Hacker News new | past | comments | ask | show | jobs | submit | mrshoe's comments login

The real problem is industry-wide adoption of metrics-driven design.

When I studied cognitive science, the method we employed for HCI was User Centered System Design. Granted, this was a backronym for UCSD, but the point remains — design should be user-centered, not metrics-driven.

When I launch Netflix, it is abundantly clear to me that their design is driven by their metrics. What I want to see at the top is my “continue watching” shows — 99% of the time I just want to watch the next episode of a show that I’m already watching. Instead, they show me row upon row of shows that I have never watched. Their metrics prove to them that getting me hooked on new shows will increase my engagement and increase the amount of time I spend in their app. Guess what? As a user, those are not my goals! Their UI is effectively one big advertisement for Netflix itself. Wonderful.

Unfortunately, virtually all tech companies have accepted metrics-driven design as conventional wisdom at this point. Run an A/B test and see which button treatment performs better, based on the metrics the company cares about — not what the user cares about.

One outlier here is Apple. They do not design based on experiments and metrics. And the Apple TV app does in fact display the shows I am already watching as the top row. Go figure.


This brings to mind my experience as an early facebook adopter (when it was still college email invite-only), to some point a while ago when they went from a chronological timeline/feed to a "curated" one. Users overwhelmingly did not want or like this, because the main use of facebook at that time was to connect and chat with people you knew in real life. However, that does not drive ads as well as what exists currently, so here we go.

If users were pickier about what they consume things would probably be better - that's also an issue (and I admit I am part of the problem). I can't even remember the last time I saw a close friend's post on my feed.


I have found a hack around facebook that partially works: remove yourself from all groups. So only friends are in your feed. Then every time a friend shares something from one of their group go hit "block all from [group]". Facebook now has less junk to show me as I've blocked most of it, and so I see my friends. I've also set a rule that when I block two groups I'm done for the session. Anytime I share something on Facebook I make sure the privacy settings is "friends of those tags only" - if I'd want more or less than that it is either too private to share at all, or it should be on a better public forum.

I hope more of you start doing the same. There is value in a Facebook type private place to communicate with friends and family (this need not be facebook, but that is where my friends are), and everything else clutters it. The more people who follow similar rules to me the more likely Facebook is to notice in their metrics that people only want Facebook for personal friends, hopefully they adjust to enable that better.


You can filter your feed to only show friends stuff (and it's chronological)

https://www.facebook.com/?filter=friends


not on mobile


Welcome to enshitification.


This seems to be almost synonymous with metrics-driven design.


>I can't even remember the last time I saw a close friend's post on my feed.

It’s probably because they aren’t on Facebook any more.


For netflix : I only watch it on my browser (my TV is basically a monitor connected to a PC), and the way I work around is by having a bookmark : https://www.netflix.com/browse/continue-watching Problem solved! (well until netflix starts blocking that too, in that case I might as well delete their membership)


That kind of criminal behavior on your part is why Netflix wants you to use the app, you lowlife. How dare you?


> Their metrics prove to them that getting me hooked on new shows will increase my engagement and increase the amount of time I spend in their app. Guess what? As a user, those are not my goals!

But here's what I find odd about Netflix. Up until extremely recently—and even now for most customers—Netflix did not and does not display ads. Netflix makes money from subscription fees, which means they get the same amount of money whether you spend zero, 10, 50, or 100 hours per month in their app, as long as you keep paying for your subscription.

Netflix has apparently concluded that users who spend more time in their app are more likely to remain subscribed. I wonder why that is, or if Netflix is even correct. I would think that for most users, the quality of their time matters more than the quantity.


Indeed, given the relatively high cost of streaming high-quality video, you'd think Netflix makes substantially more money from the 0-hour viewer compared to the 100-hour viewer.


There do seem to be "churners" out there subscribing to services only for as long as they need to watch the show they want to see, and I'd guess a wide middle of customers who are at least prepared to cancel if they feel that they're not using the service enough or that they will get better value from an alternative.

To keep these customers, you need to keep them continuously hooked on your programming.


If I was Netflix, and I was optimizing for this type of customer, I would want to have them moving through my content as slowly as possible! My goal would be 2-4 hours of Netflix per week, preferably spread across different days—just enough to ensure they haven't lost interest! Autoplay Next Episode would not be a thing, and I certainly wouldn't release new seasons all at once.


Maybe they prefer dominance over profit. Make less money but maximise revenue. They want you to spend 100 hours a month not 50 even if you pay the same so that you are not spending time at a competitor such as Disney or hanging out with friends at a bar.

Eventually they may go like Amazon with infinite upsells and hunt for whales.


How could this be good business?


Maybe the LTV is higher. Or it attracts more shows to Netflix than competitors. Ot it secures a moat.


Unless they’re selling user data.


I think it's probably because it's extremely hard to understand what *content* keeps users subscribing, and "hours spent watching this content" is maybe the lowest-hanging (or only?) "clean" metric you can use to determine what content to keep, how much to spend on content, etc. I watch a smattering of shows on netflix and so keep a subscription, but I don't watch very much. How the heck are they supposed to know whether a piece of content will keep me subscribing? And how much should they spend to produce/acquire that content? It's a very difficult supply-side question.


Worse, they loop through when you scroll horizontally and replace the thumbnails with new icons for the same shows, so you lose track of whether you’ve wrapped around or not.

I’ve been waiting for the day when we wake up and realize how cynical and impersonal the world of A/B testing is and consumers insist on better treatment.


I agree with everything except the last bit. I have purchases on Apple TV, and I'm not a subscriber, but when I launch the app, all I see is subscriber-only content. Apple Music is similar.


Apple absolutely does do this. I have first hand knowledge of doing it for them.


Sounds like something you probably should not be talking about on the internet with your real name?


I'm allowed to say that much. I know my NDAs very well.


The real problem is big-money-driven design. Big-money stakeholders don't care about non-measurable things like user experience or user well-being. Big money thinks finance, and finance is always metrics, so everything else becomes metrics too.

A brilliant essay.


Do these fields care about depth and worth-done efforts from a user ? this decade is about giving everything for free, consumption flow effortlessly, which is obvious a good point to aim, but I remember the things I liked the most are those which made me work a little to make complex things. And I can't find recent advances on this terrain.


Notably, Plex (a free media server) also displays "continue watching" stuff at the top.


Yes, and not only design, it's metrics-driven everything.

And often the metrics can be counter productive in the long term, but by then people have collected their bonuses, promotions or left the company.

Here is a example I encountered from few years back where Google Ads was loading slower then the search and ended up replacing the first result just when you were about to click it. https://twitter.com/gnyman/status/1257239940309622784

If it was intentional or not I don't know, but I recall that this persisted for many months before being fixed.

I think it's obvious (but maybe I'm wrong) that accidentally clicking on ads is not what google wants, long term. As that won't get any real results for the advertiser. Which in turn reduces their interest to spend. But short term I bet Google made a lot of buck on this.

Either way, maybe it's unavoidable in a capitalistic world, it's not like non-software companies were driven by a consumer happiness score before either. The reason it feels so bad is just that a lot of us have been here during the growth phase when end-user happiness is a more important metric.


What you describe is the capitalist perpetual September.

Long term you is less valuable than short term you.


You assume that your goals are aligned with the company’s goals. They’re not. There is no “problem”, it’s all working as intended.

From the company’s point of view, they actually are doing the right thing. They’re making money (which, I’d argue, is ultimately a good thing for everyone involved).

Netflix and Apple are no different, the goal is always $$, Apple just sells different things.


No, he's saying the exact opposite - that the company's revenue-driven goals and the user's goals are not in alignment. It is definitely possible for those things to be in alignment, that's how products have worked for the majority of modern human history.


It's a bit more complex than that right? If no part of the user wanted this, it wouldn't drive up engagement. Clearly it causes me to use the product more. Companies can try to optimize for stuff like "satisfaction after using the product" but it's genuinely much harder -- and not tying work to business outcomes is not scalable.


Most drug addicts don't want to be addicted, they just want the high. Most video watchers want to continue their current shows, but if you make it easy they will get addicted to shows they didn't know they liked yet. People get on facebook for their friends, and then the infinite scrolling keeps them on long past when they have seen almost everything their friends have done.


I think the broader point is that the User's goals and Capitalism's goals are not in alignment and the company, who is caught in the middle, has an obvious and legally required option to choose if they are shareholder owned. The user loses this fight every time.


We designed our own magnetic phone mount, which comes with the product. Once you start mounting your phone in your car (especially with our mount, which is amazing), you won't want a separate screen for your camera. :)

The advantages of mounting your smartphone, with the cutting edge hardware and active software ecosystem that come along with it, are truly huge. When I rent a car now I am incredibly frustrated without the mount. Luckily ours has a magnetic pluggable backend, so I can pull it off my adhesive mount and attach the vent clip for use in other cars.


We are using industry standard encryption algorithms for all of our wireless traffic. We are also consulting with external security experts, in addition to the expertise we have in-house.

All of our over-the-air firmware updates are cryptographically signed.


We are initially shipping only in the US, but hopefully total world domination will follow eventually. :)


We do require an OBD port, which is present on all cars 1996 and newer.

Other than that, it should work on basically all cars, assuming you don't have e.g. an existing backup camera that mechanically blocks the installation.


The solar window is actually transparent. Our PD team is second to none! They were able to make the whole solar assembly look uniformly black while not sacrificing solar efficiency.


The camera frame is indeed paired with the car adapter at the factory, which is a theft deterrent. They will not function without each other.

The RearVision also comes with a security tool that must be used to attach and remove the camera frame.


Excellent question. It takes decades for new features to fan out to the majority of new cars. After that it then takes decades more for those features to reach saturation of all cars on the road, because only 7% of the car population turns over each year.

The bottom line is it's generally 40 years from the point when features like seat belts, air bags, traction control, and backup cameras are first introduced and when they reach 95% saturation.

Throughout this adoption process, demand for these features among those whose cars don't have them increases significantly!


Excellent answer. Feature propagation & saturation is also a reason to offer/develop a version that doesn't require an OBD port.

Your customer base consists almost exclusively of those who are both willing to spend $499 on an accessory and who don't already have a reverse cam. Many potential customers won't have an OBD port in their car because OBD-I wasn't around until the '90s. Muscle car enthusiasts and parents (happy to spend $499 to prevent an accident) passing down an older car come to mind.

It would be great to see a second generation version that works without OBD.


Thanks for the very informative reply!


The RearVision comes with a security tool that is used to attach and remove the camera frame. The camera frame and the OBD adapter are also paired at the factory and will not function in isolation.


Thanks for the info...tell Tyler M I said hi!


The OBD adapter does serve a very important role in the system in that it has a power source and can wake the camera frame when necessary (before you ever launch the app!). All of our vision algorithms also run on the OBD adapter, which is a key aspect of the power model.


Both those things (remote wake and 'vision algorithms') are things that could likely work on the phone, no?


I think the remote wake is based on when the car is put into reverse, which would be challenging to tell from the phone. I imagine that only running the camera when the car is in reverse is a key part of keeping power consumption down in the frame.


Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: