SoylentNews is people
SoylentNews
For a few years now, alarms have been sounded in various quarters about Facebook's teen problem. In 2013, one author explored why teens are tiring of Facebook, and according to Time, more than 11 million young people have fled Facebook since 2011. But many of these articles theorized that teens were moving instead to Instagram (a Facebook-owned property) and other social media platforms. In other words, teen flight was a Facebook problem, not a social media problem.
Today, however, the newest data increasingly support the idea that young people are actually transitioning out of using what we might term broadcast social media – like Facebook and Twitter – and switching instead to using narrowcast tools – like Messenger or Snapchat. Instead of posting generic and sanitized updates for all to see, they are sharing their transient goofy selfies and blow-by-blow descriptions of class with only their closest friends. [...]
- As social media usage has spread beyond the young, social media have become less attractive to young people.
- Many of the students I've spoken with avoid posting on sites like Facebook because, to quote one student, "Those pics are there forever!" Having grown up with these platforms, college students are well aware that nothing posted on Facebook is ever truly forgotten, and they are increasingly wary of the implications.
- Increasingly, young people are being warned that future employers, college admissions departments and even banks will use their social media profiles to form assessments. In response, many of them seem to be using social media more strategically.
(Score: 2) by Non Sequor on Tuesday February 09 2016, @04:17PM
I'm less interested in the particular faults of particular networks than I am interested in the scaling properties of any theoretical communications network.
Building a very broad network with a low cost of sharing information may be hitting a saturation point. The implications of a public face being a liability are becoming apparent once it reaches a certain threshold.
I can't remember the name of the phenomenon but there is a typical upper bound on the number of relationships people have. Does exceeding that limit have game theory implications that paralyze people's range of actions? Is this a human limit or a constraint on systems of multiple intelligent actors?
Write your congressman. Tell him he sucks.
(Score: 2) by VLM on Tuesday February 09 2016, @04:44PM
Google doesn't find anything at the intersection of social media and Shannon's Law from the world of telecommunications, but I wonder if we have enough pre-collapse pre-threshold data to try and fit Shannon's Law to various social network observations.
You could assume the amount of signal is constant. The recent train crash in Germany is a constant aside from koan commentary about trees falling in forests. But the number of people murmuring meaningless stuff about it is a noise component that increases with the size of your social media network. Therefore the SNR varies with network size. Now Shannon sets a physics based limit on how much "real information bits" can flow given a constant BW and varying SNR and varying bit error rate. I wonder if the observed limit is anywhere near the physics based limit, or if there's a shadow of the physics based limit. Anyway lowering SNR by 10 dB by having a social network 10x bigger should increase the error rate about the recent train accident by a calculable estimable ratio at a physics based limit. So the bigger your social media network, the dumber you would be about current events. Real world may not operate anywhere near that limit, of course and if it doesn't it may or may not shadow the physics based minimums as laid down by Shannon. It could be an interesting paper, I'm surprised there's apparently nothing out there.
I would imagine there's various EE / control theory type models too. So you have a PID controller of mood or sanity or WTF and as the noise input increases via social media, the output of the controller will destabilize in a surprisingly predictable manner, at least at a large scale. Maybe the PID mood controller is depression or people acting violently. You could see how close the observed real world is to the theoretical world of control theory based predictions. Experimentally verifiable predictions like 10% more noise input from social media traffic graphs results in x% more instability so the amount of psychoactive drugs prescribed should increase x%. Might be wrong, but at least it would be testable predictions.
(Score: 2) by Non Sequor on Wednesday February 10 2016, @03:42AM
There are some fun ideas here.
I was thinking in terms of the traffic jam model brought up recently on SN where the low traffic state is like a fluid, and a traffic jam is like a transition to a crystalline solid with cars being packed tightly enough that they constrain each other into rigid motion.
If you apply the noise model you mentioned to some sort of mesh network, you have a similar state transition when you create some form of over saturation. I only know enough here to assert the similarity of the situations, not enough to convert that into a prediction though.
I'd also posit two distinct saturation effects: (1) as you mentioned limits on signal propagation due to noise and (2) a computational capacity constraint on nodes that are also doing some quantity of work unrelated to the network (day job, etc.). Actually, I'll equate (2) with your mood/sanity/stress element. Each node has an internal priority queue and at some saturation point it has to start dropping jobs or exhibiting some other failure behavior.
There is a loose equivalent of a PID controller model in economics called a dynamic stochastic general equilibrium model. In these models, you posit interaction rules for a set of actors so that each actor maximizes their utility, and that gives you differential equations for the composite system. Basically, your individual actors have responses to an error condition (P controllers). The trouble is that these models are notoriously useless for quantitative predictions. Basically they end up being used as props to illustrate a hypothesized qualitative effect. They also typically need to be naive in some portion of their specification in order to be tractable.
Basically, those models only really give weak predictions like saying that if people save 0% of their income or 100% of their income the economy will run off the side of the road into a ditch. The predictions associated with a particular savings rate or an estimate of an equilibrium savings rate are mostly useless.
Over the short term, I expect any model comparing what a social network is to what a social network does to be overwhelmed by noise in confounding variables. Right now social networks are still in a growth phase and growth among some demographic groups may have peaked while others are just ramping up. This heterogeneity would be likely to frustrate any kind of model fitting even if you have inputs and outputs that are easy to observe.
But I think I do have a qualitative prediction: Facebook's game plan of eventually being a single sign on and conduit for online comings and goings of a super majority of the world's population isn't going to pan out and social networks will stabilize into some other model than a single dominant player. Possibilities include a moderately stable oligopoly with some ebb and flow in participation or distinct niches, or a birth and death model where individual social networks grow, sometimes to large size, but ultimately eventually shrink through migrations to newer platforms which are on an upswing.
I could also be wrong and there may be iterative ways of restructuring a social network that avoid any kind of saturation effects.
Write your congressman. Tell him he sucks.
(Score: 2) by HiThere on Wednesday February 10 2016, @04:15AM
I can't remember the name either, but IIRC the limit averages at around 50, is differs significantly between people.
OTOH, there are certain assumptions buit into the calculation. Email lists expand the number of people you can have shallow, but meaningful contact with, e.g. So it probably doesn't apply strictly when you start including on-line communications, as a part of the figure was the amount of effort required to maintain the relationship.
There is still going to be a limit out there, and the original limit was stepped, with fewer non-family intimate contacts and more contacts of a more distant nature. So the actual figures aren't going to be correct when applied to on-line contacts, but the general idea will still apply. You can ask for different levels of help from different degrees of intimacy, and you will provide differing amounts of help when asked. (As such, you might consider how much help a "Facebook friend" could expect to get from you in response to a request for help.) Also note that family contacts are figured differently. "Blood is thicker than water.", etc. You don't need to put continuing effort into maintaining the existence of a family relationship, though degree of support will, clearly, vary depending on how well and how much you get along.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.