Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Friday February 23 2018, @03:38PM   Printer-friendly

The Columbia Journalism Review has some analysis of the problem of disinformation and propaganda being actively spread over social control media. As the situation is studied more, albeit belatedly, the nature of social control's business model gets more daylight.

"That fundamental goal is to get the user to stay as long as possible," Ghosh said in an interview. "Their motivations are different—for platforms, it is to maximize ad space, to collect more information about the individual, and to rake in more dollars; and for the disinformation operator, the motive is the political persuasion of the individual to make a certain decision. But until we change that alignment, we are not going to solve the problem of disinformation on these platforms."

After Mueller released his indictments, sociologist Zeynep Tufekci noted on Twitter that the indictment "shows [Russia] used social media just like any other advertiser/influencer. They used the platforms as they were designed to be used."

The phrase surveillance capitalism gets more traction as it becomes acknowledged that while social control media do not actively spread disinformation and propaganda it is a side effect of collecting as much personal information as legally (and somtimes illegally) allowed. That information is aggregated from multiple sources both internal and external to social control media itself. As a result it is getting increasingly difficult to distinguish between disinformation and authentic political speech.

Automated attacks make that differentiation that much harder. Faecebook gets the most attention, but the others, including YouTube work the same way and can thus be manipulated just as easily. (Ed: Speaking of YouTube, to single out one topic as an example, as seen recently with FCC comments on Net Neutrality, only 17%of the comments the FCC received were legitimate with the rest filled in by clumsy bots.)

Source : Fake news is part of a bigger problem: automated propaganda


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 0) by Anonymous Coward on Friday February 23 2018, @06:14PM (9 children)

    by Anonymous Coward on Friday February 23 2018, @06:14PM (#642515)

    And most people never learn critical thinking skills. Correlation or causation?

  • (Score: 3, Disagree) by HiThere on Friday February 23 2018, @06:32PM (8 children)

    by HiThere (866) Subscriber Badge on Friday February 23 2018, @06:32PM (#642529) Journal

    And most people never learn proper analysis.

    Correlation is not causation, but it is often a sign of causation. The "Correlation is not causation" is an appropriate warning because occasionally it isn't. But most of the time it is, though you can't always be certain of the direction of causality without more information. The times when there is no causal connection, direct or indirect, are rather rare. They do happen, which is why the warning is necessary, but they are rare. Causation is usually present if there is correlation. Often, admittedly, it that both are caused by some third factor or group of factors that wasn't considered, but it's almost always present.

    --
    Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
    • (Score: 0) by Anonymous Coward on Friday February 23 2018, @07:03PM (1 child)

      by Anonymous Coward on Friday February 23 2018, @07:03PM (#642557)

      Correlation is causation when you folks want it to be.

      There aren't enough women programmers? Conclusion: every last assigned male who does programming must be a sexist. And a homosexual. And they should be ashamed of being homosexuals. And programming can be mastered in under an hour.

      Fuck you. I am so fucking done with you liberals.

      • (Score: 0) by Anonymous Coward on Friday February 23 2018, @07:22PM

        by Anonymous Coward on Friday February 23 2018, @07:22PM (#642572)

        You need help brah, like professional therapy levels. You're spouting off on a thread that had zero political bias. Not a single mention of liberal, conservative, dem or rep. Get help.

    • (Score: 0) by Anonymous Coward on Friday February 23 2018, @07:30PM

      by Anonymous Coward on Friday February 23 2018, @07:30PM (#642578)

      I'm not sure why you felt the need to clarify that, was it some sort of rebuttal?? Or just a public service announcement cause you think too few people understand what "correlation or causation" meant?

    • (Score: 3, Insightful) by Thexalon on Friday February 23 2018, @07:35PM

      by Thexalon (636) on Friday February 23 2018, @07:35PM (#642582)

      Correlation is not causation, but it is often a sign of causation.

      The correct conclusions here are quite complicated, actually:
      1. If A causes B, and there exists no C that causes not-B in roughly equal amounts to A's effect, then there will be a correlation between A and B.
      2. Ergo, by the contrapositive, either there exists a correlation between A and B, or there exists some C that causes and equal amount of not-B, or A does not cause B.

      The "correlation is not causation" line is used to thwart the bad reasoning from the converse of point 1 above that "A correlates to B" necessarily implies that "A causes B". A correlation is certainly valuable information on the question of whether A causes B, but it's in no way definitive in either direction: If there is a correlation, it could be that A causes B, that B causes A, that some unknown factor D causes both A and B, or a complete coincidence that doesn't hold up on repeated testing. And if there isn't a correlation, as stated above that could mean A and B aren't related, or it could mean there's some unknown factor C undoing the effects of A.

      As an example of this kind of stumbling around with correlations rather than clear causes:
      1. You observe that rocks are lying on the ground rather than, as a general rule, flying through the air. A ha, you think: If you want to find a rock, looking on the ground is your best bet. This conclusion is so obvious that cavemen, monkeys, and small children figure it out fairly quickly. Good job! That correlation was extremely useful.

      2. You also observe that rocks that are in the air head towards the ground. OK, so now we even have a cause for the phenomenon of rocks being on the ground: All rocks are trying to get to the ground level, what we humans refer to as "down". Aristotle will write about this, and for centuries many people will believe that the cause of these phenomena were that rocks were very obviously made of the earth element, and things made of the earth element are trying to return to the earth, whereas things that don't fall to the ground in the same way from the air (e.g. birds) are clearly containing the air element and thus are trying to stay in the air, or the fire element and are thus fleeing upwards towards where they can find other fire-like things like the sun and the stars. Seriously. If you read Aristotle's stuff, and ignore what you know about Newtonian physics, it makes a remarkable amount of sense, which is why so many believed it for so long.

      3. You later get to more advanced thinking, and notice that: Some rocks are under the ground, which means that the earth that these rocks are trying to get to is clearly lower than the ground level. But at the same time not all rocks end up buried underneath the ground. Well, clearly, something is in the way stopping rocks from getting down to this level below the ground where the earth-element things are trying to get to. But for some reason, water poured on the ground seems to get right through no trouble. So from that, you start drawing further conclusions: Things with the earth element in them like rocks can't change their shape. Things made of the water element can, but can only do so to move downwards. And things with the air element, like steam and smoke, are clearly incredibly flexible and can change their shape and move every which-way. Things made of the fire element are constantly pulling upwards, and you can sometimes see the fire escaping that direction (and again, steam and smoke tend to move upwards, so they must be part-fire). You're still wrong, but this is starting to get towards the concepts of "solid", "liquid", and "gas".

      And all this is why science is hard, and in general we don't know as much as we think we know.

      --
      The only thing that stops a bad guy with a compiler is a good guy with a compiler.
    • (Score: 3, Informative) by AthanasiusKircher on Saturday February 24 2018, @01:17AM

      by AthanasiusKircher (5291) on Saturday February 24 2018, @01:17AM (#642795) Journal

      Correlation is not causation, but it is often a sign of causation. The "Correlation is not causation" is an appropriate warning because occasionally it isn't. But most of the time it is, though you can't always be certain of the direction of causality without more information. The times when there is no causal connection, direct or indirect, are rather rare. They do happen, which is why the warning is necessary, but they are rare. Causation is usually present if there is correlation.

      This is ABSOLUTELY false. And it can be statistically proven to be not only false, but a gross mischaracterization of reality.

      You can "correlate" anything with anything. I could go into a database of random demographic stat A and start comparing it to random stats B, C, D, and E. I can try out subpopulations and various subsets. I can mix and match. I can likely create trillions of potential pairwise variables to compare from such a set of data. And among them, I will undoubtedly find a large number of very highly "correlated" things, according to standard statistical metrics.

      There's a guy who made a blog [tylervigen.com] and a book using this very methodology.**

      There are an incredibly large number of potential variables in the universe that you could start comparing to other variables. And inevitably you will find things that display a ridiculously high amount of "correlation" within random choices of datasets. And the vast majority of said correlations will be meaningless and non-causal in nature.

      The unmentioned assumption in your statements is that generally when scientists or experiments start looking for correlations, they are choosing things that they feasibly think could have a correlation. That is, there's some logical connection, which would serve as the basis for causation. But the vast majority of potential pairwise variables in the universe have no such reasonable basis for connection (unless, I suppose, if you are a Jungian), and thus the vast majority of potential correlations that could exist are non-causal in nature.

      The larger point here is the problems that come about in figuring out what are reasonable assumptions about possible connections. There are possible causal connections we don't tend to imagine, because we haven't seen evidence of them before. And there are "obvious" candidates for causal connections that turn out not to be so, but we can be misled by confirmation bias in data collection, analysis, and apparent "correlations." Science has the difficult task of trying to figure out the reasonable assumptions for coming up with potential causal connections and then verifying them.

      ----
      ** Note: Yes, the author of that blog "cheats" by deliberately misusing statistics. He arbitrarily selects variables, time windows, etc. that show the best correlations. But if you enlarge the perspective to all potential measurements of all possible variables in the universe, it's pretty clear that you could come up with correlations much higher than any statistical metric standard for "significance" just by chance through combinations of quadrillions of variables with quadrillions of other random variables. While you may not be able to generate a formal mathematical proof, you can easily make a more informal proof that the number of meaningless spurious correlations is likely uncountably larger than the number of meaningful ones in the universe, at least to the level of precision demanded by standard casual metrics for "significance."

      Bottom line is that the universe has a lot more randomness than most people are willing to acknowledge, and most times you "see a pattern" where it's unexpected, it's probably just a coincidence.

    • (Score: 2) by stormwyrm on Saturday February 24 2018, @04:28AM (2 children)

      by stormwyrm (717) on Saturday February 24 2018, @04:28AM (#642879) Journal
      There has to be some kind of plausible mechanism for you to be able to conclude that correlated variables are truly causally connected, and even then you have to also be able to figure out the direction of causality. Much of science is dedicated to the determination of which observed correlations are actually causal. In contrast, a lot of superstition and pseudoscience comes about because of people assuming as you do that correlation is often a sign of correlation, that it is almost always present, without bothering to do the kind of rigorous verification that is the hallmark of science. People have been sacrificed to appease the gods because of such sloppy, uncritical thinking!
      --
      Numquam ponenda est pluralitas sine necessitate.
      • (Score: 2) by HiThere on Saturday February 24 2018, @06:38PM (1 child)

        by HiThere (866) Subscriber Badge on Saturday February 24 2018, @06:38PM (#643111) Journal

        Not knowing what the causal connection is is not itself evidence that there is no causal connection. I will agree that it's often quite useful to know the causal connection, but I will assert that it's often rather indirect, along the lines of:
        A is necessary for B to happen.
        C is necessary for D to happen.
        ...
        E increases the influence of B on G happening.
        B, D, and F are all necessary for G to happen.

        ...
        W, X, and Y are necessary for Z to happen.

        You'll have a hard time tracing out why A happening is correlated with Z happening, but there is still a causal connection (some parts of which are hidden). The default presumption should be that correlation is a sign of causation. But you shouldn't be certain of this.

        An note that the causal chain I was proposing was a very simple one. Only one of the listed links was a probabilistic one, and that's quite unusual. And I didn't only leave out links to save space, part of the reason is that a lot of the time they aren't known to the observer. (Often someone will know them, which is why a search of the literature becomes important. But also you frequently need to dig them out yourself.)

        An example I read about from several decades ago:
        One of the European Casinos had a high proportion of, I think it was black, results from one of it's wheels. This was the correlation. You may make some guesses at all the proposed causal chains that were investigated, but it eventually turned out to be due to wear on one of the bearings. The proper conclusions from the original observation was that *some* causal mechanism was involved. The one that was eventually revealed was not even considered in the original round of causal relationships.

        --
        Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
        • (Score: 0) by Anonymous Coward on Sunday February 25 2018, @07:20AM

          by Anonymous Coward on Sunday February 25 2018, @07:20AM (#643370)

          The default presumption should be that correlation is a sign of causation.

          Just no. A thousand times no. People have been misled by this kind of presumption for all of recorded history, and as mentioned, superstition and pseudoscience is driven by it. Yes, correlation can be a sign of causation, but there are four possibilities [thelogicofscience.com] that need to be considered when you have two variables X and Y that have some kind of correlation:

          1. X really is causing Y to change
          2. Y is actually causing X to change
          3. A third variable (Z) is causing both X and Y to change
          4. The relationship isn't real and the apparent correlation is due to random chance

          The best way to determine which of these four possibilities is most likely is to perform experiments that control the variables X and Y, control for possible external confounders like Z, and are done many times to reduce the possibility of random chance producing a spurious correlation.