> It’s well known inside and outside of technology that customer descriptions of their own behavior are often far out of step with reality
How do you get from "people may not be reliable witnesses of their own preferences" to "I know their preferences better"?
The first doesn't sound that improbable, but the second is a non sequitur, even if a popular one. If it's easy to lie to yourself, doesn't that suggest it's even easier to lie to yourself about what others think?
My previous comment suggested multiple reasons why people might not choose the chronological ordering even if they prefer it. Most people surely aren't aware of the trick, and using it limits the number of posts available. It's not clear that it shows all of them within a range either. I myself only use it about half the time, not because I don't want things in order, but because I have the sense that I'm missing something either way.
>I think people would find Facebook less engaging and fun to use if it didn’t tune what it showed them based on their perceived interests and their activity
I think an important point is that "engaging" is not at all the same thing as "fun to use". Meth is engaging, but in the long run, not that fun. The conflation of the two is useful for apologetics but not necessarily convincing to a lay individual.
>it wouldn’t be ideal to be put in a position to have to remove family members because their posts are boring
"Boring" sounds like a different sort of thing from "spam, porn, and shock videos". That kind of post does sound more plausible.
But Facebook provides controls to filter people if you want, and there is no reason that if they stopped choosing what you see that the user's controls would have to go away.
What is the rational argument that Facebook knows which family members of nearly 3 billion people are "boring", though?
The rational argument is pretty simple. They have data about what things you click on, comment on, react to, etc., and use these to show you more similar things and show you fewer things you just scroll past. I'm not sure how you argue against what seems like pretty objective data.
>I'm not sure how you argue against what seems like pretty objective data.
I commend you for crystallizing the argument which I think is indeed representative of what the industry believes about itself and many others reject.
Virtually everything in computing, or the real world, for that matter, is solving some sort of optimization problem, explicitly or implicitly.
I'd paraphrase your comments as, Facebook solves an optimization problem, they make billions of dollars doing it, so it is objectively the optimum.
There are two objections I think that capture most of the opposition.
- Real optimizers can't promise a true optimum every time; they sometimes converge on a local optimum. This local optimum may not be within any particular distance from the global optimum either.
- "Fewer things you just scroll past" is not necessarily what the *user* wants to optimize. It is not in fact what I want to optimize. I'm not sure if I made it clear, but the *existence* of things that I don't actually read is important contextual information. So is the relative order, the timing, the source...
These points are really abstract and general - virtually anything can have these two issues - they're not optimizing the right thing, and they're not finding the global optimum.
I think non technical people sometimes get angry because they intuitively feel that people claim objectivity without justification, but I think it's possible to provide a bloodless, abstract, logical, and specific critique more suitable for a software engineer's mentality, and that's what I tried to do above.
[One other general issue I thought of - optimizers can have the wrong constraints; everything in life normally has rules and limits on how you can pursue a goal. The ends don't necessarily justify the means. So this is another thing that is not covered by simply saying the optimizer produces the optimum]
How do you get from "people may not be reliable witnesses of their own preferences" to "I know their preferences better"?
The first doesn't sound that improbable, but the second is a non sequitur, even if a popular one. If it's easy to lie to yourself, doesn't that suggest it's even easier to lie to yourself about what others think?
My previous comment suggested multiple reasons why people might not choose the chronological ordering even if they prefer it. Most people surely aren't aware of the trick, and using it limits the number of posts available. It's not clear that it shows all of them within a range either. I myself only use it about half the time, not because I don't want things in order, but because I have the sense that I'm missing something either way.
>I think people would find Facebook less engaging and fun to use if it didn’t tune what it showed them based on their perceived interests and their activity
I think an important point is that "engaging" is not at all the same thing as "fun to use". Meth is engaging, but in the long run, not that fun. The conflation of the two is useful for apologetics but not necessarily convincing to a lay individual.
>it wouldn’t be ideal to be put in a position to have to remove family members because their posts are boring
"Boring" sounds like a different sort of thing from "spam, porn, and shock videos". That kind of post does sound more plausible.
But Facebook provides controls to filter people if you want, and there is no reason that if they stopped choosing what you see that the user's controls would have to go away.
What is the rational argument that Facebook knows which family members of nearly 3 billion people are "boring", though?