keep seeking out specialty content despite the tide, then the popular stuff is what you're going to get.
I understand, but it’s more than that.
I mean pretty soon, if you ask a kid to describe what a movie is, they might say a long video with eight to ten superheroes in it.
All this sameness is stunting our imagination.
The "sameness" makes the technical burden easier for anyone who develops a search engine. If more searches had to be treated as "unique" this would require more resources and more time to complete. Sameness allows results to be returned faster.
Thus Google and others providing "search" are invcentivized to facilitate what you call the "cultural selection" process, i.e. selection by popularity. This article reflects a common view that this process is not adequate for all type of "content".
In the early days of web search engines, sites like AltaVista or later Google, would include a line on the results page boasting about the number of returned results and the number of pages searched, drawing attention to the size of their database and presumably the variety it would contain.
The irony of this is that we see now that Google, and any other who followed their lead, does not want nor expect people rapidly digging deep into search results. The majority of queries they receive are repeat queries. They would prefer and they expect that users be content with a "top search result", never moving beyond "page one". This not only lessens the technical burden, it can be presented as "what people want", they can portray this as the optimal "experience" for the user. Whereas someone who keeps digging and digging for more would draw suspicion and be blocked. Genuine curiousity, often a hallmark of intelligence, is not accounted for.
The system is geared toward delivering the sameness that purportedly all (e.g., 80% of) users want, as efficiently as possible. They offer no option (for the remaining 20%) to sort all results alphabetically, by date, or some other field or to download a list of all results to do this offline. Even sorting by "relevance" is a non-transparent process. It is not a disclosed algorithm and we are not allowed to do our own relevance filtering on raw results. What we get has already been pre-filtered. Advertisers pay the bills.
If the majority of users never reach page two, then an ad at the top of page one can command a higher price. The order of results is all-important. The user is not allowed to control the order.
"Everything is on the web" and yet doing "real" research -- as, e.g., one would using a databse in a library - is, 30 years on, still nigh impossible. Advertising is the problem, as the Google founders published in their search engine paper announcing Google in the 1990's. Alas, they later found advertising was the solution... to commercializing what was originally meant to be academic.