I'd dispute the notion that Facebook is optimizing directionlessly. It just turns out when your entire business model is "sell ads", then profitability and popularity become synonymous.
I'm very curious if you have a proposal for how to gauge content quality at scale. Back when I majored in journalism, "community engagement" was a massive topic. People wanted to create content that wasn't just "good" in a vacuum, but actually motivated society towards positive change. The theory went that if we reported on social issues in a certain way, perhaps we could influence society for the better. This was even the topic of my capstone project.
In the end, I became disillusioned. My observations contradicted the notion that content quality (news content in my case) matters in either a commercial or social sense when conducted at scale.
In my capstone project, some people volunteered to be interviewed about their volunteer work and their motivations. It was really interesting talking to all of them, but what I found in every case was that these people were motivated by life experiences, not media. I talked to an ex-con who had his life turned around by a books program, so he volunteered with the program when he got out. I also talked to a soccer mom who organized community events for her kids.
This was admittedly a qualitative study on personal anecdotes with a small sample set so it's not hard science. But the hypothesis that content quality is irrelevant at a mass scale seems to hold up in practice.
This also tangentially ties into a hypothesis I have on US presidential elections, which is that once it's down to two candidates, name recognition is all that matters. Every time (I'm aware of) that a president dominated the mass media of the time, he won the election. From FDR on radio to Kennedy on TV to Obama on social media and now Trump who rode on the public obsession with gossiping about every stupid thing he did or said.
Okay, rant over. I am legitimately interested, though, in any ideas for determining content quality at scale just because it seems like an interesting project.
I'm not sure, really. I assume that it would probably be mostly "theory-less" in practice. Google built ways of accessing web page quality. This seems somewhat related. pagerank? amoorthy's project seems to be citation-based. That sounds like a good start, I think.
From what I hear, FB are now pretty worried about fake news and other issues relating to the last US elections. I suspect they're also going over a few other big political events where FB played a big role. After seeing it happen at home, it probably feels more real. Brexit. The Syrian protests, later civil war. ISIS. The Turkish coup attempt. FB played a bigger role than any other media in all these.
I suspect what they are doing about true fake news is probably spam filtering.
What I'd be interested in seeing them try is the stack overflow approach. Have a subjective opinion about what a good contribution to FB is. Design features to promote that kind of content.
We've written an algorithm that scores news for quality. It's a work-in-progress but you may be interested in checking out our chrome extension at http://www.civikowl.com/
>This also tangentially ties into a hypothesis I have on US presidential elections, which is that once it's down to two candidates, name recognition is all that matters. Every time (I'm aware of) that a president dominated the mass media of the time, he won the election. From FDR on radio to Kennedy on TV to Obama on social media and now Trump who rode on the public obsession with gossiping about every stupid thing he did or said.
Makes more sense than blaming Russia and Facebook.
Why would you think the two thesis cannot work together as force-multipliers? If name recognition matters, then why can't Russian troll farms influence an election by hacking what names you recognize most using Facebook?
Because occam's razor says that the mundane reasons (terrible Democratic candidate, poor economy, Trump's name brand value) are more plausible than sinister conspiracy theories involving unimpressive ad budgets pushed by media orgs who just happen to be large DNC donors.
>Terrible democratic candidate translates to "she's a woman"
No, it translates to:
* Obvious corporate shill and corrupt beltway insider - made even more obvious in light of the content of the leaked emails.
* A candidate who had a chequered political career that included dog whistling racists - an event that Trump made massive political capital out of and used to get large portions of a base she was relying upon - black voters - to stay at home.
* A candidate whose approach was to present herself of defender of the status quo at a time of obvious economic distress for most Americans which has never worked, ever.
* And yes, warmonger (that much was obvious from her term as secretary of state)
* A candidate whose strategy to get people to vote for her was to hypocritically either imply or outright state that you were sexist or racist if you considered voting for the other guy. Funnily enough that backfired massively but people (like you) are still flogging that dead horse in what I can only assume is an errant attempt to ensure that Trump wins a second term in office.
The really puzzling part of this whole thing is the number of people on Hacker News who will argue out of one side of their mouth that Facebook ads area a scam that simply don't work and on the other hand that spending what is essentially political pocket change on them will buy you a presidency. Crazy.
Defender of the status quo? Her voting record shows otherwise(1). She defended those who needed it most, going back far longer than her Senate tenure. My daughter's life was saved by a clinic that would not have existed but for the rural health program in Arkansas she championed.
So what you consider quality is probably different that what the (mode) average person considers quality. Perhaps the issue is joe sixpacs are the bulk of the bell curve.
Compounding this is most people are also information-overloaded anyway and don't have time for quality. Emotional appeals and manipulation on the other hand are hard to ignore even if you don't really have the time or desire to engage with them ..
I'm very curious if you have a proposal for how to gauge content quality at scale. Back when I majored in journalism, "community engagement" was a massive topic. People wanted to create content that wasn't just "good" in a vacuum, but actually motivated society towards positive change. The theory went that if we reported on social issues in a certain way, perhaps we could influence society for the better. This was even the topic of my capstone project.
In the end, I became disillusioned. My observations contradicted the notion that content quality (news content in my case) matters in either a commercial or social sense when conducted at scale.
In my capstone project, some people volunteered to be interviewed about their volunteer work and their motivations. It was really interesting talking to all of them, but what I found in every case was that these people were motivated by life experiences, not media. I talked to an ex-con who had his life turned around by a books program, so he volunteered with the program when he got out. I also talked to a soccer mom who organized community events for her kids.
This was admittedly a qualitative study on personal anecdotes with a small sample set so it's not hard science. But the hypothesis that content quality is irrelevant at a mass scale seems to hold up in practice.
This also tangentially ties into a hypothesis I have on US presidential elections, which is that once it's down to two candidates, name recognition is all that matters. Every time (I'm aware of) that a president dominated the mass media of the time, he won the election. From FDR on radio to Kennedy on TV to Obama on social media and now Trump who rode on the public obsession with gossiping about every stupid thing he did or said.
Okay, rant over. I am legitimately interested, though, in any ideas for determining content quality at scale just because it seems like an interesting project.