3 ms·

Maybe. I don't know that there's an objective way for us to convince each other. Subjectively, I just can't see "streaming media player that is good enough" as

Maybe. I don't know that there's an objective way for us to convince each other. Subjectively, I just can't see "streaming media player that is good enough" as a long term differentiator. Seems like too limited a domain. I'm not dismissing "good software is hard," but this isn't iphone vs android. It's more comparable to a single app. Recommendation algorithms... I honestly think this is overemphasized. It's selecting from a pretty shallow content pool. This isn't TikTok/Youtube/Instagram/spotify etc. Discovery just isn't that big a problem when there are circa 2000 options. Online streaming is a content game, ultimately. You can't win on tech, I think.

jjk166 6y ago
There are only a few thousand options on the streaming network, you are correct in pointing out that picking a few recommendations for the user from this list is not differentiating. For picking what content should be created and added to the streaming network, there are infinite possibilities. Selecting correctly there is very difficult to do reliably. We've all watched movies that made us ask "what the hell were the producers thinking?" Even experienced professionals can do little more than guess what audiences are looking for. When you wonder why a weak movie has 7 sequels, it's because re-using what worked in the past is a very simple risk reduction heuristic that humans are good at applying. With little to go off of beyond past ticket sales, a few test screenings, and personal anecdote, there's little a human can do to improve upon this. You get the occasional genius who has some fundamental understanding of what people like and they regularly make hits, but these people are few and far between, and still they are generally limited to a handful of genres they understand and enjoy enough to work with. If you had pitched to me an idea for a documentary about a gay redneck roadside zookeeping convicted felon, I would find the idea that any significant amount of people would watch it laughable. No sane person would think of giving this top billing. But Netflix's algorithm spit out that for some arcane reason this would be a hit, and alas now everyone and their mother has watched Tiger King. Such is the power of analyzing tens of billions of viewing experiences in exceptional detail. Moving beyond this, there is a widely held belief that the key to a strong streaming service is having the best content. Certainly if you are choosing between two movies at the theater you're going to watch the one you think you'll like more. Ditto for if you're flipping through the channels to see which show you want to watch. But streaming isn't the same as going to the movies or flipping through the channels. In those circumstances you have to choose the best of a small number of options that are offered right now; with streaming you watch what you want when you want. This leads to fundamentally different viewing habits. People will binge a season of TV in a weekend, or throw on a movie they've seen 15 times to have on in the background, or will watch a new movie every night for a month, and so forth. Further, the economics of streaming are different. Movies are about getting as many people as possible to pay the price of admission, television is about getting as many people as possible to watch ads, streaming is about maximizing subscription-months. It doesn't matter if 99.9% of a service's offering is crap, if they have 12 releases a year that you want to watch, and they space them out appropriately, you'll remain subscribed. On the other hand, a much anticipated blockbuster that has something for everyone but doesn't have much re-watch value isn't going to get anyone past the free trial period. Success in streaming is all about having the right combination of content being released on the right schedule so that as few users as possible go more than a few weeks without wanting to utilize the platform. When you have millions of users, this is an incredibly difficult instance of combinatorial optimization. This is why, for example, Amazon entered the space and succeeded with ease despite no experience with content creation: online streaming is not a content game, it's a big data game. Ultimately, all such innovation implosions are a result of a mismatch between the technical challenges perceived by those looking to get into a market and the real technical challenges that actually face those in the market. The old institutions definitely have the resources necessary to reach technical parity, but they squander those resources solving the problems they think are important, not the ones that actually are. At the end of the day, you don't really know what it takes to win until you've won, and even then you may have just gotten lucky somewhere along the way. Success requires the humility to question the validity of our experience.

jjk166 6y ago

There are only a few thousand options on the streaming network, you are correct in pointing out that picking a few recommendations for the user from this list is not differentiating. For picking what content should be created and added to the streaming network, there are infinite possibilities. Selecting correctly there is very difficult to do reliably. We've all watched movies that made us ask "what the hell were the producers thinking?" Even experienced professionals can do little more than guess what audiences are looking for. When you wonder why a weak movie has 7 sequels, it's because re-using what worked in the past is a very simple risk reduction heuristic that humans are good at applying. With little to go off of beyond past ticket sales, a few test screenings, and personal anecdote, there's little a human can do to improve upon this. You get the occasional genius who has some fundamental understanding of what people like and they regularly make hits, but these people are few and far between, and still they are generally limited to a handful of genres they understand and enjoy enough to work with. If you had pitched to me an idea for a documentary about a gay redneck roadside zookeeping convicted felon, I would find the idea that any significant amount of people would watch it laughable. No sane person would think of giving this top billing. But Netflix's algorithm spit out that for some arcane reason this would be a hit, and alas now everyone and their mother has watched Tiger King. Such is the power of analyzing tens of billions of viewing experiences in exceptional detail. Moving beyond this, there is a widely held belief that the key to a strong streaming service is having the best content. Certainly if you are choosing between two movies at the theater you're going to watch the one you think you'll like more. Ditto for if you're flipping through the channels to see which show you want to watch. But streaming isn't the same as going to the movies or flipping through the channels. In those circumstances you have to choose the best of a small number of options that are offered right now; with streaming you watch what you want when you want. This leads to fundamentally different viewing habits. People will binge a season of TV in a weekend, or throw on a movie they've seen 15 times to have on in the background, or will watch a new movie every night for a month, and so forth. Further, the economics of streaming are different. Movies are about getting as many people as possible to pay the price of admission, television is about getting as many people as possible to watch ads, streaming is about maximizing subscription-months. It doesn't matter if 99.9% of a service's offering is crap, if they have 12 releases a year that you want to watch, and they space them out appropriately, you'll remain subscribed. On the other hand, a much anticipated blockbuster that has something for everyone but doesn't have much re-watch value isn't going to get anyone past the free trial period. Success in streaming is all about having the right combination of content being released on the right schedule so that as few users as possible go more than a few weeks without wanting to utilize the platform. When you have millions of users, this is an incredibly difficult instance of combinatorial optimization. This is why, for example, Amazon entered the space and succeeded with ease despite no experience with content creation: online streaming is not a content game, it's a big data game. Ultimately, all such innovation implosions are a result of a mismatch between the technical challenges perceived by those looking to get into a market and the real technical challenges that actually face those in the market. The old institutions definitely have the resources necessary to reach technical parity, but they squander those resources solving the problems they think are important, not the ones that actually are. At the end of the day, you don't really know what it takes to win until you've won, and even then you may have just gotten lucky somewhere along the way. Success requires the humility to question the validity of our experience.