A quick prelude - this is content excerpted and edited from an essay written on the much more specific topic of personal sources of "idiolect" change. Expect some density in prose.
The internet crowds out pretty much any other source in terms of its amount of impact on my language - it's not that it by itself is more compelling than the rest: it just happens to cannibalize more and more time out of my life and in turn represent even more of the media I - we - consume. This happens to be very good news for language in general if we remember that the internet's lifeblood is very much text right now. It means, for example, that we're reading and writing (but mostly reading, to be terribly honest) on a scale never seen before, on topics we genuinely like.
When we consume information on this scale the notion of "efficiency" (or "sloppiness", depending on your interpretation of the phenomenon, although I prefer the former since the behavior is natural) starts to creep in - our natural gravitation, over time, to uses of language that for some reason or another are much more "obvious". While I can't argue humans are more likely to interpret a word a certain way exclusively because of the way our brains are wired, I do believe that some of the odds are stacked because of previous experience with English. (What "previous experience" means in this day and age is all up in the air - and I am thankful for that.)
But it this type of change doesn't always have to be based off complete re-interpretation of a word from its "first principles" - I'm sure we hold onto many words precisely because they and their meanings seem counterintuitive (the better to be different) or because they resist easy decipherment. I suppose the one generalization we can make is that we take the shortest path possible to claiming a word as ours. The process seems perfectly natural and intuitive like that.
There are larger sociological factors at work, though: to squeeze out
the most utility out of the internet one has to use English, for
better or for worse. While one may point to the tradition of using English in programming (once conflated with the concept of
"computing" as a whole) for names and comments as progenitor to this trend, a more likely cause is much more contrived and outsite: the internet's origins in academia, and, in particular, the enforced-English environment at CERN. This trend is sustained by the fact that most innovation on the internet (China's own separate one-way alternet aside) does come from the US - hence one can guarantee an initial userbase of English speakers for most things up-and-coming. The internet is inextricably linked with English - this will most likely provide trade off longetivity for stability (far less than in previous centuries) in the scope of centuries.
Ultimately, though, the fact that the internet has very much "arrived" puts the most accessible example of this type of change out of commission - texting shorthand no longer represents the state of language on the internet. We instead have everything in between for services on the internet that do span the gamut from brief to encouraging depth. My belief, though, is that the community and its trends and fads will provide much more language augmentation than the actual services - after all, inventing something radically new is harder than incremental evolution, and what ultimately matters is whether a community can cohere around a service.
I would argue, though, that the brunt of language change has followed anything resembling counterculture on the internet into media less mainstream. As a result I don't think we'll see a return to obfuscation simply for the sake of it - at least not when the purpose of communication is more about actually communicating than asserting identity. (To be honest, it isn't just the fact that the mainstream has moved into the business of different media - it's also the fact that most of these media have an explicit "social" edge to them, and thus the need for clarity... i.e. Twitter - same limitations, but little syntactic change: just perhaps a word or two) The canonical example of "inbred language" is interesting symbiosis of image and text called the "imageboard" (the concept is simple enough - one image and a piece of text that is most expressly not caption), for example, that has gained notoriety as the medium used by the site 4chan - its own fame stemming from its very libertarian approach to identity and moderation and the resulting drama and fad creation it has caused. The site itself is an interesting "application" of English which lies inbetween paragraphic exposition and quick bursts: when much of what you want to convey, like base emotion or an example, is much more clearly explained with an augmenting image, written text must skip to the details - exposition without a premise. So much of the written internet is left to those without any ingrained interest in continuing traditions like these merely for the connotations the heavily abbreviated word confers - the great unwashed - the hoi polloi ("efficiency" in action - do I have to know that "hoi" translates to "the" in order to use it effectively? No. It's easier if I don't - it's incredibly easy to determine what it means from context, and I would argue that the assumption that both words join together to make a single concept is perfectly harmless) compared to the elite but paradoxically more open and egalitarian ranks of first adopters. And thus the cycle continues, as with anything deemed exclusive.
We're seeing people "on" the internet starting to value clarity in discourse over everything else - efficiency, yes, but nothing seemingly contrived. No one is immune to the fads that sweep the internet, of course, but when one does make the rounds these get introduced with quirky little items that quickly establish a specific usage for each of them. Some generalization is unavoidable, of course, and that is when a word enters our lexicon for real. But never forget the gauntlet inbetween: staying power is and should be inscrutable, otherwise I'd be a damned rich media mogul.
Change on the scale of single media at this rate is very much uncharted territory. I expect that as we start contorting our language around whatever happens to come and pass, this concept of efficiency will come to dominate so as to expedite this process. What will not change, though, is the need for clarity in language and that is very much in line with efficiency. If a language can provide both - English aside since it is joined at the hip with the internet - then it is guaranteed a place on the internet as a native language. Resisting this type of top-down change ensures artificial existence as a second-class citizen - if one cannot express a concept by loaning a word or two, one simply resorts to English. The natural result is a cultural disconnect - despite China's ascendancy to New World Order status, I couldn't give a single flip about what of cultural significance is churned out there because there is ultimately no one to share it with.
Ultimately, there's nothing here to be cynical about - nor is there any need to explicitly embrace change. So the internet is ultimately our best, first, and possibly only (while there is currently no historic comparison to be made, any successor will be judged in terms of the internet) shot at providing a meta-medium for experimenting with communication, considering the freedom it enjoys and the noticeable absence of disruptive tollboothing. If one subscribes wholeheartedly to the concept, what's not to like? Or, more importantly, what's to lose?