Taste is increasingly dictated by algorithms • Hacker News Board

Re: Taste is increasingly dictated by algorithms

#51

Post by melq » Tue, Apr 17, 2018, 7:29 PM UTC

kristianc wrote: post #48

tomc1985 wrote: post #43

Earlier quoted context omitted.

The difference is that those old ways allowed for greater personal agency. Perusing old books requires selection and even a bit of curation on part of the buyer, whereas algorithmic recommendations make deciding fairly mindless. You write as if the filters you lament are bad. I think, given the overabundance of mischaracterized crap that seems to clutter up 'algorithmically' filtered exchanges (YouTube, Netflix, etc)…

> Pandora and Spotify still can't build or execute a playlist like a good DJ can. Spotify's playlists are at least in part human curated, so not sure how you got to that conclusion.

He's obviously talking about the ones that aren't, such as the 'Discover Weekly' and the way it will guess at what to play next after an album finishes.

I believe his point is that the algorithms don't generate playlists or dj mixes with the same polish as a professional human and I'm inclined to agree. That said, I do think their recommendation engine produces decent results, but I don't think the composition of the list is something it focuses on. Rather its just a list of unrelated tracks it thinks you'd be inclined to enjoy on their own.

Re: Taste is increasingly dictated by algorithms

#52

Post by dang » Tue, Apr 17, 2018, 7:31 PM UTC

This submission is rather more substantive than the titles it's had so far ("Have Algorithms Destroyed Personal Taste?", which we renamed to "Style Is an Algorithm" per https://news.ycombinator.com/newsguidelines.html), so I've taken another crack at it. If anyone suggests a better title—one that is accurate, neutral, and preferably uses representative language from the article—we can change it again.

Re: Taste is increasingly dictated by algorithms

#53

Post by andrewjl » Tue, Apr 17, 2018, 7:41 PM UTC

tomc1985 wrote: post #43

al2o3cr wrote: post #33

Shorter this article: "I liked it better when my taste was dictated by who paid the most to advertising agencies, not algorithms" In all seriousness, consider the bit listed under "how to resist the algorithm" (quotes indented): I might only read books I stumble across in used bookstores, ...which have been filtered by publishers deciding which manuscripts to publish, a past consumer deciding which book to buy, AND a…

The difference is that those old ways allowed for greater personal agency. Perusing old books requires selection and even a bit of curation on part of the buyer, whereas algorithmic recommendations make deciding fairly mindless. You write as if the filters you lament are bad. I think, given the overabundance of mischaracterized crap that seems to clutter up 'algorithmically' filtered exchanges (YouTube, Netflix, etc)…

Cuts both ways, for me domain-specific blogs and goodreads (with reviews by other people) influenced my book buying choices way more than any any algorithm and I suspect I'm not the only one.

Re: Taste is increasingly dictated by algorithms

#54

Post by tomc1985 » Tue, Apr 17, 2018, 7:42 PM UTC

melq wrote: post #51

kristianc wrote: post #48

Earlier quoted context omitted.

> Pandora and Spotify still can't build or execute a playlist like a good DJ can. Spotify's playlists are at least in part human curated, so not sure how you got to that conclusion.

He's obviously talking about the ones that aren't, such as the 'Discover Weekly' and the way it will guess at what to play next after an album finishes. I believe his point is that the algorithms don't generate playlists or dj mixes with the same polish as a professional human and I'm inclined to agree. That said, I do think their recommendation engine produces decent results, but I don't think the composition of the…

Exactly. A playlist full of tracks chosen at seeming random lacks the inter-track context that a DJ can add through his choice in selections. Auto-generated playlists have no sense of flow and transition.

You can similarly ruin a good, curated playlist by hitting shuffle.

Re: Taste is increasingly dictated by algorithms

#55

Post by neap24 » Tue, Apr 17, 2018, 7:43 PM UTC

It is pretty difficult for Netflix or Amazon to have recommendations that appeal to me because their datasets are limited to their specific domains. In reality, my tastes and interests pull from such a wide range of fields.

One year, I decided to only watch movies filmed in, or about, the state where I live. One year I caught up on famous horror movies I hadn't seen. None of these ideas was "suggested" to me directly. Nor, do they really seem related to each other. For me, at least, it seems a long way off before computer algorithms are affecting the way my brain chooses something it likes or wants to try.

Re: Taste is increasingly dictated by algorithms

#56

Post by chc » Tue, Apr 17, 2018, 7:44 PM UTC

wavefunction wrote: post #32

greggarious wrote: post #11

I'm a film buff and I think so. A lot of people don't seem to go out of their way to obtain films aside from looking at Netflix recommendations. Back when I used OkCupid I could tell when something older went on because it immediately got listed under "favorite movies" by a certain type of hipster ;) (I distinctly remember how overnight everyone looooovveeed Amelie when it went up) Sadly it's a lot harder to find new…

Presumably the people who liked Amelie would have enjoyed it both before and after they saw it, but how could they know they liked it until they knew enough about other people liking it and why those people liked it to like it themselves? I can't tell if you think all/most the people who claimed they liked Amelie are dishonestly reporting they liked it because it is important for them to like something other people l…

I think the point is not that they are being disingenuous, but that even people who are deliberately trying to like obscure things nowadays are primarily just consuming what Netflix shows them.

Re: Taste is increasingly dictated by algorithms

#57

Post by gascan » Tue, Apr 17, 2018, 7:48 PM UTC

mattkrause wrote: post #19

greggarious wrote: post #11

I'm a film buff and I think so. A lot of people don't seem to go out of their way to obtain films aside from looking at Netflix recommendations. Back when I used OkCupid I could tell when something older went on because it immediately got listed under "favorite movies" by a certain type of hipster ;) (I distinctly remember how overnight everyone looooovveeed Amelie when it went up) Sadly it's a lot harder to find new…

Do you think people actually like—or are claiming to like—a movie just because Netflix recommends it? To me, it seems more likely that this just reflects increase availability: more people suddenly have access to a movie, so more watch it and some of those will like it. I can see how people, especially on a dating site, would feel pressured to mention something highbrow, but the social pressure itself isn’t coming fr…

They weren't pressured to like it, but without the "algorithm" showing it to them, they never would have discovered it.

Re: Taste is increasingly dictated by algorithms

#58

Post by Asdfbla » Tue, Apr 17, 2018, 8:30 PM UTC

On the topic of algorithms helping you discover your own 'taste' (whatever that may be), some time ago when the topic came up on HN someone linked this interesting paper: "Recommender Systems for Self-Actualization" [1], which suggests some methods how filter bubbles could be avoided. I really liked it and its suggestions like having extra recommendation lists for things the user will likely hate, items that are very polarizing in the user base or things no one ever rated before.

[1] https://sci-hub.tw/https://dl.acm.org/citation.cfm?id=295918...

Re: Taste is increasingly dictated by algorithms

#59

Post by blattimwind » Tue, Apr 17, 2018, 9:04 PM UTC

bitwize wrote: post #44

firasd wrote: post #9

Really interesting article. Weird thing is that in the social media era, humans ourselves seem behave in algorithmic ways. My Twitter feed is full of Beyonce worship every few months even though I've been mostly-indifferent to her music for 20 years. But when people come together like a hive and retweet how they excited they are about some shared interest, there doesn't need to be a recommendation engine pushing a sp…

When I was in Japan in 2011, nearly every girl I talked to had the same favorite singer -- Lady Gaga. I noticed that Japanese pop culture diversifies in time rather than space: something reaches critical mass and then sweeps the entire nation like cherry blossom season, only to be swept away by the next big thing weeks later. Also trending at that time were the anime One Piece and the Disney character Stitch (who had…

> Japanese pop culture diversifies in time rather than space

That's not diversity, it's simply pop culture.

Re: Taste is increasingly dictated by algorithms

#60

Post by c22 » Tue, Apr 17, 2018, 9:06 PM UTC

greggarious wrote: post #11

I'm a film buff and I think so. A lot of people don't seem to go out of their way to obtain films aside from looking at Netflix recommendations. Back when I used OkCupid I could tell when something older went on because it immediately got listed under "favorite movies" by a certain type of hipster ;) (I distinctly remember how overnight everyone looooovveeed Amelie when it went up) Sadly it's a lot harder to find new…

This is obviously useless for almost everyone, likely including you, but I urge anyone in or near Seattle who has this problem to check out Scarecrow Video[0] which has over 100,000 films in a variety of formats organized by director.

[0] https://en.wikipedia.org/wiki/Scarecrow_Video