> On a related note, why is Netflix's interface so locked down?
A/B testing. The product org has always been driven by A/B testing, which means it sometimes finds a local maxima which they then stick to. They've also been driven by keeping things simple, both for the user and the developers. Having a single interface means the user gets a consistent experience (ironically broken by having so many A/B tests) and also means that testing on the backend is easier, because you don't have a geometric explosion of combinations of settings to test.
The fewer settings there are the fewer things to test when you want to change something.
> Why would it be bad for me to turn off autoplay for trailers?
Because A/B testing has shown that overall it's better for customer retention. They're willing to give up some users in exchange for the gains they get from having it on.
> Why can't I say "never show me this show again, I'm never going to watch it"?
Because people lie to themselves. A lot of people will say, "Schindler's List is an amazing movie!" and keep it on their watch list, but yet never watch it again, and then will say "Jackass 3 is the dumbest shit I've ever seen" and never put it on their list but watch it 15 times. People's intentions and actions don't always match. Netflix used to have a way to remove stuff from being shown, and then get calls into customer service from people who had chosen that button asking how to get the movie back because they actually want to watch it.
> And why, in this age of AI, and they still putting white text over a white background?
While you could fix this with AI, it would take a heck of a lot of computer power for a relatively small fix for few people (not a lot of people use captions). You'd have to review every film with every set of captions. That's a lot of hours of AI review.
> That last one, especially, made me say "do people at Netflix eat their own dogfood"?
When I worked there, most everyone watched Netflix every day. In fact, the only thing we didn't dogfood was the billing system because we all had free Netflix, and then when it broke, they gave everyone an $8/mo raise and told us all to sign up for paid accounts so that we could test the billing system with different billing days and different payment methods.
But very few that I knew of who used captioning. Most people don't like it and find it distracting. So when there were captioning errors they were usually caught by customers.
A/B testing. The product org has always been driven by A/B testing, which means it sometimes finds a local maxima which they then stick to. They've also been driven by keeping things simple, both for the user and the developers. Having a single interface means the user gets a consistent experience (ironically broken by having so many A/B tests) and also means that testing on the backend is easier, because you don't have a geometric explosion of combinations of settings to test.
The fewer settings there are the fewer things to test when you want to change something.
> Why would it be bad for me to turn off autoplay for trailers?
Because A/B testing has shown that overall it's better for customer retention. They're willing to give up some users in exchange for the gains they get from having it on.
> Why can't I say "never show me this show again, I'm never going to watch it"?
Because people lie to themselves. A lot of people will say, "Schindler's List is an amazing movie!" and keep it on their watch list, but yet never watch it again, and then will say "Jackass 3 is the dumbest shit I've ever seen" and never put it on their list but watch it 15 times. People's intentions and actions don't always match. Netflix used to have a way to remove stuff from being shown, and then get calls into customer service from people who had chosen that button asking how to get the movie back because they actually want to watch it.
> And why, in this age of AI, and they still putting white text over a white background?
While you could fix this with AI, it would take a heck of a lot of computer power for a relatively small fix for few people (not a lot of people use captions). You'd have to review every film with every set of captions. That's a lot of hours of AI review.
> That last one, especially, made me say "do people at Netflix eat their own dogfood"?
When I worked there, most everyone watched Netflix every day. In fact, the only thing we didn't dogfood was the billing system because we all had free Netflix, and then when it broke, they gave everyone an $8/mo raise and told us all to sign up for paid accounts so that we could test the billing system with different billing days and different payment methods.
But very few that I knew of who used captioning. Most people don't like it and find it distracting. So when there were captioning errors they were usually caught by customers.