I personally think you are in for a very rude awakening. This is exactly what people are trying not to do with Machine Learning. It is always, almost, about profiling you, about finding out who you are in order to offer you what you already know, to limit your horizon, to "tailor" your "experience" and the span of possibilities.
As a very technical fellow told me: "when you read three political news in a row we set a badge, you are now politically minded."
This, sadly, doesn't fit your yearning for us growing "beyond changing ourselves to fit the algorithm."
My prediction is that it will get worst before it gets any better at all.
The first thing we'd probably do with cuttlefish suits anyways is stick ads all over them. Then the company that makes them would get acquired
Of course it will be abused, and it will be abused from the outset. The powers that be already define hate speech vs dissent, propaganda vs news, terrorists vs. freedom fighters, scientism vs science, et cetera.
Big Tech and Western governments together control of the flow of information. Previously, the blue-flavored regime held that control and abused it terribly. Now, the red-flavored regime has it, and people are finally noticing. God help us if people far, far worse gain that control in the future and start defining what is unacceptable speech.
This is an easy target. Whenever you get that argument you go all: "it's a bit unsettling that you find censor-happy human-out-of-the-loop algorimhms good, but I thank you for you honesty."
Then you recall that you are a programmer: "you know what, I should definitely program an algorithm that censors all positive talk about censor-happy human-out-of-the-loop algorithms. Thanks for the idea."
The corollary is 30 ton tractor trailer changing lanes on the highway, and concluding that there's no traffic because it would prefer not to see it.
Censoring the rear view mirrors is a bad idea no matter how much you hate what you see in them.
Facebook was an effective rear view mirror, for a whole lot of us. The more savvy among us could see the storm clouds gathering on the horizon literally almost since the beginning , but certainly by late 2014/early 2015 it was plainly obvious that psyops were afoot, though even to a lot of plugged in hackernews type people the full-blown industrial scale of it all wasn't immediately obvious.
That's whats so insidious. It's the model. It's the first person perspective that facebook locks you into. It looks like reality, and it contains people you know, but it is not real.
You don't really know who wrote what, who liked what, etc. There's no global view of facebook for a user. I think that's the problem.
The first person view that you're locked into with facebook, makes it trivial to manufacture the appearance of group consensus when no such thing exists in the real world (if you have access to the backend).
The problem really is facebook, not the concept of social media all together. The problem is also that Facebook knew what it was doing. They were building a political doomsday machine from the get-go, they knew that's what they were doing, and they designed specifically to that goal for a decade with practically limitless funding.
Perhaps was my comment too hypothetical? :]
Thankfully, this sort of thing is exactly what the Internet was designed for ..
Senator Grassley also said yesterday at the very end of the Zuckerberg hearing that he hopes someone with Zuck's influence and other such platforms would do something to "reduce the cynicism people have towards Congress."
That sounded pretty scary. I get that the mainstream media (and more recently Facebook propaganda, but this is definitely a much older issue) is partly to blame for polarizing America, but asking companies to actively try to influence how people think about their governments sounds like the wrong direction to take to me. I also don't think it's any better than say Fox News trying to polarize people.
https://www.judiciary.senate.gov/meetings/facebook-social-me...
This is what many don't understand, or don't want to understand, this is why some big players are pissed at Facebook, very very pissed.
She was supposed to win, they had to help her win.
https://wikileaks.org/podesta-emails/emailid/56638#efmAAGAA2...
https://wikileaks.org/podesta-emails/emailid/19070
https://wikileaks.org/podesta-emails/emailid/15092
---
From:[email protected] To: [email protected] Date: 2016-01-02 22:07 Subject: RE: Happy New Year
I am thrilled at the progress Hillary is making.
[...]
2015 was challenging, but we ended in a good place thanks to your help and support.
Look forward to working with you to elect the first woman President of the United States.
Have a great New Year.
-John
I was hoping that when they started realizing the potential, they'd feel threatened and require it to cease working that way. Instead they all run PR campaigns on social media, and they probably collect data on their voters too if they can. It's just too disgusting to bear. Not just the US either, this is happening all across Europe.
At first I stopped in from time to time to get my outrage on (I never interact with them, just read) and to feel smug that I am intellectually superior to them and their followers.
But then I realized that if way back, at the very beginning of your thought process you start from a false premise, everything else can follow reasonable and logical methods. It's almost like chaos theory, small differences in the starting position can have vastly outsized differences at the end.
I've used this experience to go back and look at my "starting" positions. And I've horrified myself too. So on the whole it's been an enlightening experience.
So now I can say that I am firmly on my way to understanding them now (and myself too) even though I diagram with them vehemently.
I don't use Facebook much, but if I did, I would want to be able to tell the algorithm "take 50% of what you want to show me based on my preferences and show me the exact opposite"
The problem is that your definition of bullshit might be different than mine, so any filter you come up with will have issues.
If you think you've got yourself into a situation where you're immune to the filter bubble, that just means you've created one so comfortable that you never get prodded by anything.
For me this is where these YT tropes get interesting. When you get past the thumbnails of the mega channels with multiple full time employees who are obviously doing something very baity and strategic, you’re left with the ordinary channels who mimic these tropes likely without deep consideration or even consciousness.
It’s just more of a will to survive kind of thing, and thus an entire ecosystem shapes its own desires by way of unknowingly optimizing for an algorithm which itself is based on honing in on community trends.
Can we not design a better fitness function?
People do write replies, but I feel the conversation is much terse these days. I enjoy HN because in general the crowd here is more mature and willing to invest in an informative thoughtout reply instead of just witty one-liners for karma (which I'm glad people here downvote).
But yeah, I wonder what if we took away the "likes". We had an internet and community without them before.
PS Thanks for the book recommendation.
I'm not sure how recent that phenomenon actually is. I think it's more that the social tools where people congregate made it easier and even incentivize responses.
A larger number of people with less to say will respond because these systems are designed to cultivate response.
You can ignore prime top level directives only for so long.
This is an interesting thought. It does simplify expression, but in sum I don't think it's that much different than life before the like button.
We press the like button all the time in real life. Giving thumbs up, a quick nod of the head, or clapping.
From that standpoint, the like button is just the digital manifestation of how we showed engagement with "content".
Where I see we run into problems is that a "like" has different meanings depending on the person and context. Unfortunately algorithms aren't able to fully understand these nuances yet, so we end up with the same stream of youtube videos in our feed or the same 5 songs in our pandora playlist. Which means on the backend we're stuck with a certain set of "almost-but-not-quite-right" labels in a system which will then make "almost-but-not-quite-right" decisions.
Folks at the edges may get incorrectly lumped into the wrong group and, in milder cases, receive content they aren't interested in, in extreme cases, be flagged as a "potential terrorist".
It's tough. I do think my life is generally better off with intelligent algos removing friction, aiding in discovery, and anticipating my needs. But to do so, those same algos need an almost infinite amount of info about me to be able to infer context and nuance.
> We press the like button all the time in real life. Giving thumbs up, a quick nod of the head, or clapping.
Oh no, I don't think real life expressions can be summed or approximated into a binary thumbs up vs no thumbs up (there's no thumbs down). For example, we express a lot more by how eagerly we raise our hands while giving a thumbs up, or how our eyes approve (or disapprove) while giving the quick nod, or how hard (or sarcastically mild) we clap.
I think there is always an amount of uncertainty in real life expression that gets lost the moment we transform them into certain form.
My point is more about the way people use the like button has corresponding actions in "real life".
A thumbs up can be used to convey a variety of emotions. Happiness, irony, and other meanings depending on culture.
However on FB, the like button has an implied set of narrow usage. You use it to acknowledge, approve, like things. It's not as useful as in real-life because you can't as easily interpret context or add accompanying signals like a furrowed brow or a smile.
The usage set is more narrow/different and I think most people recognize that when they give a thumbs up on FB vs. when they do so in real life.
I personally have been noticing more and more how technology and algorithms seem to bounding expression online and off, particularly around content creation, but also with influencers and personal branding and so on. It feels like 90% of the shit being pumped out is super amplified to draw notice or for signaling, and while this is not a new phenomenon of course, it still feels weird to see ordinary people at restaurants taking pictures of their food, or sucking up to some corporation online for the chance to win a prize
I feel this pressure too. The pressure to get noticed. To tailor content to what will rack of the most views. To post that tweet that will make me look like a smart, funny guy who has something valuable to say and oh by the way you should totally follow me. And I hate it. And I try not to care about numbers. But it still feels really shitty when something you spent a month working on and genuinely find really cool gets a total of ten views while a picture of a goat in a sweater just when viral. And what if the audience is right?
There has to be something better. I hope that we can grow beyond changing ourselves to fit the algorithm, and re-focus on using technology to create new ways of sharing and expressing and relating to each other. One fun example of this that Lanier gave: creating technology that would let people communicate like cuttlefish (specifically the shifting skin colors and textures). That optimistic vision still feels as radical and fresh today as it did when I first read the book in 2010. I just hope it's not too late