5 ms·
If I was Medium, I would use machine learning to curate a feed for each individual subscriber. Their personalized curated feed is already really good. I had to
by Swizec 10y ago
If I was Medium, I would use machine learning to curate a feed for each individual subscriber.
Their personalized curated feed is already really good. I had to stop looking at their weekly email because my clickthrough rate was too high and it was a time sink.
- bootload 10y ago"If I was Medium, I would use machine learning to curate a feed for each individual subscriber." If I was Medium I'd hire a team (three or four) experienced newspaper editors [0] and use their expertise to bring back some discipline back to the story process. Do you remember newspapers? There were hundreds of competing stories but the news was carefully selected on a daily basis. This is what is lacking. The machine is being gamed. Maybe the editorial teams' suggestions should be used to train some aspects of story and category selection? Machine learning is a tool and if Facebook with all it's propellor heads have a problem, what hope have Medium? A different approach is needed. reference [0] Lots of newspaper people are out of work or under utilised and as such their availability is high. For example this tweet highlighting guardian journalists getting the sack: https://twitter.com/Hadas_Gold/status/844580929226072065 https://twitter.com/Hadas_Gold/status/844580929226072065
- Swizec 10y ago> Do you remember newspapers? Honestly? No. They're that thing my dad would buy on the weekends, read one page of, and then we'd use it for potato peels, lining the floor when painting rooms, and stuff like that. By the time I was old enough to even consider reading newspapers, they had devolved into clickbait sensationalist garbage. Designed primarily to find whatever boogeyman would make each newspaper's audience most likely to buy the issue. For left newspapers the top story was always about a businessman doing something bad, for right newspapers the top story was about how this or that nationality is threatening our economy. Sometimes both sides had the top story as this or that natural disaster or war that happened so far away that it's irrelevant. At least the trashy magazines were always honest about being trashy. Their top story was this or that random celebrity doing some outrageous act that nobody cares about. Honestly, the problem is tying remuneration to readership. The other problems are emergent. Edit: the intro of this System of a Down video sums up my thoughts excellently --> https://www.youtube.com/watch?v=5vBGOrI6yBk https://www.youtube.com/watch?v=5vBGOrI6yBk
- arkitaip 10y agoWhatever your feelings about newspapers are, they have existed for hundreds of years and learnt some incredibly valuable lessons about news, journalism, reader psychology, etc. The collective wisdom of newspapers of record could fuel countless of business ventures and avoid lots of costly mistakes.
- Swizec 10y agoTotally. I just don't think I've ever experienced that. For as long as I can remember news has been about sensationalism at all cost. The question was do I personally remember newspapers and the answer is no. I remember them as a thing that exists, yes, but not as a thing that provides value.
- hkmurakami 10y agoAs one of the other commenters suggested, I encourage you to pick up a copy of the Economist.
- bshimmin 10y agoI dearly love The Economist (and, indeed, have a small stack of them about a foot from me right now), but I do sometimes get the feeling that reading a newspaper with which I agree so strongly on almost all issues (hell, even the advertising feels like it's targeted at people just like me) is probably a little unhelpful in terms of giving me a balanced perspective on the world. I'm not sure what one can do to balance this, though - I picked up a copy of Nexus (https://en.wikipedia.org/wiki/Nexus_(magazine) https://en.wikipedia.org/wiki/Nexus_(magazine)) recently and that didn't really much help either, though it was pretty damn entertaining.
- dagw 10y agoI'm not sure what one can do to balance this, though Try reading magazines like The Spectator, The National Review and The American Conservative. Even if you don't agree with everything they write (or especially if you don't agree with everything they write), they're on the whole well written and present their view in a pretty well argued manner.
- majormajor 10y agoThis might work for most people in five to ten years. Today, Spotify's algorithms couldn't figure out that I might be interested in a new album released by one of my most-listened-to bands last year. I didn't know about it until seeing it on Wikipedia, of all places, since I rely (over-rely) on Spotify for discovery. Netflix is great at telling me I might be interested in stuff I've heard of but intentionally not watched cause I'm not interested in it despite superficial similarities with stuff I do like. They seem rather useless at detecting that I like a certain type/style/"quality" (this is too pretentious, but I can't come up with a better word at the moment) of movie/show regardless of genre rather than a certain set of genre tags. And so on. What I wouldn't do to get most machine-learning curated recommendation engines with human-curated ones. Think of it as the difference between reading The Wirecutter and buying based on whatever Amazon.com's homepage pops up. Or comparing a daily digest of The New Yorker with whatever shit articles Facebook bubbles up to me.
- YPCrumble 10y agoYou're implying that it was a time sink because your clickthrough rate was too high. If reading a news source isn't worth doing once a week it probably isn't worth doing once a month, once a year, or ever. I don't think the issue is your clickthrough rate. I think the issue is that once you click through to the articles they don't contain anything of value. Your email sounds like a curated list of clickbait articles without any substance. It sounds like they've used machine learning to curate the most clickable titles (short-term gains) while ignoring quality of content (long-term retention). In a high-growth environment, for a startup that's less than five years old, that isn't surprising.