Comment by brianstorms | Hacker News Reader

Comment by brianstorms | Hacker News Reader

Parent Full thread

Have been a Netflix customer since about 2000. I used to blog about them all the time in 2000-2001, did full writeups on their UI design, even the red envelope design. Had a dialog going to Neil Hunt their CTO/product guy. Even visited Netflix and had lunch with him once over the blog posts. I'd say they made about half the changes I urged them to make. Which was fun to see, that it made a difference.

As for their recommendation system: I always found it to be abysmal and worthless. It recommended films in genres I had never watched, had no interest in, like children's films, sports films, other niche genres. So as an experiment I started rating films Netflix would recommend in those genres to me, as having 1 out of 5 stars, the lowest rating you could give. [At some point they added "Not Interested" and I'd give it that rating. Ha! It was about as effective as pressing a pedestrian walk button at an intersection.]

I did this for months. Any spare time I had while waiting for something else to get done, I'd pop into Netflix and rate rate rate. Got really good at it. Kept doing it for more months. Eventually I rated over 20,000 movies, all in genres I had absolutely zero interest in ever watching anything in.

It made no difference. Netflix still recommended films in all kinds of genres, regardless of even if I had rated EVERY TITLE IN THAT GENRE as "Not Interested" or 1 out of 5 stars.

I once got on an elevator with Reed Hastings at a conference. As we rode the elevator I told him the story. He was ROTFL by the end. Laughter but as in a "too bad for you, I could care less" kind of way. He said no way they would not fix anything. He absolutely did not care whatsoever that Netflix recommended movies at the genre level when I absolutely hated the genre.

It was quite revealing, his "I couldn't care less" reaction.

Probably the same thing going on inside Google with user feedback on search results and ads.

At some point, after years of clicking the almost invisible [x] on the top of "hot singles near you" and ads for mail order brides from various countries and patiently telling Google it wasn't relevant, I got fed up, tilted, clicked the ad, found the contact address and explained in sufficient detail how Google scammed them by including me and others who absolutely weren't interested at all.

Sometime after that it suddenly stopped and now I only get irrelevant ads, not disgusting, embarrassing or insulting ones.

I can't say if what I did made the difference or not, but I can say that years of telling Google did change nothing. (Or to be fair: when I said I didn't want Russian mail order brides they would admittedly show me some other nationality next time).

Also today I have almost completely[1] given up Google and I don't feel I break an unspoken contract by blocking ads anymore.

[1]: Visited Google the other day and noticed they had brought back a working context menu with a working "view cached result" option, but it would take 200 more such fixes, many of them much bigger, to restore Google.

Pedestrian buttons change the lights. This is easily testable. What an odd analogy.

Oh phooey. It's a longstanding joke that the "Don't Walk" lights don't change to "Walk" any sooner if you press the button. Just like the "close door" button in elevators doesn't make the doors close any sooner than they normally would.

> Just like the "close door" button in elevators doesn't make the doors close any sooner than they normally would.

I had never thought about this. In Thailand they definitely do close the door sooner, or straight away, and everybody does it all the time.

In europe they usually close sooner, as well. It is allmost as if the world is not made by one standard.

> Just like the "close door" button in elevators doesn't make the doors close any sooner than they normally would.

I recently read an article (maybe even here on HN) that explained this fully.

Basically, in the US, under the Americans with Disabilities Act, the doors must stay open for a minimum time (which depends on some variables). The door close button can only have an effect after that time is elapsed, but in many (if not most bases) the time after which the door automatically closes is set to that same minimum anyway.

My understanding has always been that, at least in the US, it is usually the case that the "close door" button doesn't do anything unless there is a firefighter key inserted and the elevator in fire mode. That's really there to allow a firefighter to manually control the door open/close functions during fire-ground operations where you really don't want the elevator doing things on its own.

There are, of course, always exceptions. Some reading I did earlier suggests that there is some variation based on elevator brand, and possibly even then whims of the building owner (in terms of what instructions they gave the elevator installer about how they wanted things to work). Curiously though the "door open" button does usually pre-empt the (automatic) closing of the door, even in "normal" mode. At least IME.

It depends on the light. Some lights don't change unless they sense a car or pedestrian going the other way. Some run on a timer so they change automatically. On others, the light will change but the walk light will not come on if you didn't push the button (these ones drive me insane). Then there's the ones that only activate a chirp for the blind people. Then, finally, there are the fabled buttons of lore that do absolutely nothing. Oh, and some are just broken.

I still haven't come across the placebo buttons though so I'm willing to doubt their existence.

> I still haven't come across the placebo buttons though so I'm willing to doubt their existence.

IIRC the "placebo" buttons, where they exist, were not installed as such, but simply deactivated after the traffic light control system changed, since uninstalling them would be more expensive.

> I still haven't come across the placebo buttons though so I'm willing to doubt their existence.

How do you know you haven't?

In my home town in Germany (and probably other cities as well, only I've realised I've never really paid attention to that subject while visiting other cities), the little tactile arrow marking the crosswalk direction plus a tactile buzzer for blind people is mounted together with those buttons, so even when a certain direction and its parallel crosswalk always get green by default (and therefore there's no need ever to push a button), a button housing still gets installed for the buzzer and arrow.

The only difference is that those kind of non-buttons are painted a different colour, plus with the older style of buttons that you physically have to push in order to operate a switch, on those non-buttons the push-plate is fixed in place and cannot actually be pushed. (Some people occasionally still try to.) With the newer style of button that operates via some sort of electronic touch sensor and therefore has no moveable parts to begin with, the only distinction might be the colour? (Not sure, because the intersection I regularly pass by on the way to work still uses the old-style buttons…)

Not in NYC

In many places in the US, pedestrian buttons don't change the lights, certainly not within 1-3 min, if it's a busy or major intersection with lots of vehicle traffic. Also, the controller behavior can change during the day/night depending on traffic volume, or maybe time of day. One time I counted it took the lights 7 minutes to change at a three-lane intersection late at night, even though there were no cars.

https://news.ycombinator.com/item?id=12330251

https://news.ycombinator.com/item?id=17900576#17906527

BBC: "Does pressing the pedestrian crossing button actually do anything?" https://news.ycombinator.com/item?id=6326588

Not necessarily that testable.

For example I used to live near a major junction, for about 18-20 hours per day the buttons have no effect because there is traffic, the junction's computer is adding walk cycles where they fit in its schedule anyway so there's no difference. Testing would show no effect.

Late at night you can see the buttons work, there is no traffic so the junction is left with the fast through lanes green indefinitely, but pressing a pedestrian button causes your walk cycle to run immediately.

On the other hand I work on a University campus, the University would prefer to shut the road through the middle of it, local government said "No" so the pedestrian crossings are set to run very frequently with no penalty delay, meaning most of the day if you push the button you get a walk cycle immediately. During term time it's probably quicker to drive around the campus to the far side than use that road - because somebody will push the button, probably more than once, there are like four crossings with this treatment. So on campus it's very obvious that the buttons work.

Reply on news.ycombinator.com