Hacker News Reader

> The problem with that is good art or even entertainment is rarely made by analyzing a spreadsheet to determine what audiences want.

Exactly. I made this point a month or two ago. I cancelled my Netflix account for the first time since 2004. Because with all their data scientists and two decades of viewer data (more detailed than Nielsen ratings could ever provide), and billions of dollars... they still can't compete with AMC, HBO, etc. It's inexcusable. It's laughable. And it makes me think we are in a serious big data/data analytics bubble. When I unsubscribed, their crowning achievement was a "Top 10" leaderboard. Something any junior dev could throw together with a few SQL reports and a day or two of free time. That feature also highlighted the bargain bin nature of Netflix's catalog.

The other problem they have is they have eroded viewer trust. You can't actually watch new shows because you know they will be cancelled after a season or two. So new shows come out and no one watches and Netflix cancels them because of no one watching. It's a snake eating its own tail.

Plus people have realized they would rather wait until all seasons are out to binge watch than have to wait large gaps of time.

This habit of cancellation is, for me, the biggest failure of the lot. Some of their shows were very high quality and quite unique: One Day At A Time, The OA, The Baby-Sitter’s Club. You know what else all of these shows have in common? They were cancelled at the top of their game.

People forget that many huge shows, e.g. Breaking Bad, Battlestar Galactica, only became hugely popular several years into their run. The Wire is even more extreme: it only really became huge after the last episode was broadcast. But in each of those cases, someone made the call that a high-quality show was worth running, either because they hoped it would find a wider audience or because they believed it enhanced their brand.

Oh I miss Patriot Act with Hasan Minhaj. Netflix cancel it.

I'm still really angry they cancelled the OA, it might have not been the most popular show but I thought it was incredibly original and engaging... and it was clear the story had so much more to give.

I wish they had, at least, given it one extra season to wrap things up. I think there's a huge difference between cancelling a show and telling a show "okay you have one more season". Both in trust with filmmakers/writers/etc and viewers.

For this reason I have mostly just been watching shows that I know are either finished, 1 season, or guaranteed to be renewed due to massive popularity. I was sort of traumatised when they cancelled the OA.

FWIW the creators have been quite clear they didn’t think a “wrap it up” season would have worked. I can see why. Their whole storytelling method was unlike anything else in the business. Ironically, since they more than most shows just wildly ignored the convention that episodes should be the same length, it was uniquely suited to streaming. (Master of None does this, very few others do.)

>> The problem with that is good art or even entertainment is rarely made by analyzing a spreadsheet to determine what audiences want.

To navel gaze, what makes good anything? (Forewarning I'm eventually heading around to talk about guessing ahead on whether we're making good software or not.) Good furniture, good cars, good bicycles... how & when are groups effective in these places? Or even in these cases, perhaps there secretly are individual craftspeople (I'd consider a moviemaker a certain type of craftsperson) defining the outcome, steering; we think of these industrial productions as group efforts but perhaps there really there are embedded decision makers.

Is good software made by analyzing a spreadsheet to determine what audiences want? In what cases? Who can determine what to build, based off what? Who directs the software, makes it happen, based off what?

How are we defining good, who gets to pick? Many more layers of questions here. In software, are we seeing & judging software as an appliance/application, or as technology itself is? If every engineer working on a thing says, this thing sucks, it's just the worst, but people love & use the thing, what does that say? What gives rise to tech that has a strong cult following?

Movies are interesting because we're incredibly bad at grabbing numbers about them. We can count the number of actors, budget & allocation of budget. But how do we rate & estimate story, it's value? How do we calibrate & rate the chemistry of the cast & the audience's reaction to these actors they may well already have seen in dozens of other films, how do we assess the photography department's choice of shots, or the special effect's team's spectacularness or believability? Everything is radically under-specifiable, as compared to a bike, where there are defined parameters & form, generally. Software too is another ill-defined thing, is more formless in nature; true to it's name. It has such a mound of cofactors constantly defining it. There's so much taste & necessary continual discernment, picking well, that keeps compounding, that comprises the whole, like a movie.

Let's reverse engineer what a PM will think with data:

1. 80/20 rule: Target best performing shows, replicate success to capitalize on fan loyalty/momentum (new seasons or spins): The Crown, Money Heist, Designated Survivor, Narcos...

2. Validation: Look at high movie ratings, copy those categories: Korean dramas, documentaries, telenovelas, reality TV...

3. Culture: Follow trending themes on social media - feminism, financial freedom, woke...incorporate them into the screenplay, regardless of bias, craft, or message

4. Engagement: As long as we capture user attention and addiction, reinforce this habit via recommendations with AI algorithms

Like Facebook, they are focused on creating an addiction machine for short term gain, and this strategy is reflected on their increasingly low quality shows - this affects the brand, leading to low trust and rising churn

>no one is subscribing to a service to watch reality TV

I subscribe to Hulu to host Bachelor series watch parties.

However, I strongly agree that Netflix content is braindead and the service charges you more than its content is worth.

If you haven’t watched it, I highly recommend “UnReal”. It gives you the experience of watching reality TV and prestige TV at the same time.

It's there a term for this phenomenon where a 'thousand cuts' of small optimizations that improve x or y metric kill a brand?

I'm not above neologizing.

Garden Path Optimizing

transitive verb

1. to be misled by progressive or incremental optimization.

2. to achieve sub-optimality by pursuing incremental optimization.

I like this name. Like some kind of greedy algorithm type that makes the immediate optimum choice without regard to he larger picture.

It's both the drop in quality and the increase in cost. That turns Netflix into a really bad deal in the face of cheaper competition with more quality shows.

Netflix used to rule with affordable access to the best new shows, but now they offer neither.