Replies: 4 comments 6 replies
-
AI slop
One example we might wish to look at is Kagi's StopSlop, which allows users to downrank slop websites. You might also want to play SlopDetective! PI maintenance
I'd also gently suggest that most "normal" podcasters have no idea that Podcast Index exists. And that's fine. But after five years of Podcast Index, I think we should be really careful about trying to launch a creator-focused thing like "movemypodcast" - it's a lot of marketing to get its name out, and most podcasters will need it once or twice in their lives. I think we should be better at using the signals we receive from others: whether that's a 301 redirect or the Apple Podcasts API returning a different feed address - both are signals from the podcast creator that things have changed, and we should respect both of those equally. (As an aside - what worries me is that there's nowhere to discuss this stuff. Not even here, where it's been three days of silence. Not on podcastindex.social, where conversations like this fizzle out without any ending.) |
Beta Was this translation helpful? Give feedback.
5 replies
-
|
Well this doesn't happen often in my life... I agree 100% on every single thing James wrote. From top to bottom. |
Beta Was this translation helpful? Give feedback.
-
|
I can always go back and edit it |
Beta Was this translation helpful? Give feedback.
-
|
I also agree but at 99%. The 1% left it's that if AI Slop is a problem for the infrastructure, I don't have a problem with deleting it as with other SPAM feeds. As a user, just the score system it's enough (there will apps that let me hidden it). |
Beta Was this translation helpful? Give feedback.
-
|
I do hear you. However, the index could decide that a slopscore of >= 7 would a) stop the index from ever checking the feed (essentially setting it as “dead = 1”); and a slopscore of 10 would mean it would ignore podpings, too (which is, as I understand it, an easy way to bring a show in suspended animation back to life). In effect, the amount of spam wouldn’t hurt the index - it wouldn’t be checking these feeds, and the worst thing that happens is that it’s a little extra storage. Storage itself is cheap; it’s compute that costs. |
Beta Was this translation helpful? Give feedback.
-
|
"a little extra storage" - It's database storage (block storage) though, and that is expensive. Databases always grow. They never compact without offline maintenance. You're point is legit though. |
Beta Was this translation helpful? Give feedback.
-
|
If it helps visualise the issue with PI's data integrity - here are about five hours of logs, where the PodcastIndex feedURL ("PI") is different from the feed url Podnews has ("PN"), which is typically grabbed from Apple Podcasts. The size of the issue here is much bigger than I thought it was. There's particular issues with Podbean and Libsyn shows which have changed domain (for the better), and also issues with http verses https. The feedURL is currently the unique ID for each podcast - at least, that's the way we've been conditioned to think about things - and there are lots of differences here. I'm not sure that these should differ quite so much. (The "i" code at the front of these are Podnews IDs - For the second, as one example - the Podcast Index feed redirects to the "feed.podbean.com" one that Apple holds. Why isn't PI correcting the feed that it holds? (Of note - both Podbean and Libsyn are returning a 302). |
Beta Was this translation helpful? Give feedback.
1 reply
-
|
I'm actively working to try and solve the data consistency issue. But, I need to come to a mental picture of how best to do it first since Apple can be wrong too. I had a Fountain show this week where the owner had moved to Fountain but not updated Apple Podcasts, so we were right and Apple wrong. I'm afraid to start changing things just yet since it might make things even worse. But, I am at least writing code now. |
Beta Was this translation helpful? Give feedback.
-
|
I agree with scoring instead of censoring. My only concern is literal self defense against being overwhelmed. If the AI slope keeps getting worse and worse it's going to mean sizing up our infrastructure ($$$) to accomodate millions of shows nobody even knows exist and only exist for the purpose of affiliate spam, SEO linkbacks and DAI fraud. It's a hard issue, but I do agree with not blocking unless it gets to be existential. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
To leave the info more accesible. We now have: |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hi @daveajones and @agates , I think last board meeting was a lot to chew. Here are my 2 cents before I forget half of it.
AI Slop / SPAM
Hosting that ask if the show is AI, could also exposed on a tag that would be added to the reputation mix but not marked for removal with out more "points". A disclosure tag may not be the solution, but still I think could be a part of the puzzle.
Normal noise / PI maintenance
3rd directories
Grey areas
CC @nathangathright
Beta Was this translation helpful? Give feedback.
All reactions