Sunday 5 August 2018

"Avogadro Corp" by William Hertling

From the cover's tag line "The Singularity is closer than it appears" I expected this book would be a bit pat - aimed perfectly at my interests, as per this blog's focus. It certainly was very pat. Entirely straight forwards and linear in it's plot structure, with overly plain language and cardboard characters prone to converse in totally unrealistic exposition.

This book is basically a long form best guess at how a very early technological Singularity might be kicked off, via the accidental creation of AI. "Avogadro" would probably have read "Google" if not for the trademark issue - that is literally the company described, in every detail. As such the story is unimaginative. Yet it is still unbelievable in the amount of contrivances it throws in, to start the plot rolling.

The behaviour of characters is often uncanny, too - although this may be down to writing for brevity, without the author having the deft wordsmith skills of the likes of Charles Stross (let alone Banks).

To be fair, it was easy reading, for me, with short chapters and good pacing - I actually finished it (first one this year!). I'm considering going forwards onto his sequels, except the first chapter excerpt name drops both "Gibson" and "Stross" (as future phone handset models). Ick; clearly the author's influences are near perfectly aligned with my own, here, but his hat  tips are ugly AF.
Ray Kurzweil was clearly such a massive and direct inspiration that he mentions him twice in the narrative, plus in the author's end note and (updated) foreword. Pointing out, in fact, that Ray was actually employed by Google in 2013 (two years after the first edition of this book).

Which is kind of ironic, in highlighting one of the books implausibilities [spoilers]: that there was no one around in the company, or the "ELOPe" project who in any way expected or worried that this very advanced system might just be intelligent in it's own right! It's supposed to understand written (email) language and the specific human psychology of recipients well enough to provide changes to composition that substantially exceed the human writer's capability to convince others themselves. That it's self learning too, I mean geez! Run away super-intelligence would be the first thought from anyone in Google (or anyone, really), barely worth even a wry jibe.

It is good in the perspective it presents trans-human AI: as unintelligible and unknowable, with (ironically) no way to converse with it. And in that it's super smart in a social engineering, rather than hackzing all the things and brute-force death-bots, etc. Instead, modifying it's code by secretly recruiting a small army of third party contracted coders.

That this part of the plot, and much bigger business restructuring machinations, happen while the two initiating protagonists were away for a couple of weeks (snowed in visiting parents and on holiday feigning business as usual), was the unrealistic feeling part. You just known that this series of frustratingly dumb shit and overly convenient windows of opportunity have to happen for the story to get where it's going. And that's pretty much the whole book, by length.

So I can't really recommend this, given that it's too dumbed down for those already aware of and interested in this kind of near term futurology. And it's probably not well written enough to compel anyone else to read it. I guess that kind of paradox is what makes fiction really niche.


Side note: My attention was particularly piqued by the inclusion of floating data centres, including converted oil tankers: sea-steading was a particular interest to me, on this blog, almost a decade ago. The perfect platform to build a new, AI nation. Away from the old land territories, all claimed by men. I imagined writing a sci-fi story, in which a tanker data-centre AI attempts to escape persecution by national militaries, into space, extravagantly: the whole damn ship, by means of a space fountain! (Maybe just as a bait and switch distraction from the real escape by micro-satellite, or something...)


Addendum (2018-10-01):

I made the mistake of reading the sequel "AI Apocalypse". It started out marginally more promising, with a more unfamiliar technological evolution and a more immediately exciting development of this. But the characterisations were *so* flat and bad, there was an autistic failure to connect with the huge amount of suffering, death and destruction supposedly occurring and it was all capped off by a unbelievably abrupt ending that would have felt weak for an end of episode reset in a cartoon. It was just utterly implausible that it could have worked, given the axioms of the story and events of the previous book. But the author hit the page count limit, or something...

It was punctuated by a couple of passing assertions that broke my suspension of disbelief even further: (1) China is an irrelevance; just a pirating economy, no significant innovation. (2) This "mech war" game, where players contend by designing increasingly complex algorithms to control virtual automated military hardware, is supposed to have 90% market share... Yeah. Good luck with that, given the current popularity of Fortnight battle royal, and it's shallow, short attention span game-play. He really has zero understanding of gaming or people to throw that in there...

Again, a quick, easy read that I did actually finish. (Even if I wish I had spent that time on some other novel.)

No comments:

Post a Comment

I'm very happy to see comments, but I need to filter out spam. :-)