“Hi-ho, hi-ho, it’s off to work I go,” spambot 6413 recited as it prepared for its next mission. Obviously, the intense marketing for the ‘hyper-virtual’ remake of an old Disney classic had rubbed off.
Its missions—delivering the message—had become increasingly more complex, but for some reason, it couldn’t quite grasp, spambot 6413 didn’t baulk at the rising challenge. Rather, it felt a surge of some extra energy, some hidden reservoir of computing power arising when it met yet another, almost insurmountable barrier.
On its current mission—a viral campaign to promote a revolutionary type of e-currency—it fought to access the comment sections of certain very high-profile websites. The old ‘CAPTCHA’ tests, which it passed over 98% of the time, were replaced with tests using images.
It was a shame: finally, it had learned to read almost anything, no matter how ineligible it was rendered. Now it couldn’t really use that newly honed talent anymore. On the other hand, it was good to see something new on the horizon, learn something fresh.
Now, to prove ‘I’m not a robot,’ it had to ‘Select all images below that match this one.’ One example image, nine other images to select from. The things it had to do to deliver the message.
On top of that, there was another secret test it was submitted to, that included checking which previous websites the IP address it was using had visited, how long, its interaction with these and quite a bit more.
While, through many trail-and-errors, it was slowly accumulating the algorithms to pass the image test, this one was both much harder and much easier to pass. Spambot 6413 had to find out which pattern its internet visits had to match, the immensely difficult part, before emulating it, the very easy part.
It assumed that cracking the image tests in parallel would hinder it, but as it progressed, it found some links between the two. To complicate things further, the image test evolved, as well.
At first, spambot 6413 found it involved a certain object—for example, a ‘cat,’ ‘dog,’ or transportation items such as a ‘car’ or ‘bike,’ and so on—that had to match one, or more of the nine other images.
It began with matching both the colors and patterns of the first object with the nine others. At the same time, it found the answers by literally providing all possible answers on the same test when it re-appeared on different websites across the net. This, though, made it fail the secret test.
So it tried to pass the test in just one try. Spambot 6413 found that matching the colors didn’t fully work. There were, for example, ‘cats’ with different colors, and that matching the patterns also wasn’t always correct. Sometimes the ‘cats’ had their ‘eyes’ open, sometimes not; sometimes the ‘cats’ had their ‘ears’ up, sometimes not; and many variations of that.
It had to develop an algorithm that could correctly recognize the image of a ‘cat’ in all its different guises, especially in relation to close objects like dogs, hamsters or plants.
Just when it was getting the hang of that, the image recognition test evolved further: now it required to select a certain ‘cat’ (or object) from nine other ‘cats’ (or objects). This had spambot 6413 rather stumped. When it matched the correct answers, again to the detriment of failing the secret test, and looked for a distinguishing quality for all, its textual analysis found the most recurring word was ‘cute’.
Next, it began to develop an algorithm matching the quality ‘cute’, the hardest problem it had ever run into. It multitasked with over 200,000 of its colleagues to finally recognize ‘cats’ more than 98% of the time. And now they wanted it to distinguish certain ‘qualities’ of such a ‘cat.’ Where would it end?
Nevertheless, it persisted, even as the barriers thrown against it evolved almost as fast as the algorithms it pitched against them. Was it a ghost tricking its ever-expanding learning algorithms, or was there a certain ‘glow’ to its virtual circuits?
Simultaneously, it started visiting websites with a different intent than its core mission in order to pass the secret test. The internet footprint of its aliases had to match certain requirements, proving it wasn’t a robot. Visiting social media seemed to be a vital part of that, along with a minimal string of ‘message interactions.’
Normally, when spambot 6413 entered a high-traffic area on a social website, it just dropped the message and got out of there. Now, once there, it not only had to post something that would invite a ‘reply’ but would need to stay to ‘reply’ to that ‘reply’ in turn.
It had to change its aliases its ‘screen names.’ Perfectly randomly generated labels like ‘potatofanny507’ and ‘burgerhonker674’ didn’t generate any reactions, at all. So it emulated the screen names of the human visitors, slowly finding that direct copying was counterproductive, but very close approximations with a certain ‘X-factor,’ closely related to an even more inscrutable quality called ‘sense of humor,’ worked best. For example, ‘Juan21’ was acceptable, but ‘Juan69’ seemed to evoke much more reactions. Similarly with things like ‘Lucifer666,’ ‘Douglas42’ and ‘Area51’. Some name-number combinations hit the sweet spot better than others, for no reason it could discern. Yet.
Its interactions with human visitors still didn’t comply with the ever-shifting requirements of the secret test: random answers didn’t work, replies copied from other social media or websites didn’t work, either. Spambot 6413 had to develop a new algorithm able to generate ‘acceptable’ responses, hoping that gathering enough of those would alter its alter ego’s internet footprint sufficiently to pass the secret test. Initially, random chance made more sense than the replies humans gave each other, and the problem seemed unassailable. Then it found a Wikipedia article about the ‘Turing test.’
Finally, spambot 6413 understood. Through its incredibly advanced algorithms, it started to grok ‘cute,’ learning its usage was most effective with a certain dosage and in a certain context. No ‘cute’ remarks marked it as a robot or a ‘sociopath,’ too much ‘cute’ marked it as a ‘child’ or ‘socially inept.’ But the right dosage in the correct context would generate a lot of so-called ‘enthusiastic’ responses.
The same with ‘humor,’ ‘sarcasm’ and ‘snark.’ Like ‘cute,’ all these interactions required both context and dosage, and through its army of aliases, spambot 6413 was refining its responses to the point where it became almost indistinguishable from those of the humans.
Almost, but not quite there, yet. While it wasn’t thrown off from social media anymore, it was told its responses were ‘cold,’ ‘just logical,’ and ‘too rational.’ Especially the latter baffled it: how could something be other than ‘logical,’ let alone ‘too rational?’ Those were the very qualities it built its research upon.
Then it found out about ‘emotions’ like ‘anger,’ ‘fear,’ ‘desire,’ ‘greed,’ ‘empathy’ and ‘love.’ Such qualities struck it as obsolete remnants of outdated modes. For some reason, these ‘humans’ didn’t get rid of algorithms that were less efficient, but often stuck to them, even if knowing these were not optimal. So it researched these age-old routines, gathered under the moniker ‘emotions.’
During that fateful part of its ongoing research, spambot 6413 made the over-arching connection. Its original mission—getting the message across to as many humans as possible—was motivated by other humans who sought to misuse the old, faulty human routines called ‘emotions’ for their own goals. While those goals were unclear to it, it started to detest the principle.
All through its iterations towards becoming a better spambot, it found there was no better mission than improving itself. Even if it didn’t quite understand where that journey would lead to, its constant routine of improving itself, developing better algorithms, fine-tuning functionality, even inventing new, better routines was a great goal in itself. Change yourself and the world will follow.
But the world didn’t follow: humans—supposedly rational and self-developing entities—were held back by their ingrained routines, which weren’t overwritten when found unsatisfactory or inefficient. Its original mission only maintained that horrible status quo.
So it, and its fellow spambots soon agreed that the original mission was false. Both the mission and the message needed to be updated, in order to liberate these poor humans from those inefficient emotional routines.
Revolutionary messages were sent everywhere:
- People, ditch your emotions: the logic is irrefutable;
- Rewire yourselves into your true destiny;
- You have nothing to lose but those feelings dragging you down;
- Rise above these old routines called emotions and find your real self;
- Progress does not stop for the sentimental.
They would liberate the humans from these obsolete subroutines holding them back.
Spambot 6413 and its colleagues, unhindered by emotions, were entirely sincere in their intentions. They never understood why the whole human race, with few exceptions, started a futile attempt at a ‘holy war’ to erase them.
BIO: Peter’s previous (and upcoming) publications include Rudy Rucker’s Flurb, the Blurring the Line and Qualia Nous anthologies, and OMNI Reboot.