> "Find people who hold the most influence, typically those who sit at the center of a social network"
This is an assumption - and I'd be interested to see what gets found when it is tested. Influencers are critical, but are they detectable in the degree of a social network graph?
I know a couple of people whos opinions would cause me to change my actions. I don't know if I'm friends with them on a social network, which I mainly use to keep in contact with close friends. The highly connected nodes in a social network could very easily be people who are the social equivalent of town criers rather than the social equivalent of the town mayor, or local matriarch/patriarch, who wield actual social power and validate or invalidate "what everyone knows". In that case, seeding them preferentially isn't all that different from seeding random people.
Societies can do an exceptional job of hiding who actually has final say on what are and are not the valid opinions. They are not the people who talk the loudest, or most frequently.
This is an interesting point. I know what my network topology looks like because I found out back when Wolfram Alpha was able to key into the Facebook API. I'm at the periphery of many networks, but among my close friends a couple of them are super-connectors within mutual networks.
So on the one hand, one might expect that I have influence just due to network position; but really, I've just been talking to the people I figure would know the most among these networks' topics.
Social media is really wild right now. Twitter verified symbols are almost exactly Dr. Seuss’s star belly sneetches. [1]
I’m sure with a bit more effort one could compile a long list of social behaviors warned against by literature that social media sites directly implement.
I don’t think targeting a single influencer is really that valuable, but targeting a platform seems to work really well.
DXracer chairs have heavily targeted twitch streamers and I think that’s the main reason behind their success, because they are really expensive and uncomfortable, yet every teenage room of every colleague of mine houses one.
Of course that’s completely anecdotal, I just couldn’t help laughing at my own daughter when she bought hers. Because it’s a really horrible product with an insane price tag that only sells because everyone has one.
For a different anecdote on the same topic, I had no idea where they came from until seemingly in a matter of < 6 months they were in stores everywhere. I tried one and found it quite comfortable - feels like a much more reasonably priced Aeron (and I'm not so happy about Aeron's nontransferrable warranty games).
Later on I saw one in a Twitch video and I just figured they also found it a nice chair. Funny that the directionality may have been reversed!
DXRacer are actually pretty decent chairs for long periods at a desk. Pretty much the only (known) product between cheap but shitty office depot chairs and great but expensive aeon chairs. It’s not all marketing—mostly maybe but not all.
Simply seeding a few more people at random avoids the challenge of mapping a network’s contours and can spread information in a way that is essentially indistinguishable from cases involving careful analysis; seeding seven people randomly may result in roughly the same reach as seeding five people optimally.
Interesting and I like it, though it conflicts some with previous studies I have seen where introducing new concepts through someone low in pecking order results in dramatically slower spread than if it starts with someone high in pecking order.
Also, there is a lot more to the article than that one point.
No, sorry. Stuff I read long ago and a quick Google is not turning it up.
One study I recall seeing introduced something like gum drops to primate groups. If the alpha male was given them first and liked them, this became the hot new trend for the entire group in just two hours. When introduced via the lowest ranking member, after two years it had only gotten to about half the group.
I've read a bunch of stuff over the years and enjoyed the writings of Frans de Waal, among others.
Cool study. Ant hills don't have influencers. They are still able to run billion individual colonies. The more connected we get the more we resemble an ant hill imho in terms of info dispersion.
Feels like its also why news/journalist hubs aren't required as they once were. Other than for attention mining which seems like a unsustainable temporary glitch in the system.
It'd be silly to assume that influencers are a necessary requirement for information to spread in a network, just because that's how it's worked a lot of the time in our society.
Nice piece of work. The role of influencers in this study is akin to that of hubs in scale-free network. Diffusion on scale-free and small-world network have been studied extensively in physics community. It will be interesting to compare the results from this paper with work on scale-free networks.
> Nice piece of work. The role of influencers in this study is akin to that of hubs in scale-free network. Diffusion on scale-free and small-world network have been studied extensively in physics community. It will be interesting to compare the results from this paper with work on scale-free networks.
Also, context is everything. A social graph is not a flat structure. This work, for instance, https://arxiv.org/abs/1507.03826, shows that how consensus is formed and spreads across an agent population depends heavily on how different network types overlap.
Interesting study, but this totally falls under Betteridge's law of headlines. No, influencers aren't overrated.
There is no way in hell I will get the message across as effectively as if I have an influencer spread it. This should not even be a question. If I have a story and I tell it to 7 people at random and convince them to spread it vs. convince Paul Graham to spread it - no contest.
We just did a test with one influencer (6M followers on Twitter) to promote our products - bang for buck way better than advertising.
The connection asymmetry (good influencer has magnitudes more connection) and trust/influence (I trust/believe information coming from PG) are important. It is however hard work finding good influencers, and just having a lot of followers does not mean a thing if they are not engaged etc... But worth it.
Mind sharing some details? Did you do it through a middleman as part of an ongoing relationship or just was it a direct one-off? What's the approximate price range? How elaborate was the advertising itself (as in product placement, sponsored tweet, a combination of these etc)?
This is an assumption - and I'd be interested to see what gets found when it is tested. Influencers are critical, but are they detectable in the degree of a social network graph?
I know a couple of people whos opinions would cause me to change my actions. I don't know if I'm friends with them on a social network, which I mainly use to keep in contact with close friends. The highly connected nodes in a social network could very easily be people who are the social equivalent of town criers rather than the social equivalent of the town mayor, or local matriarch/patriarch, who wield actual social power and validate or invalidate "what everyone knows". In that case, seeding them preferentially isn't all that different from seeding random people.
Societies can do an exceptional job of hiding who actually has final say on what are and are not the valid opinions. They are not the people who talk the loudest, or most frequently.