Tagged: citation Toggle Comment Threads | Keyboard Shortcuts

  • Profile photo of nmw

    nmw 15:27:59 on 2016/07/12 Permalink
    Tags: academia, academic, , , , , , , , bandwagon, bandwagon effect, , citation, , , , , , , , compute, , corrupt, corrupted, corruption, , , , , , , group think, groupthink, , , , , , , , , , , , , majority, , , , , populism, populist, , , rason, , , , , , , , , , systemic, , , trusted, , , universities, , valid, validity, vote, votes, voting, ,   

    The Spectre of Populism 

    There is a spectre haunting the Web: That spectre is populism.

    Let me backtrack a moment. This piece is a part of an ongoing series of posts about „rational media“ – a concept that is still not completely hard and fast. I have a hunch that the notion of „trust“ is going to play a central role… and trust itself is also an extremely complex issue. In many developed societies, trust is at least in part based on socially sanctioned institutions (cf. e.g. „The Social Construction of Reality“) – for example: public education, institutions for higher education, academia, etc. Such institutions permeate all of society – be it a traffic sign at the side of a road, or a crucifix as a central focal element on the alter in a church, or even the shoes people buy and walk around with on a daily basis.

    The Web has significantly affected the role many such institutions play in our daily lives. For example: one single web site (i.e. the information resources available at a web location) may be more trusted today than an encyclopedia produced by thousands of writers ever were – whether centuries ago, decades ago, or even still just a few years past.

    Similarly, another web site may very well be trusted by a majority of the population to answer any and all questions whatsoever – whether of encyclopedic nature or not. Perhaps such a web site might use algorithms – basically formulas – to arrive at a score for the „information value“ of a particular web page (the HTML encoded at one sub-location of a particular web site). A large part of this formula might involve a kind of „voting“ performed anonymously – each vote might be no more than a scratch mark presumed to indicate a sign of approval (an „approval rating“) given from disparate, unknown sources. Perhaps a company might develop more advanced methods in order to help guage whether the vote is reliable or whether it is suspect (for example: one such method is commonly referred to as a „nofollow tag“ – a marker indicating that the vote should not be trusted).

    What many such algorithms have in common is that on a very basic level, they usually rely quite heavily on some sort of voting mechanism. This means they are fundamentally oriented towards populism – the most popular opinion is usually viewed as the most valid point of view. This approach is very much at odds with logic, the scientific method and other methods that have traditionally (for several centuries, at least) be used in academic institutions and similar „research“ settings. At their core, such populist algorithms are not „computational“ – since they rely not on any kind of technological solution to questions, but rather scan and tally up the views of a large number of human (and/or perhaps robotic) „users“. While such populist approaches are heralded as technologically advanced, they are actually – on a fundamental level – very simplistic. While I might employ such methods to decide which color of sugar-coated chocolate to eat, I doubt very much that I, personally, would rely on such methods to make more important – for example: „medical“ – decisions (such as whether or not to undergo surgery). I, personally, would not rely on such populist methods much more than I would rely on chance. As an example of the kind of errors that might arise from employing such populist methods, consider the rather simple and straightforward case that some of the people voting could in fact be color-blind.

    Yet that is just the beginning. Many more problems lurk under the surface, beyond the grasp of merely superficial thinkers. Take, for example, the so-called „bandwagon effect“ – namely, that many people are prone to fall into a sort of „follow the leader“ kind of „groupthink“. Similarly, it is quite plausible that such bandwagon effects could even influence not only people’s answers, but even also the kinds of questions they feel comfortable asking (see also my previous post). On a more advanced level, complex systems may be also be influenced by the elements they comprise. For example: While originally citation indexes were designed with the assumption that such citation data ought to be reliable, over the years it was demonstrated that such citations are indeed very prone to be corrupted by a wide variety of corruption errors and that citation analysis is indeed not at all a reliable method. While citation data may have been somewhat reliable originally, it became clear that eventually citation fraud corrupted the system.

     
  • Profile photo of nmw

    nmw 18:49:07 on 2016/06/25 Permalink
    Tags: , , , , , citation, citation analysis, , , , , few, , , , , , , , many, , , sample, sampling, , , ,   

    Don’t Listen to One Single Piece of Good Advice — Listen to Many 

    Several months ago, I mentioned on one of my other blogs that I enjoy listening to Gretchen Rubin’s „Happier“ podcast. I still do, even though I think content sponsored by advertising is by and large fake.

    Recently, Gretchen (and Elizabeth) asked her (their) listeners what the best piece of advice was that they ever got. I responded (they asked for people to phone in their comments – I think my remarks may have arrived a little too late for episode 70, but perhaps they might appear in episode 71(?).

    This was the gist of my message: Don’t Listen to One Single Piece of Good Advice — Listen to Many!

    This is also something Jason Calacanis mentioned in a recent episode of his „This Week in Startups“ podcast, but I can’t remember which one – that you should never rely on just one source of information. I remember thinking as I listened to Jason (and of course I had heard such advice decades before from many of my school teachers): „does that mean if you search for information you will not only listen to Google?“ Stange as it may seem, my hunch is that for the vast majority of the population, this is not the case. Indeed, my experience has been that most people will only search for information using Google’s algorithms – if they do not see anything that appeals to them via Google, they will assume that no such thing exists.

    Incidentally, there is also another kind of parochialism that I feel is closely related to this fanatical belief in Google’s scoring algorithm. In a recent episode published by HBR’s „Ideacast” podcast, Todd Rose was interviewed about a book he had recently written (” The End of Average: How to Succeed in a World That Values Sameness”) about measurement and statistics. His argument echoes something I have long held to be true (and I think I recall that one of my comments regarding this matter also appeared on a German radio program – perhaps 5 or more years ago).

    Oddly, Google fan-boys (and fan-girls, too, of course) often overlook the fact that Google also ranks results according to such „cooked“ statistics. In fact the situation is even worse: when Google calculates its metrics for websites, then those metrics are applied regardless of how relevant they are (or aren’t). So while SAT scores attempt to measure both mathematical ability and verbal ability, Google’s statistical measurement for quality (which was shown to be totally bogus decades ago) is applied whether or not the source is reliable for the search query. It is essentially a „one-size fits-all“ metric (which also happens to be totally unreliable). Yet very few people really care, because most people use Google mainly to search for domain names anyways (in other words: they „search“ for ebay because they are too lazy to type in ebay.com. I bet if people stopped doing that, then the reduction in energy required might actually reduce global warming significantly! 😉

     
c
Compose new post
j
Next post/Next comment
k
Previous post/Previous comment
r
Reply
e
Edit
o
Show/Hide comments
t
Go to top
l
Go to login
h
Show/Hide help
shift + esc
Cancel
Skip to toolbar