Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[silverspawn]

the solution to everything is just more mindfulness

[silverspawn]

[markusin]

there's probably some kind of a tree structure going on in the background

yes, there is

I still think it's fair to say that having a datastructure that you can find elements in O(ln(n)) time is the main idea of a data base? Yes, you can also do that with a text file. You can have text in a text file sorted. You can have several text files that structure your data. You can in-principle keep a structure that gives you O(ln(n)) access time forever, along with optimality on all the other operations. You can have a text file such that you maintain an implicit tree over it.

But then you're just emulating the properties of a data base manually instead of automatically. I mean, ultimately there is no sharp distinction between any of this stuff. "Database" is not a primitive object in the computer, ultimately it's all processor operations over binary data.

So maybe a more accurate description would be "there's a bunch of nice properties that you'd like your data to have, and if your use case requires large amounts of data, at some points it becomes sensible to automate those, and that's a data base. But like, pedagogically simplifying it to "databases keep your elements sorted and that gives you faster access time"... still seems reasonable to me

Worth noting, SQLite does store its database in a single file.

[silverspawn]

Trump has a 36% of being nominated and a 40% chance of becoming president. Wut?

[silverspawn]

So in movies you sometimes see "honorable" criminals that have a code of morals which says that killing 18+ yo people (especially men) is totally fine, but killing children is extremely super bad.

Is there any steelman to this position? It seems pretty nonsensical to me. If anything, the value you assign to a life should be a bridge shape, peaking somewhere at 20 or something. I get that it's not honorable if the other person can't fight back, but that doesn't make any sense if you use guns and hit your victims unprepared (since then no-one can fight back).

[Awaclus]

So in movies you sometimes see "honorable" criminals that have a code of morals which says that killing 18+ yo people (especially men) is totally fine, but killing children is extremely super bad.

Is there any steelman to this position? It seems pretty nonsensical to me. If anything, the value you assign to a life should be a bridge shape, peaking somewhere at 20 or something. I get that it's not honorable if the other person can't fight back, but that doesn't make any sense if you use guns and hit your victims unprepared (since then no-one can fight back).

Children are cute, and therefore their lives are more valuable. Adult men are not cute, and therefore their lives are less valuable.

[silverspawn]

This is my suspicion as well, but not a steelman!

[faust]

So in movies you sometimes see "honorable" criminals that have a code of morals which says that killing 18+ yo people (especially men) is totally fine, but killing children is extremely super bad.

Is there any steelman to this position? It seems pretty nonsensical to me. If anything, the value you assign to a life should be a bridge shape, peaking somewhere at 20 or something. I get that it's not honorable if the other person can't fight back, but that doesn't make any sense if you use guns and hit your victims unprepared (since then no-one can fight back).

I can try this.

The criminal justice system treats children differently from adults. This is because they are not fully formed moral being and as such do not have full responsibility for their actions.

If you decide it is okay to kill people, you presumably have some sort of justification for why they deserve that. You are, in effect, acting as the justice system yourself. In that way, it makes sense to treat children differently for the same reason the criminal justice system treats them differently.

[silverspawn]

That's pretty good!

Although it doesn't make sense if you're willing to kill people you know nothing about. Some adults gotta be innocent as well

[silverspawn]

New attempt to understand IMDB ratings

The Game of Thrones Finale has a rating of 4.1. It was criticized for bad writing. The Rise of Skywalker has a rating of 6.5.

So instead letting my head explode, my new theory is that my tendency to have a globally consistent set of rules is actually just very unusual and that's not how most people rate things. In an absolute sense, the final episode of GoT has infinity better writing than Star Wars, but in GoT was disappointing because it used to be really good and Star Wars wasn't. I mean don't actually agree that the final episode is super bad compared to the rest of the show, but I can see how you think that.

Basically this new model is that people aren't consistent across different genres.

But it's not good model because I don't know how people apply their standards. I guess with these two examples it's kind of obvious, but with stand-alone films... maybe it depends a lot on the presentation.

[silverspawn]

And the Rise of Skywalker does seem to me to be super bad even by Star Wars standards, so it's not just about relative difference but also that Star Wars viewers just care far less about writing quality.

[silverspawn]

Adventure Time finale has a rating of 9.7

[Awaclus]

I have never seen any GoT or Rise of Skywalker, but isn't the former based on a book and the latter original? A ShiTty adaptation of great source material can be more compelling to watch than an underwhelming original work, but as a work of art, it is inferior. Most people probably won't consider it that philosophically, but they'll intuitively get more or less the same result anyway because they feel like the adaptation ShaT all over the original and then they rate that.

[Jimmmmm]

And the Rise of Skywalker does seem to me to be super bad even by Star Wars standards, so it's not just about relative difference but also that Star Wars viewers just care far less about writing quality.

I think Star Wars ratings are a mess in general. 6.5 does seem high for TRoS, but I would suggest that it's boosted by those who hate TLJ and see it as a big middle finger to it, and also by those who are anti-anti-sequels.

There seems to be a major divide among Star Wars fans between those who want to enjoy each new instalment on its own merits, and those who care as much or more about the effect that it has on the overall story. I get the sense that those on each side are pushed further towards loving or hating something in response to the other side.

[vidicate]

IMDb ratings. A couple thoughts-

Comparing user ratings for movies with user ratings for shows is quite inconsistent.

Even across genres-
A comedy film with a rating in the 6s is probably a good watch, for a comedy. A drama film in the 6s is probably mediocre (for a drama), though if you’re a fan of a particular aspect of the movie (an actor, the writer/director, the book, the niche sub-genre, the type of art/cinematography) you will probably enjoy it.

For user ratings of individual show episodes, two quirks to keep in mind-
  One is that the later seasons are watched and rated predominantly by the die hard fans (who will love everything, until they don’t); each season will have a smaller, more dedicated following than the previous, typically.
  Two is that for shows which air one episode at a time, the ratings/reviews will be heavily skewed by the viewers at the time each episode aired (without the foresight of peers telling them it gets better, nor the hindsight of bingeing ahead and coming back to rate it). This is significant for long-form stories, shows that later have a big relevant “reveal”, and shows that later go through a change in style, pacing, direction, etc.
  For example, I was once checking episode ratings of The Walking Dead season 10. I had almost finished bingeing two seasons, and enjoyed the last few episodes of 10, which were “bottle-episodes” mainly focusing on some of my favorite characters. I discovered that, when they aired on AMC, these were “bonus” day-in-the-life episodes tacked on after the main story’s full run (i.e. the main story still got the usual amount of screen time for the season). As they aired each week the viewers hated them. They were hungry for something meatier after waiting week to week. I didn’t care if they took some time to give some extra love to the show; by the time I got to these episodes, season 11 was already airing and I didn’t have that “what’s this crap I need to know what happens next” feeling.

[faust]

That's pretty good!

Although it doesn't make sense if you're willing to kill people you know nothing about. Some adults gotta be innocent as well

Well, you could adopt the idea that no adult is innocent. Christianity does this, so it's not particularly out there. I will admit that this makes it harder to argue why children deserve different treatment but it's still possible (e.g. if the sin stems from sexuality).

And often enough a criminal knows at least something about their victims. Possibly they are cops, and well, ACAB. Or maybe you've abducted a plane and are holding the people in it hostage. Well you know they're rich enough to fly, and not environmentally conscious enough to avoid it, and either may be enough to condemn them.

[silverspawn]

IMDb ratings. A couple thoughts-

Comparing user ratings for movies with user ratings for shows is quite inconsistent.

Even across genres-
A comedy film with a rating in the 6s is probably a good watch, for a comedy. A drama film in the 6s is probably mediocre (for a drama), though if you’re a fan of a particular aspect of the movie (an actor, the writer/director, the book, the niche sub-genre, the type of art/cinematography) you will probably enjoy it.

A testable hypothesis! I'd say I'd verify it, but I can only do it for drama since I don't really watch comedy films.

For user ratings of individual show episodes, two quirks to keep in mind-
  One is that the later seasons are watched and rated predominantly by the die hard fans (who will love everything, until they don’t); each season will have a smaller, more dedicated following than the previous, typically.
  Two is that for shows which air one episode at a time, the ratings/reviews will be heavily skewed by the viewers at the time each episode aired (without the foresight of peers telling them it gets better, nor the hindsight of bingeing ahead and coming back to rate it). This is significant for long-form stories, shows that later have a big relevant “reveal”, and shows that later go through a change in style, pacing, direction, etc.
  For example, I was once checking episode ratings of The Walking Dead season 10. I had almost finished bingeing two seasons, and enjoyed the last few episodes of 10, which were “bottle-episodes” mainly focusing on some of my favorite characters. I discovered that, when they aired on AMC, these were “bonus” day-in-the-life episodes tacked on after the main story’s full run (i.e. the main story still got the usual amount of screen time for the season). As they aired each week the viewers hated them. They were hungry for something meatier after waiting week to week. I didn’t care if they took some time to give some extra love to the show; by the time I got to these episodes, season 11 was already airing and I didn’t have that “what’s this crap I need to know what happens next” feeling.

Ya, that all rings true

(Also, I had no idea TWD went on for 10 seasons  I dropped it somewhere in season 2.)

[silverspawn]

Random thought about C. before I forget it

A lot of people imagine a close link between how much an agent satisfies its preferences and how good it feels. Or when applicable, a close match between the reinforcement signals of a reinforcement learning agent and how good the agent feels. Like in humans!

But in an AI, the reward signal is totally arbitrary. A reinforcement learning agent only optimizes for high reward because it's explicitly programmed to optimize for high reward. So what do these people expect happens when you change one character in your code to make it optimize for low reward instead? Presumably, happiness is not literally tied to the value of a single number, so the agent should now feel good if it achieves low reward (since  that's what it "wants")? But even then, you'd still get totally different valence from changing one character.

It sure does seem like humans don't optimize for avoiding pain because it was arbitrarily chosen, but that pain is an objective thing that systems naturally want to avoid, or something. Like, it feels like the pleasure/pain axis is inherent and evolution accessed it to get reinforcement learning behavior for free. But that picture doesn't square with functionalism.

I think that argument is worth polishing up and using somewhere

[silverspawn]

And the Rise of Skywalker does seem to me to be super bad even by Star Wars standards, so it's not just about relative difference but also that Star Wars viewers just care far less about writing quality.

I think Star Wars ratings are a mess in general. 6.5 does seem high for TRoS, but I would suggest that it's boosted by those who hate TLJ and see it as a big middle finger to it, and also by those who are anti-anti-sequels.

There seems to be a major divide among Star Wars fans between those who want to enjoy each new instalment on its own merits, and those who care as much or more about the effect that it has on the overall story. I get the sense that those on each side are pushed further towards loving or hating something in response to the other side.

Worth noting that the reviews are much more negative, which sort of fits since the people who like it out of "principle" will have a harder time articulating that well

[silverspawn]

Also, isn't there regular reinforcement learning in humans where we optimize for stuff that doesn't actually feel good, and you can recognize this and try to stop optimizing?  Thinking about certain status-related instincts here, but there are probably other examples

[silverspawn]

That's pretty good!

Although it doesn't make sense if you're willing to kill people you know nothing about. Some adults gotta be innocent as well

Well, you could adopt the idea that no adult is innocent. Christianity does this, so it's not particularly out there. I will admit that this makes it harder to argue why children deserve different treatment but it's still possible (e.g. if the sin stems from sexuality).

And often enough a criminal knows at least something about their victims. Possibly they are cops, and well, ACAB. Or maybe you've abducted a plane and are holding the people in it hostage. Well you know they're rich enough to fly, and not environmentally conscious enough to avoid it, and either may be enough to condemn them.

I do think a concept of innocence with 0 entropy beyond age is stupid, but nonetheless, good steelman!

[silverspawn]

I have never seen any GoT or Rise of Skywalker, but isn't the former based on a book and the latter original? A ShiTty adaptation of great source material can be more compelling to watch than an underwhelming original work, but as a work of art, it is inferior. Most people probably won't consider it that philosophically, but they'll intuitively get more or less the same result anyway because they feel like the adaptation ShaT all over the original and then they rate that.

Well, GoT used to be based on a book, but George R R Martin still hasn't finished book 5, which (along with natural diversions due to the butterfly effect) meant that the series diverged from the books increasingly sharply. Which did come with the inevitable decrease in quality, especially with things like how fast characters can move from one point to another.

I think season 8 is almost entirely separate from the books.

[silverspawn]

Correct: still hasn't finished book 6. SOmetimes I do wonder whether he will manage that before his death, let alone book 7.

[silverspawn]

correction: correct should be correction. now it's correct.

[Jimmmmm]

Yeah GoT 8 and the SW sequels are similar in that they're the continuation of a story without much or any of the original writer's input.