How I Think, Part Two: Distrusting Individuals

post by Richard Henage (richard-henage) · 2023-11-08T04:06:13.546Z · LW · GW · 6 comments

Contents

6 comments

I've talked about movies before, so I guess I'll continue with that. I spoke with someone who seemed to distrust the people that decide which movies win the Academy Awards (at the time of writing, it's a group of over 7,000 people who have to have movie credits and be invited by a board to join the voting pool according to Wikipedia). Now, I don't think those people always get it right. I remember reading on Wikipedia that some of them said if they could go back in time, they would have voted for All The President's Men instead of Rocky to win Best Picture. If that's true, obviously someone is wrong. So I don't put a whole ton of stock into what they say. But still, I believe that it's probably one of the best ways to judge good movies: how many Academy Awards they've won.

Other people are crazy. I was crazy. I used to put a ton of stock into what people I met face-to-face said. If someone said that a certain movie was their favorite movie, I assumed that movie was probably good. If someone said a movie was not worth watching, I believed them. Now I just nod and smile and let most of what they say slip off me and ignore it. That sounds rude. I guess I probably say something like "I'll check it out". If the person goes into more detail about what they like about the movie, like how it's the greatest action movie because it's super realistic about the way guns work or something, I jot it down in my head and look it up later.

Don't trust individuals. Especially when talking in-person. First of all, there are way too many steps in the process for things to get garbled. Maybe the person just has weird or unusual tastes. There's always a chance they are lying or exaggerating or something, too. Then, of course, they could have remembered the name of the movie wrong, or, separately, said it wrong. You could hear the name of the movie wrong, or, separately, remember it wrong later. That's six potential ways of getting things wrong right there, and there have to be about fifty more.

Think about it this way. You have easy access to dozens of different movie rating systems where you can see how hundreds of thousands of other people feel about movies. Why would you pick just one of those reviews and trust that one instead of the average of the group?

Now, I think it's fun to write down when someone recommends a movie, and then later, if I ever end up watching it, go back and tell that person that I watched it and (hopefully) liked it.

Okay, that's how it works for movies. Facts are even worse. I think people can be wrong about how the feel about a movie (maybe they just really want to like it or something, I don't know). But for the most part, people should kinda be trusted to know how they feel about something. But when they're not reporting on their own feelings, but on a fact? Depending on the person, there's like a fifty percent chance that they're just completely wrong. I've had so many times where I fact-check what someone said after our conversation, and there's just no truth to what they were adamantly asserting, whatsoever. Then, even if they were on to something, what's the chance that they remembered and said everything correctly? I mean, what's the chance that you look it up on Britannica or Wikipedia later, and it says pretty much exactly what that person said? I'd say pretty low. Even if everything they said is true, the overall message might not be. The way they emphasized the importance of parts of their information might be off or something. 

Of course, you can't trust me on any of this, since I'm just one person. I think believing what someone wrote is a little better than believing what they said. But not much better. A couple of the error-prone steps are gone when using writing, but certainly not all of them, or even the most important ones.

 There are probably some topics that individual people are comparable with in accuracy vs a trusted internet site. And there are some things that the internet can't really do at all (without AI), like tailored advice. And it's probably important and valuable to get different peoples' opinions on like political issues or something. I'm not sure if the selection bias for political opinions is worse online or in-person.

I guess I trust people kinda like how I trust Chat GPT. If they say something interesting, I look it up and verify it. They might sound really confidant about it. They might have all these details attached to it that make it seem like it's more likely to true. They might even cite their sources. But I mean, they're just this kinda black box that spits out information. Unless they are actually reading off the information from a website in front of you, there's too many unknown factors.

Maybe the worst thing about trusting individuals is that it snowballs and people start repeating things they hear other people say. Maybe the worst thing is that when you really believed someone, learning later that they were completely wrong can damage your relationship. Or maybe not trusting people is a cure worse than the disease. But I don't think so. I've still had some pretty meaningful conversations with people I don't really trust to do anything except to report their own feelings accurately. I've sometimes learned interesting things from people that turned out to be true after looking them up. It does make it hard to have an extended factual discussion with someone. I mean, I just end up saying "huh, interesting" a lot, or "if that's true, I wonder what it means about X". Sometimes I can pull out my phone and start verifying claims, which makes me more comfortable making my own claims, and overall lets me engage a lot more in the discussion. It probably makes it less fun for the other person, though. Or not.

6 comments

Comments sorted by top scores.

comment by CrimsonChin · 2023-11-08T18:56:37.132Z · LW(p) · GW(p)

Maybe if you have very "standard" tastes this is good advice to get movies that appeal to large audiences.

If you were the type of person who ONLY likes movies where the guns are super realistic, you aren't going to love Titanic and other top-rated movies. 

I have certain people I categorize as "well-aligned movie watchers" like my brother that I grew up with. We have similar tastes. I find that gets me further than aggregate rating systems. 

Replies from: richard-henage
comment by Richard Henage (richard-henage) · 2024-06-14T00:40:10.547Z · LW(p) · GW(p)

I have certain people I categorize as "well-aligned movie watchers" like my brother that I grew up with. We have similar tastes. I find that gets me further than aggregate rating systems. 


For some reason, that statement has stuck with me for a while. I finally realized why it doesn't sit well with me. I agree with your message to some extent, but here are some problems I see with it:

1) Your brother still needs a way to find good movies (you can't pull each other up by your bootstraps)

2) I wonder if you are thinking of movies as a binary "like" or "don't like". If your bother's recommendations provide you a system for only watching movies that you end up liking, that's a valuable resource. But I see movies as more boundless in how good they can be. Sure, there are movies that I "liked", for example, Greyhound. I enjoyed it and didn't have any problems with it. But it didn't have a big impact on me and wasn't particularly memorable. I would give that movie a Thumbs Up. Then there are movies like Lord of the Rings, 12 Angry Men, and Memento. These are movies that are very meaningful or amazing to me. I want to watch them many more times throughout my life. I would give them each a Thumbs Up as well, but that doesn't really do it justice. I could give them a 10/10, but that doesn't quite fit either, since I assume that I'll eventually find movies I like even more than those ones[1]. So for me, finding someone who has similar preferences to me isn't enough. I need something that can sort through the hundreds of thousands of movies out there and point me to the ones that I'll like the most of all of them. If movies are more of a casual thing to you and you're not trying to optimize your experience, the "like", "don't like" system makes sense. Otherwise, I'd like to hear your thoughts so I can further optimize my system (currently, I'm using a spreadsheet that combines data from multiple online sources).

Of course, if your statement "I find that gets me further than aggregate rating systems" really is true, then what I said here doesn't matter.

  1. ^

    In this way, I like IMDb more than Rotten Tomatoes, since Rotten Tomatoes has 500 movies with a 100% rating, while IMDb has only seven with a 9.0+ rating and none with a 10.0 rating, meaning there's still room to grow. On a similar note, they have to give Oscars to somebody, so that evidence doesn't count for as much. If the police had to arrest somebody, they might end up arresting some random homeless person just because he was the most suspicious person they could find. If the Academy chose not to give out Oscars some years, (and some years gave out multiple), it would theoretically increase the trustworthiness of the award.

Replies from: CrimsonChin
comment by CrimsonChin · 2024-06-24T09:45:52.404Z · LW(p) · GW(p)

That's fair. I guess if you will allow me to re-state my idea with your ideas in mind:

Aggregate rating systems are best. But occasionally they are wrong. Sometimes I will see a movie has a high rotten tomatoes score but I still hated it. I don't watch a lot of movies so this happens often actually. Having someone similar to me, who is watching highly rated movies can usually save me time by predicting whether I will like something or not.

You are right aggregate is best. But I think having an aligned friend who knows your preferences can help build on top of that. Or rather help me filter.

I'm my original post I failed to realize that my brother and I are both watching movies that are highly rated. I think using both in conjunction works great. So I'm not disagreeing with you, rather building off your thoughts.

But to pose an interesting question. If tomorrow, rotten tomatoes had an option to find a user who EXACTLY matched the ratings you gave. For instance both of you gave Titanic a 7, Forest Gump a 8, etc. Then the algorithm tells you this user that most matches you, watched Shrek 6 yesterday and gave it a 10/10. Would you prefer that algorithms suggestion over just an aggregate average rating?

There are a lot of cool algorithms that could be applied to this, even a neural net that could take your past ratings and "predict" movies you would like next. I'm sure one might be more accurate than averaging method. SVN or KNN algorithms seem promising off the top of my head.

comment by Viliam · 2023-11-08T09:44:49.749Z · LW(p) · GW(p)

I guess I trust people kinda like how I trust Chat GPT.

Haha, perfect!

I think there is also recency bias in other people's recommendations. Fresh emotions are stronger in mind. I have seen some good movies recently that I would recommend now, but I am not sure how I will feel about them 1 year later.

A possibly better solution (but lot of paperwork) would be to write down the recommendations, and ask those people how they feel about them 1 year later.

*

Also, different people like different kind of movies. That makes me think... should we treat differently recommendations that people make within their preferred categories, compared to those they make outside their preferred categories?

As an example, I really liked The Eccentric Family (anime), and Before Sunrise (romantic movie). If I tell you that in general I love anime and hate romantic movies, should you treat these two recommendations differently? How specifically? If I like a movie from a category I generally hate, does it suggest it might be an exceptionally great movie? Or just that it is unusual within its category... in some direction that I like... which may be completely irrelevant for you, unless you have a reason to believe that we have similar preferences.

comment by cata · 2024-06-14T21:27:07.525Z · LW(p) · GW(p)

It definitely depends. I think there are lots of people for which there are lots of domains of information for which they are highly trustworthy in realtime conversation. For example, if I am working as a programmer, and I talk to my smart, productive coworker and ask him some normal questions about the system he built recently, I expect him to be highly confident and well calibrated on what he knows. Or if I talk to my friend with a physics PhD and ask him some question like what makes there be friction, I expect him to be highly confident and well calibrated. Certainly he isn't likely to say something confident and then I look on Wikipedia and he was totally wrong.

In general I take more seriously what people say if

  • They are a person who has a source of information that could be good about the thing they are saying.
  • They are a person who is capable of saying they don't know instead of bullshitting me, when they don't know. And in general they respect the value of expressing uncertainty.
  • The thing they are saying could be something that is easier to actually know and understand and remember, instead of super hard. For example, maybe it is part of a kind of consistent gears-level model of some domain, so if they forgot or got it mixed up, they may notice their error.
comment by Sune · 2023-11-08T11:00:29.671Z · LW(p) · GW(p)

Corollary: don’t trust yourself!