Welcome! Log In Create A New Profile

Advanced

Algorhythms

Algorhythms
December 04, 2015 12:34PM
Have any of you had any success with the internet's RIYL (Recommend If You Like) systems, in which computer algorithms predict your tastes?

I don't like someone just because they sound like someone else I like. I love Kate Bush, and can't stand the terribly precious ladies of the Tori Amos ilk that get recommended to me. I don't like Bush cuz she's an artsy piano-playing woman. What I like is her experimental approach to pop songs (a la pop-era Eno, her sometimes collaborator Peter Gabriel, etc.) And her wondrous voice. Amos has none of those qualities, for me at least.

And on the other hand, I really like Zappa-influenced and related artists, but generally can't stand the man himself, what with his overly fussy music and soulless misanthropy. But Beefheart, and various post-Zappa obscure weirdos like Zoogz Rift and Rudy Schwartz Project make me happy.

Why are they doing this, what is the point of these systems, are they actually successful? How can a computer predict what I will like?! Like how Pandora keeps playing the Misfits on my Cramps channel. Blech. I've said it before, I'll say it again: Danzig sings like Cher.



Post Edited (12-04-15 14:36)
Re: Algorhythms
December 04, 2015 12:50PM
I've found the "related artists" list on Spotify to be pretty useful, especially on genres I had little access to during their heydays.

For instance, I pulled up The Field Mice a year or two ago, and there were a plethora of minor C86 types in their related artists listing that I was unfamiliar with at the time, like Another Sunny Day, The Hit Parade and 14 Iced Bears, that I've come to enjoy quite a bit, and those led me to many more minor UK 80s bands, many of whom turned up as a Nugget of the Day at one point or another.

The problem with someone like Kate Bush is that her followers are a lot easier to categorize than her peers. Looking at what Spotify lists as her related artists, you get Amos and Sinead, but also Marianne Faithfull, Gabriel, Roxy Music, Prefab Sprout, Bowie, David Sylvian, Tanita Tikaram, Talk Talk, Bjork and The Pretenders, who I have to think are there only by virtue of having a female vocalist. And Sting...

Some of those names are useful comparisons, some not. And generally speaking, they aren't going to help the curious Kate Bush fan discover anything they're not already familiar with.
Re: Algorhythms
December 04, 2015 01:03PM
Hmm, I'm not on Spotify, but Pandora was sometimes useful if I started with someone fairly obscure to begin with. Sounds like what you did with Spotify. Pandora actually starts with humans coding the songs before the data gets fed into HAL, so that probably helps.

I don't believe that Amazon, however, has ever suggested a single useful thing to me.



Post Edited (12-04-15 16:41)
Re: Algorhythms
December 07, 2015 03:55PM
Discovered the odd cool thing via Pandora but not necessarily obvious in terms of the initial entry so I guess that's a lottery-like situation. I've had Amazon recommend some cool books but not music.
Re: Algorhythms
December 04, 2015 01:39PM
Those algorithms probably are aimed more at the "average" listener than at the true music mensch.

When CMJ New Music Monthly was in publication, each of its album reviews included an RIYL line. That led me to some cool artists, but in retrospect, I probably overlooked some potentially great ones whose music didn't apparently dovetail with what I already knew. So I guess I did myself a disservice, following that line too closely.

Re: Algorhythms
December 05, 2015 06:09AM
I dunno. The Amazon system has more often than not made me aware of an obscuro classic that I had no idea was finally available on CD based on what I was looking at. But that's not really introducing new things to me, really. I can't speak for any other systems because I don't get out much.



Former TP subscriber [81, 82, 83, 84]

[postpunkmonk.com]
For further rumination on the Fresh New Sound of Yesterday®
BCE
Re: Algorhythms
December 05, 2015 08:35AM
"Have any of you had any success with the internet's RIYL (Recommend If You Like) systems, in which computer algorithms predict your tastes?"

No, and doesn't this very concept (the genesis of Pandora) contradict the idea of discovering new music? If I tell Pandora I like Social Distortion and Velvet Underground and The Pogues, I don't want to hear bands that sound like Social Distortion and Velvet Underground and The Pogues -- and, yet, that's exactly what seems to happen. The "right of this dial" alt-radio station here is all computerized (after a flip to alt from AOR) and incredibly, incredibly boring and predictable -- due in large part to those stupid algorhythms and music directors who look to them for the best rotations.
Re: Algorhythms
December 05, 2015 10:04AM
Amazon's RIYL has thrown up some stuff I've gotten into (ditto Forced Exposure), though it can start to become predictable in what it shows ("this recommendation again?"). Same goes for Youtube and Discogs. But I can't say I've ever had any luck with Pandora.
Sorry, only registered users may post in this forum.

Click here to login