|
|
|
|
|
|
by mg
808 days ago
|
|
|
Regarding whether it is "a good thing for social networks to provide such stats": It's a central part of social media, even here on HN. Look at the front page. Each post shows 6 pieces of information: Title, Domain, Points, Author, Time, Comments
2 of them = 33% are numeric indicators we trust because we trust HN.Likes and follows are probably two of the main reasons that centralized social media sites have displaced websites, rss, forums, mailing lists, newsgroups etc. |
|
|
And, hell, while "author" is the most trivial feature of all to get right as it simply can't be forged in a signature-based system, it is incredible to me how many people on Hacker News actively try to ignore who wrote a comment (not me! the first thing I look at is your name as I think it is actively asocial to dehumanize the people you are communicating with).
I honestly think you will be surprised at how usable the Hacker News feed would still be if you remove everything except the URL itself, and had your client generate a URL preview, and encourage you to try hiding some of these fields for a while using a local browser extension / script / stylesheet.
The real magic of Hacker News -- both of the feed and of the comment thread -- is the algorithmic sorting that makes sense of the cacophony of submissions, not the silly indicators (which are likely serving more as distractions than as beacons) it is bothering to surface to you: those are merely serving the purpose of making its decisions seem credible (and if you go hunting you will realize there are many posts you likely didn't see for many reasons despite having gotten many likes or even comments).
The question then is really: "how can I algorithmically rank content on a decentralized network", and I think that phrasing of the issue unlocks tons of doors that myopically insisting on a scant handful of specific statistics that are merely one of many many inputs to the current system that actually rules your usage of the website uses precludes.
Sometimes, information the user might actively be using to argue with the recommendations is even best hidden from the user: TikTok notably will happily recommend a post from years ago and hides the timestamp from you as people doing manual mental filtering incorrectly deweight the value of old content and/or find it awkward to interact with.
Now, is sorting and filtering in a decentralized system for sure solvable? I don't know... I think it is, and have tons of ideas for how to do it! Yet, I would not be shocked to try really hard and fail, or even to discover some "trivial" (in retrospect) proof that it is impossible. But, either way, one thing I am very confident of is that likes and comments isn't it, as real world systems -- including Facebook and TikTok -- manage to surface tons of interesting content to me that have low numbers of likes: the world simply isn't best sorted by thumbs up.