4 ms·

>Said problem is that warehousing and distributing content scales insanely well but curation does not I don't think it's a problem any more than weather is a "

>Said problem is that warehousing and distributing content scales insanely well but curation does not I don't think it's a problem any more than weather is a "problem" that required the invention of the roof. As far as user-generated content goes, it's easy to invent the hose, but the faucet is much more complex. It's irresponsible to provide only acceleration ("reach," "virality," etc.) for information without also implementing brakes. Moderation has been a known fix for decades, but people like Zuck -- in disposition and inexperience -- won't launch anything human-powered, and moderation requires a human eye. Fuck us, right?

ummwhat 7y ago
I think you're view will change when you start putting numbers on it. I'm going to talk about the games in the app store example because it's the instance of the problem that I first thought about and it is easier to put a monetary value on an app than on a comment. Let's suppose Apple wants to make sure every game in their store gets at least 5 ratings. To do this they are going to pay people to play each game for 5 min and then assign a rating. In an hour a curator can get through 12 games. Over an 8 hour day thats 96 games. Apple would need to hire about 7 (iirc) full time employees just to give every game one rating. We wanted every game to have 5 ratings. Assuming that users on average provide one rating (but sometimes 0), we still need curators to provide the other 4. That means Apple would be employing 28 full time employees just to get their content minimally curated. Let's use the salary from the article and say each curator costs 30k a year. That means this whole department of curation is going to cost about 900k a year. Most of what they are sorting is going to be shovelware anyway. To justify the cost of this department, the curators need to find gems, content that's really good but would have otherwise been looked over. They need to find 900k of gems per year. Considering the average earning of an app, this is just barely plausible. If you want to do anything more advanced with your curation, like have the curators generate key words or figure out which demographic might like it, your budget is blown for sure. IDK how the economics of comment curation works out, but I bet it's even worse. After all, no one is even paying for comments, and making a comment is way less effort than making an app.
- BRAlNlAC 7y ago
  I don’t understand your argument at all. Is 900k/year to curate a store that’s on almost all of their products (or even one) supposed to be a lot for Apple? Seems like a rounding error to me. Not too mention that the real benefit isn’t from finding gems but from not allowing rotten content to spoil your platform. If they(FAANGs) can poach neuroscientists with 7 figure salaries like we saw yesterday, they can afford to have humans do moderation and not treat those moderators like garbage. Also they aren’t doing comment moderation, but content, mostly videos in the article it seemed. The fact of the matter is causing PTSD in your workers is an enormous (not to mention cruel) externality that Facebook absolutely needs to be held accountable for.
  - ummwhat 7y ago
    The confusion arises because I'm not at all talking about filtering rotten content (aka facebooks instance of the problem). I'm talking about sorting for good content, possibly identifying a niche hit in a stack of shovelware. On the surface these are very different tasks but they are both manifestations of what I'm calling the fundemental problem. The task of curating, that is identifying the content of incoming data to either sort or filter it, must still be done by hand and thus rarely if ever pays for itself. Sure 900k is a rounding error to Apple, but how much do they make off that 900k? And keep in mind the 900k figure was a bare minimum just to accomplish 5 ratings. The issue is that if you want to identify content as "fun", "offensive", "dramatic", "tense", "crass" or any other of the myriad of labels that haven't been machine learned, very quickly the labor to generate these labels will cost more than the value generated by these labels.
    - BRAlNlAC 7y ago
      Thanks, this makes much more sense. Typically we’ve used the market as a judge/pre-judge, but low inspiration me-too content in an opaque marketplace seems to disrupt this model significantly. It seems to me the value, especially for Apple, would be from having a better store UX. Tags that find users good stuff may not lead to Gems that you can pump to get more sales, but it might mean on-boarding a % more users per quarter to your service/ecosystem—especially over time as reputation builds. Conversely, if the App Store becomes too cluttered with garbage users may lose interest in the device or even ecosystem entirely if the quality of new apps they discover drops off dramatically due to a lack of tools/curation. An efficient curated store certainly improves the feeling of an experience versus one loaded with junk. Tagging isn’t trivial, however it can lead to a much better experience.

ummwhat 7y ago

I think you're view will change when you start putting numbers on it. I'm going to talk about the games in the app store example because it's the instance of the problem that I first thought about and it is easier to put a monetary value on an app than on a comment. Let's suppose Apple wants to make sure every game in their store gets at least 5 ratings. To do this they are going to pay people to play each game for 5 min and then assign a rating. In an hour a curator can get through 12 games. Over an 8 hour day thats 96 games. Apple would need to hire about 7 (iirc) full time employees just to give every game one rating. We wanted every game to have 5 ratings. Assuming that users on average provide one rating (but sometimes 0), we still need curators to provide the other 4. That means Apple would be employing 28 full time employees just to get their content minimally curated. Let's use the salary from the article and say each curator costs 30k a year. That means this whole department of curation is going to cost about 900k a year. Most of what they are sorting is going to be shovelware anyway. To justify the cost of this department, the curators need to find gems, content that's really good but would have otherwise been looked over. They need to find 900k of gems per year. Considering the average earning of an app, this is just barely plausible. If you want to do anything more advanced with your curation, like have the curators generate key words or figure out which demographic might like it, your budget is blown for sure. IDK how the economics of comment curation works out, but I bet it's even worse. After all, no one is even paying for comments, and making a comment is way less effort than making an app.

BRAlNlAC 7y ago
I don’t understand your argument at all. Is 900k/year to curate a store that’s on almost all of their products (or even one) supposed to be a lot for Apple? Seems like a rounding error to me. Not too mention that the real benefit isn’t from finding gems but from not allowing rotten content to spoil your platform. If they(FAANGs) can poach neuroscientists with 7 figure salaries like we saw yesterday, they can afford to have humans do moderation and not treat those moderators like garbage. Also they aren’t doing comment moderation, but content, mostly videos in the article it seemed. The fact of the matter is causing PTSD in your workers is an enormous (not to mention cruel) externality that Facebook absolutely needs to be held accountable for.
- ummwhat 7y ago
  The confusion arises because I'm not at all talking about filtering rotten content (aka facebooks instance of the problem). I'm talking about sorting for good content, possibly identifying a niche hit in a stack of shovelware. On the surface these are very different tasks but they are both manifestations of what I'm calling the fundemental problem. The task of curating, that is identifying the content of incoming data to either sort or filter it, must still be done by hand and thus rarely if ever pays for itself. Sure 900k is a rounding error to Apple, but how much do they make off that 900k? And keep in mind the 900k figure was a bare minimum just to accomplish 5 ratings. The issue is that if you want to identify content as "fun", "offensive", "dramatic", "tense", "crass" or any other of the myriad of labels that haven't been machine learned, very quickly the labor to generate these labels will cost more than the value generated by these labels.
  - BRAlNlAC 7y ago
    Thanks, this makes much more sense. Typically we’ve used the market as a judge/pre-judge, but low inspiration me-too content in an opaque marketplace seems to disrupt this model significantly. It seems to me the value, especially for Apple, would be from having a better store UX. Tags that find users good stuff may not lead to Gems that you can pump to get more sales, but it might mean on-boarding a % more users per quarter to your service/ecosystem—especially over time as reputation builds. Conversely, if the App Store becomes too cluttered with garbage users may lose interest in the device or even ecosystem entirely if the quality of new apps they discover drops off dramatically due to a lack of tools/curation. An efficient curated store certainly improves the feeling of an experience versus one loaded with junk. Tagging isn’t trivial, however it can lead to a much better experience.