I often claim to live in the present; the past is a foreign land, and many will know how I feel about those. One of the many implications of this attitude is that I forget... and unfortunately, this includes forgetting 'ideas', which tend to blaze through my awareness only to fizzle in the atmosphere of my ongoing lethargy. When I went to write this post, I finally remembered a meta idea.
Oddly, however, I'm intensely proud of having ideas. So, in one more effort to build not this pride upon sand, I thought I'd try to 'open-source' my brain, spending my limited enthusiasm for an idea by writing it on the indestructible clay of the internet instead of merely being intensely distracted for a few hours.
What odds that this will be the last post I make on this blog? *
Anyway, today's idea, which isn't all that exciting after all that waffle:
Auto-completion seems to be pervasive nowadays, what with phones and word processors finishing words, and Google helping us answer important questions at near light-speed. Yet it tends to drive me nuts... but what would drive me nuts less - and might be genuinely useful for people who struggle to put together a sentence - would be to exploit "big data" for more than just word/phrase completion.
Suppose, for instance, you had a burning desire to write a pseudo-Austen novel (ok, I still haven't read this possibly amusing travesty), but you wanted to churn it out quickly so you could get on to something more interesting. Say that your fancy quad-core computer, instead of sitting around idling while your brain was churning, spent a few cycles analysing whatever sentence you were working on and subtly popped up related sentences at the bottom of the screen: not only ones suggested by your latest couple of words, but by 'significant' words in your existing sentence, or even combining that information with what was happening in previous sentences. The sentences, or fragments, could come from Austen, or other period writers, or ...
In short, plagiarism for dummies.
But who said it was such a bad thing? For one, you've got all Gutenberg to lean on if desired (and I suspect there are probably better sources around), and for two, a more realistic scenario might be writing an internal report for your company. Plagiarism in this case is habitual; why not make it more efficient?
So, does it exist? If so, why isn't everyone using it?
[I refrained from using fancy phrases like n-grams, language model, concordance, corpus, etc. But I hope you know that I know them and this pre-postscript suitably creates or reinforces some positive impression.]
* Yes, I know, such stereotypical pessimism. One has to hold on to personal tropes.
Oddly, however, I'm intensely proud of having ideas. So, in one more effort to build not this pride upon sand, I thought I'd try to 'open-source' my brain, spending my limited enthusiasm for an idea by writing it on the indestructible clay of the internet instead of merely being intensely distracted for a few hours.
What odds that this will be the last post I make on this blog? *
Anyway, today's idea, which isn't all that exciting after all that waffle:
Auto-completion seems to be pervasive nowadays, what with phones and word processors finishing words, and Google helping us answer important questions at near light-speed. Yet it tends to drive me nuts... but what would drive me nuts less - and might be genuinely useful for people who struggle to put together a sentence - would be to exploit "big data" for more than just word/phrase completion.
Suppose, for instance, you had a burning desire to write a pseudo-Austen novel (ok, I still haven't read this possibly amusing travesty), but you wanted to churn it out quickly so you could get on to something more interesting. Say that your fancy quad-core computer, instead of sitting around idling while your brain was churning, spent a few cycles analysing whatever sentence you were working on and subtly popped up related sentences at the bottom of the screen: not only ones suggested by your latest couple of words, but by 'significant' words in your existing sentence, or even combining that information with what was happening in previous sentences. The sentences, or fragments, could come from Austen, or other period writers, or ...
In short, plagiarism for dummies.
But who said it was such a bad thing? For one, you've got all Gutenberg to lean on if desired (and I suspect there are probably better sources around), and for two, a more realistic scenario might be writing an internal report for your company. Plagiarism in this case is habitual; why not make it more efficient?
So, does it exist? If so, why isn't everyone using it?
[I refrained from using fancy phrases like n-grams, language model, concordance, corpus, etc. But I hope you know that I know them and this pre-postscript suitably creates or reinforces some positive impression.]
* Yes, I know, such stereotypical pessimism. One has to hold on to personal tropes.
50 shades of grey.
ReplyDeleteLet's make some money.
So, what source books are you going to feed in for this one? I'm looking for legal plagiarism, y'know, and I don't think 19th century erotica would sell well nowadays...
Deletehmm, maybe it would? (*adds note to imaginary reading list*)
Being unable to fake even a passing familiarity with such terms as n-grams, etc., I'll move along to something I can fake a passing familiarity with: expertise, and the cutting of corners.
ReplyDeleteSo, way back in the primordial mists of my youth I dreamt of being a ninja warrior (false) and so engaged in the study and practice of karate (true). Rather than being taught graceful and deadly punches and kicks however, I was taught very formal and exaggerated versions of such - all of which seemed highly inappropriate for either actual combat situations or Van Damme films. When I cast my gaze upon those lofty peaks of my aspiration (the black belts), I observed none of the cumbersome movements I was encouraged to practice. It frustrated me at the time but, remembering the lessons of Mr Miyagi, I persevered in the trust that all this waxing on and off would eventually deliver me the respect of my peers and the girl of my dreams.
As I progressed, I found that the misshapen and awkward rock of my formal punches, when placed in the river of thousands of repetitions, were polished into something smooth and elegant. The extraneous and exaggerated aspects of the movement were pared away to reveal it's essential points, and so my karate became more graceful (maybe) and more deadly (not really).
So it all worked out (well, not the bit about the respect and the chicks), but I have to wonder if the approach was really necessary. Couldn't I have just been taught the movements of the black belts, rather than have them reveal themselves to me over the passage of many variously-coloured trouser-supports? I would say no: that, with respect to karate at least, there are no short-cuts. I should acknowledge though, that the salt this view needs to be seasoned with is that I did, in fact, execute thousand of punches, and that's a big investment to turn around say was a complete waste of time.
Anyway, how does this relate to the OP? In many ways, computers are placing the power to cut corners in the hands of metaphorical white belts. Of course, for the expert, the cutting of corners is a convenience, and a greatly appreciated one. But for the beginner, the cutting of corners is a developmental road-block. If technology allows white belts to write like black belts, where will we get our actual black belts from? Am I stretching this metaphor too far?
Now, I'm not even remotely interested in proposing that we deliberately inconvenience ourselves by hobbling technology. I'm just observing that there seems to be something corrosive about convenience. In conquering technology, are we vanquishing ourselves?
I'm suitably embarrassed at how long it took me to work out who that was (damnit, why didn't I actually pay attention to the username?).
ReplyDeleteI am, however, suspicious of arguments that are along the lines of 'you have to suffer through this tedious initiation to truly be able to do this'. They tend to crop up each time there's some technological innovation, and ultimately fall by the wayside as "probably not as important as we thought" (hand-writing, spelling, assembly language programming, long division, ...).
And black-belts will happen naturally. Some people will be committed enough to understand assembly even when they don't strictly need to...