Great questions
Hive might not have the most content compared to WEB2 services, but that doesn't make it statistically easier to find the content one is looking for. If 1 in 100 posts are really good, it will take 100 tries to find a good post no matter what the total number of posts is. An algorithm can act as a filter that makes it easier to find those gems. Good posts still get buried by shitposts here simply due to the sortation of content by payout.
The Steem whitepaper is a joke and an utter failure for a lot of reasons.
Could write an entire post on that and then some.
It is not a selling point that the network lacks basic tools.
In fact an algorithm isn't a selling point at all for any blockchain, because an algorithm is centralized in nature no matter what. It does not empower the blockchain, but rather the frontends connected to the blockchain, which are all a single server and single controlling entity. An algo on Peakd would not be the same as an algo on Leo or Ecency or Liketu or Hive.Blog. The entire point of the algo is to differentiate different front ends to fill a certain niche. The initial code for them could be open source and cloned, but they'd need to be tweaked a bit for each frontend to get the desired result.
Ultimately an algo for WEB3 would have completely different incentives than WEB2. WEB2 algos exist to hook the user and farm data, whereas a WEB3 algo could have much more pure intentions and generate value solely for the users.
True, it probably made sense back in 2016, when it was released, but latest after the hype of 2017, it was clear that some of the assumptions in there do not hold. It would have been good, if someone went back and proposed an updated version, but @dan was already gone and STINC was already lacking in vision and execution.
Centralized or decentralized is not an issue, as long as you have consensus on the data you're going to use (which is exactly what the blockchain does). The idea of different algorithms is quite interesting and you could even further push it, by giving each user an individual algorithm, running locally on the user's device. That would give a similar or even better experience than centralized social media platforms, but that is technically quite challenging. So different algorithms for different front-ends seem more likely.
Pioneering choice of algorithm would indeed be great indeed a great feature for Web3 social media. However, most users would not care and therefore it would only be interesting for a few users.