How spammers are gaming the podcast charts | Hacker News Reader

How spammers are gaming the podcast charts

chartable.com

93 pointsvoodooranger31 comments

A podcaster (and Youtuber) who I really enjoy is CGP Grey. He has been proposing allowing voluntary removal of view numbers from Youtube and similar platforms.

I think Apple should fix or remove their rankings but, please, don't add any more metrics.

The podcast world is currently mostly immune from so much of the clickbaity trash on Youtube (that arguably isn't even YT's fault).

I admit it's very counterintuitive or almost subversive these days to suggest that internet points be kept secret.

Edit: Here's a CGP Grey video where he talks indirectly about how he sees the internet fracturing his (our?) ability to pay attention and focus. https://youtu.be/wf2VxeIm1no

> I admit it's very counterintuitive or almost subversive these days to suggest that internet points be kept secret.

Media popularity rankings are toxic for everyone but distribution middlemen and advertisers. They turn what should be heterogeneous markets for content producers and their audiences in many different niches (geographic area, interests, subcultures, etc), into a global winner-take-all popularity race in a single market (owned by the distribution middlemen, like iTunes, Amazon, Google Play, etc). The kind of market where every seller is ranked by a single metric and only the top few are rewarded makes sense for things like professional sports, but very little else.

Which of the two groups of people are benefitting from Hacker News showing metrics? Stories show points and total number of comments. Anyone can see anyone’s total karma. And you personally can see your comment karma.

YouTube and and other social media would still get advertising and such without public metrics since you could give private access to them when trying to make deals.

So overall I don’t see how either groups of people mentioned are benefitting much from HN showing metrics.

> Which of the two groups of people are benefitting from Hacker News showing metrics?

That is a bit of a nonsense question. Neither benefit, because there is no advertising on HN, and HN is not a media distribution business. HN is an online forum, and the original point of post ratings on forums was crowdsourced moderation. The point of online forum moderation is to remove spam, trolling, kooks, and other things that waste your reading time. So moderation is a strictly negative system. Using post ratings for online reputation is what turns it into a positive system.

IMO online reputation based on upvotes is a game with no winners, only losers. Problems include gamification and addiction, groupthink, privacy risks (it is very easy to correlate bits of information to deanonymize pseudonyms, the persistent use of which the point system encourages). What's the benefit? It is much more valuable to focus on the moderation aspects. The problem with online communities today is keeping out spam, trolls, kooks, idiots, and other bad actors. Modding down inane and ignorant postings so that they don't waste people's time is much more valuable in keeping forums interesting and relevant than another 10 upvotes. "Harmless" but useless and time-wasting posts will drive participants away and erode the community. That is the one of the lessons of Eternal September.

> I admit it's very counterintuitive or almost subversive these days to suggest that internet points be kept secret.

Interesting point. I'm not sure if you already know this, but Hacker News used to display points for each comment previously. This was turned off in...I dunno, early 2011?

I don't think it's harmed user activity on the site at all!

I did not know that but I'm glad they did. Displaying "internet points" causes herd mentality to propagate and removes critical thinking or at least diminishes one's capacity to genuinely listen to one's own feeling. The same goes beyond intellectual content; ditto for creative content (e.g. YouTube videos) etc.

I liked what he was putting down in this (even if it does mean he's on a HN fast!) when I first saw it

I've used uBlock Origin to block the score-related elements in Reddit as well as my own comment scores here. I don't think I pay much attention to the numbers in YouTube etc but will do so there too if I notice I'm checking them

  www.reddit.com##.score-hidden, .score
  news.ycombinator.com##.score

I think this article should be retitled to remove the word "How" at the beginning, since it doesn't even really speculate how one would go about doing this. It's interesting nonetheless.

Any ideas? Assume that someone decided to harness one of the many click-farms to do automated podcast subscriptions?

I suppose it's also possible that Apple did an ill-advised algorithm change or something that changed the weights very heavily, and the spammers who had been quietly doing their thing for months suddenly all got promoted.

We can't say for sure how the charts are being gamed, but there is strong evidence that its related to a pay per play scheme to pump subscription numbers in a small time window. The implication of the graph visualization is that the same iTunes users are subscribing to all of the suspect shows, and all on one storefront (the U.S.).

I'm reminded of the images from China where one person is in front of rows and rows of phones clicking on stuff and that is their job. Would that do the job here?

Apple doesn't publicly comment on how the chart rankings are calculated, which is why we can't draw strong conclusions, but our understanding is that this would do the job. Since the somewhat catastrophic failure of the charts a week ago its clear that Apple did update their algorithms, but its not clear how (and some of the same players identified in the blog are still ranking highly).

It's going to be just like record charts. People are looking for those popularity lists to find the interesting and good stuff that other people are listening to. We'll have scandals, companies will be pushing up their own ratings via nefarious means. Then you get more viewers, more money from advertisers, your network benefits - there must be podcast celebrities who get the equivalent of payola to say they like another podcast. History repeats itself.

The algorithm seems to be weighted to boost new subscriptions - looking at the Chartable top US-all there is one at #20 which recently launched (has 6 episodes) which is above This American Life at #28

While Apple doesn't comment on how the rankings are calculated, the common knowledge is indeed that its heavily biased towards recent subscriptions.

Maybe it should be more weighted towards regular releases over a longer period of time... points for releasing on average of once a day, or once a week for over 2 years after hitting X subscribers. I can't speak for anyone else, but those are the podcasts that I would be most inclined to want to hear/see.

I have never used a podcast top X chart before. I usually think, hmmm, I wonder if there's a podcast about <some subject> and search the internet for that.

I guess I care more about a subject than popularity.

Startup idea for anyone so inclined: Create a podcast distribution service where podcasters pay the current cost of a postage stamp per listener to push podcasts onto devices. If the recipient listens to the podcast, the money is refunded to the podcaster. If they delete it or ignore it for 30 days, they keep the money. The podcasters are allowed to see anonymized metrics for who listens to what. Listeners can use their earnings to tip their favorite podcasters, or just cash out.

Sweet! I can now sign up to as many podcasts as I want, never listen to anything, and make money.

And so can everyone else.

And podcasters would love to use this distribution system that will net them $0 after risking capital, ONLY if they magically achieve a 100% listen rate in spite of the incredible incentives to have the listen rate be 0%. Otherwise it’ll cost them money.

And for all that hassle and risk, they get a distribution system no better than the free and open one they enjoy now.

Perhaps I’ve misunderstood something, otherwise I don’t think this idea is viable :)

You don't have to sign up for any podcasts. You just sign up and start receiving recommendations. At first, recommendations are free or virtually free for the podcaster, since there is no data about your interests. You start accepting the ones that interest you and rejecting the ones that don't.

Podcasters get access to the entire data set for every listener and can run clustering and regression to their hearts' content. As you listen to podcasts and perhaps rate them, the cost to target you (i.e. your random ID) goes up because the podcasters have better data to work with and should know better than to send you politics podcasts when all you listen to is vampire erotica.

If you become a listener, then they've acquired a listener and you've acquired a podcast that you like. If you're not interested, then they pay the cost of poor targeting, and you are compensated for the inconvenience. Proper incentives all around!

The net cost to the podcaster doesn't have to be $0 for the service to be worthwhile. It only has to be more effective than AdWords and other venues where they pay to expose themselves.

Sorry, this would never work. Maybe if Apple launched this so that you had already overcome chicken and the egg. But I’d never ever sign up for this as a podcaster (or any kind of audience builder) unless there was literally no other choice because that’s where ALL the listeners are. Even then, I’d probably just do something else rather than allow some sleazy middle-man to screw me over while trying to connect with my fans.

You’re totally missing the point that podcasts are pull, not push. I listen to dozens of them, none of them have ever “spammed” me. That literally makes no sense. I subscribed to them. There’s no way for them to push their podcast on someone who doesn’t want it.

It's not a chicken-and-egg problem. Clearly, building the audience is the harder problem and the priority. When you have an audience, podcasters pay to reach that audience, which is usually done by pestering people with web ads until they install an ad blocker or by hiring sleazy "PR consultants" to spam the ratings and reviews and social media. This would just cut out those middlemen and compensate the audience directly whenever the messaging misses the mark.

No users will sign up for a podcast app / network that has no podcasters on it. And no podcasters are going to pay to reach a non-existent userbase.

You're in such a rush to shoot down the idea that you haven't taken 3 seconds to realize how trivial it is to create a podcast site/app with thousands of podcasts but no users. Users are the chicken and the egg, just as they are for the podcasters who spend more time and money on audience-building than on podcasting.

Couldn't we just skip the technology part and have podcasters send money directly to scammers in exchange for the scammers not listening to their podcasts?

"Scammer" here is someone who signs up and then never listens to anything? Don't send them anything. Scammer and spammer problems solved.

How do you verify if someone actually listened to a podcast vs. someone who has 100+ accounts, a wall of 100 mobile phones constantly 'listening' to 100 podcasts simultaneously to game the system?

That would be a really counterproductive way to game the system. :)

So you're paying people to not listen to podcasts?

That seems like the mother of all perverse incentives.

Sure, kind of like direct mail marketing except the victim gets the money (minus a bit) instead of the post office. It wouldn't be very profitable because non-activity would show up in the metrics and podcasters would not distribute to them--unless the podcaster is just indiscriminately spamming everyone, in which case, they're just paying people to ignore them until they go broke.

or: How to Game to Podcast Chart

There is no “how” in the article other than “click farms”. It should be “Spammers are gaming podcast charts”.

Reply on news.ycombinator.com