Pages

Tuesday, 21 February 2017

Video Is Bad For You

As a statistics student one of the things you'll do is examine residual plots for signal -- patterns that indicate the model needs further work. However, as people we identify more patterns than what actually exists. This is why Spotify redid its "random" algorithm, Dilbert has a comic with a "dude" saying 9 repeatedly in his capacity as a random number generator and why I look at the pattern of my multi-choice answers when guessing (to see if it's something someone thinks is random; yes this is problematic... am I assuming I am better at perceiving randomness? do I ignore that a computer has probably sorted it in this day and age? Point is, I still do it).

Obviously this is a big problem if we're modelling something. If we're interested in causality "fake patterns" will compromise our inferences. If we're interested in prediction, we could build a model of the data rather than the process -- data are assumed to come from a process -- and, hence, we're predicting a para-reality. But, it seems to me, from an evolutionary point of view, this is very interesting because it seems very useful.

Imagine, if you will, that you're out and about in the jungle. Now, this is a scary place to be for anyone who isn't equipped with fairly modern weapons (and, even then, in many respects it still is anyway) but it's also a very busy situation. As a species we're a great prey animal. Bad hearing. Crap sense of smell. Lots of mass, i.e. food. Limited capacity to run, climb or swim to safety. Oh, and we have no defensive weapons -- no canines, no claws and no power (a chimpanzee is a lot stronger than a human). Basically all we can do is notice a lot of stuff (mostly visually) and link it together, and give it a meaning. We can also easily pass on this meaning to other people.

One of the big ideas in statistics is the null. Basically, a statistician assumes innocence or boringness -- this thing did not do/is not associated with this cool/interesting thing. In other words, statistics is very cautious and (intellectually, not politically) conservative.  But this is a taught characteristic and equivalence. What is conservative ultimately depends on a lot less than what is cautious. The conservative reading is that of meaningless. Or, put another way, science generally assumes squares living in square worlds. But for our jungle walker, cats are real and it's a dog eat man world. Caution, there, means assuming meaning -- but in both situations it's about harm minimisation. Specifically it's about the relative harm. If you're not cautious in the jungle, you'll ignore the noisy rabbits (saving energy) but you're probably going to be eaten by a tiger. If you're not cautious in academia, you could be "ahead of the curve" but realistically you're going to be caught out on closer examination. But thinking that "twig snap" = tiger isn't conservative, even if it is cautious. That's the point.

Now, this is an interesting point and it's probably worth dwelling on. But it also gives me a sense of deja vu -- that, perhaps, I've seen a video about, if not the stats points, the evolutionary example. The thing is, you read the above a lot faster than you would've watched a video saying the same thing. If you don't believe me, time yourself reading a paragraph aloud and silently. In fact, think about how much you learn when you're doing that. It is the same right? A video of the above discussion "teaches" exactly the same content as the above discussion as is, right? Which is to say, the lesson I've been building to is that the information load of videos is pretty God damn awful. Huh. But, does this matter? Is it practically significant or just some "true but trivial" lesson paraded by pedants? Well, I rather think it is something we should be concerned about. Quite concerned, even. I think an example will help explain this.

I used to have a clear morning routine that invilved reading the news on TVNZ's website, the NZ Herald's and the BBC's. These days, though, it's months between visits to TVNZ. Why? Hopefully you're thinking videos -- although I don't like the horizontal layout, indeed, I feel it actively harms my perusal of the site. But we need more details, yeah?

There is no way that I could ever have read all the articles on all three sites or, even, all the latest ones in each main category. But what I could do was exercise much more control over how I consumed information. A video can look pretty and professional (thus hiding the uncertainties so much more easily) but if a fast-forward it's like reading every other paragraph. If I need to speed up a search for "meaning" (the pattern of things I care about) with "written" articles (although, in truth, while I may be happy to call practically anything an article.., videos aren't articles) I can scan the text. I don't understand or read all of it -- but I am able to spot patterns for meaning (signal indicating newsworthiness) and maybe slow down where appropriate (rather than just skipping material). I can also read the whole thing faster than the video plays, which also matters. (And if you're thinking about, say, playing at 1.2 times the speed, as opposed to fast forwarding, read on.)

We live, today, in a world concerned less about infotainment than it is about "fake news" and "alternative facts" (remember when infotainment was the Devil?). Indeed, we live in a meme-news era: I accidentally discovered the origin of the phrase "alternative facts" as the in-vogue expression via clickbait. But this misses the point -- the real news has to be entertainment too. In the past, this probably begab with "if it bleeds, it leads" and then that evolved into "human interest" delivery -- you know, things where, say, the bombing of Hiroshima would be reported on by cutting to [grandchild] of [granny visiting Hiroshima], when not glorified versions of "cat in tree" stories.* These days, while these things are still true, real news must now be about video. Video, it seems, is the final frontier.

I'm not saying that if people watched fewer videos, (other) people would be less concerned about fake news due to being less conned. However, I really am saying that in a world confronted by "fake new" a news-audience needs to process information individually. The truth is that it's not good enough to read half an article whilst listening to the "news" video plating in the other tab and thus get nothing from either (we suck at multi-tasking, live with it) or subconsciously melding them instead. It's not enough to be reliant on a medium unsuited to scepticism... a typo, I think, could really make you doubt a fake news piece but how do you notice that in a constantly perpetuated video? (Typos are indicative of shoddy editing, which suggests shoddiness, which suggests the reporting may be shoddy.) And, it must be said, practice makes perfect... the more you read of "real news" the better your sense of the variation in "normal" reporting ("fake news" being "pathological" reporting). But maybe it's hopelessly naive of me to think this. A lot of "fake news" is written, so maybe people who don't notice now aren't going to realise that it's outside the aforementioned variation. And maybe "fake news" has nothing to do with modes of consumption and people are right to analyse it from the echo chamber... i.e. modes of thought (e,g. believing what you already believed). But I obviously think that our current modes of consumption are conducive to the proliferation of spurious truths.

What to make of this? Am I some would-be enfant terrible yearning for a return to a lost-era I never experienced where the internet was dominated by the written word? Or have I stumbled across something of genuine interest? Could it even be that we need to consider what people are likely to read? That is, does it matter that comments sections in news articles generally follow smoothly from the article and with videos the comments are generally clearly separated? Does this change how often comments are used? Is the reason why no-one refers people to comments sections because no-one actually reads them? Could it therefore be that videos exacerbate and foster echo-chambers? Certainly, it's harder to reply to a video than it is to a written document: at least, without a transcript. Does this make videos the ideal vehicle to parade ideas that one doesn't want critiqued? These are important questions. and this is an important topic. Videos are bad for us. But sometimes they're fun to watch.

*An inconvenient truth about Campbell Live lost in the hooha surrounding its cancellation was that it's so-called "hard reporting" was mostly this kind of story, plus infrequent actual hard reporting.

No comments:

Post a Comment