[thanatropism]

Netflix has about as many original great shows as I remember HBO having 10-15 years ago: two to three a year. They just produce A LOT of stuff, and for people with a habit to consume TV four times a week, it runs out.

They have the advantage of being able to use consumer viewing patterns to optimize their new shows, but then the seams are too often visible. Often the shows with better plots suffer from bad acting too. The best of their shows are the ones they seem to have bought as an entire proposition, like "1983" and "Suburra".

[mlthoughts2018]

> “They have the advantage of being able to use consumer viewing patterns to optimize their new show”

From a cultural or art critical point of view, this is a negative though. Movies end up being made to appeal to lowest common denominator tastes or themes for a broad audience.

The recent movie Bird Box is a great example. Total garbage, derivative movie, clumsy, contrived villains, manufactured tension with no actual plot driving any of it. Yet became somewhat of a viral hit.

If this is what Netflix’s data lets them optimize towards, count me out.

[thanatropism]

It is.

I must have expressed myself really badly, since people are reading the opposite of what I intended.

Netflix bet the farm on computer-generated plots, which is partly how they have such an immense quantity of content - yet the badness shines through. OTOH the handful of decent shows they have are all fully fledged concepts that they bought.

Like most people I really have to rethink the value-per-$ of this proposition, but if Netflix just erased all the crap maybe 10 up-to-snuff and 2 to 3 really good shows a year would remain. Worth paying what it is? If every new show was like "1983" or "Fauda"...

[tome]

"Computer-generated plots"? What? That's bizarre. Tell me more!

[jerf]

Netflix isn't outsourcing all of it's script work to computers literally, of course, but they can definitely do "People like historical drama... people like boats... let's do a historical drama on a boat!" sorts of things, which may not literally be coming straight out of a computer, but it might as well be.

When my kids were younger, my wife and I used to joke about some of the kids shows that way, too... "kids like dinosaurs... kids like trains... Dinosaur Train!" If you don't have kids or aren't in the US or something... I'm not just making up a joke: https://en.wikipedia.org/wiki/Dinosaur_Train

[ardy42]

> When my kids were younger, my wife and I used to joke about some of the kids shows that way, too... "kids like dinosaurs... kids like trains... Dinosaur Train!" If you don't have kids or aren't in the US or something... I'm not just making up a joke: https://en.wikipedia.org/wiki/Dinosaur_Train

Netflix is literally doing the exact same thing: https://en.wikipedia.org/wiki/Dinotrux:

> Dinotrux is an American computer-animated web series. It features a fictional prehistoric world inhabited by hybrid characters that are part dinosaur and part mechanical construction vehicle. The larger Dinotrux are accompanied by Rotilian Reptools who are smaller reptiles combined with mechanic's tools.[3] The series debuted on August 14, 2015 on Netflix...

[brazzy]

> they can definitely do "People like historical drama... people like boats... let's do a historical drama on a boat!" sorts of things

They can do a lot more than that. They know precisely at what points people pause episodes or give up viewing a series entirely, for example. I'm sure someone must have thought of using that to determine what plot structures and pacing captivates viewers.

[treis]

>they can definitely do "People like historical drama... people like boats... let's do a historical drama on a boat!" sorts of things,

Which is the Achilles Heel of all recommendation engines. Just because I watched one historical drama doesn't mean that I want all historical dramas all the time. Take a show like Stranger Things. The campy nostalgic thing worked for it, but that doesn't mean I want my next 5 shows to be campy and nostalgic. I want my next show to be something original like Stranger Things was.

[dasmoth]

Ken Liu wrote an enjoyable short story ("Real Artists") about exactly that. Just looked it up and it was published in 2011. One of those unusual bits of SF that becomes, if anything, more credible over time.

[brazzy]

I think that was hyperbole for "plots that are to some degree based on datamining Netflix' amazingly detailed knowledge of people's viewing habits".

[taurath]

> From a cultural or art critical point of view, this is a negative though

And keep in mind, this isn’t actually an advantage they have - scroll through any channel list and you’ll find a TON of crapware tv entirely designed to hook the lowest common denominator for as cheap as possible. That is reality TV. Yes they have fewer barriers to entry for promotion vs having to get it onto a tv schedule, but data points overall just select for optimization along one or two axes (cost, viewership numbers being the most frequent).

Shark week is an example of a viral hit - it’s not that Netflix makes more because they have more genre data.

[erikbe]

It's highly questionable that you can produce _great_ shows based on data.

You can't just go to a producer and tell them, make something about Japan, because it's trending, and use these two actors because they're the most popular at the moment. Creatives don't work based on directives.

[thanatropism]

Which is why the best shows Netflix has tend to be full fledged pitches that they acquired. Compare "Great News" to "The Kominsky Method" or that utterly-unwatchable-beyond-the-second-episode Australian dystopian political show with the Polish dystopian political show "1983".