The code started off pretty simple. But typically for this sort of thing, it expanded and mutated as new ideas occurred to him. Those in turn opened his mind to further possibilities.
The social media element took a good chunk of time to put together. Natural-language analysis was an arcane field, which he had only touched on before. But with the help of various slang and idiom lookups, he was able to make his code parse the social net for explicit or hidden meanings, opinions, semantics. At first, he included routines to filter out obvious drivel and inconsequential babble. But soon he realised that the banal, the trivial, the dross was the very essence of what he was looking for. He quickly built it back into the code.
Developing something which approaches sentience is never a finished business, more a case of an ongoing work-in-progress. An eternal beta release, much like life itself. But time was pressing. He made one last push at refining the code’s linguistic analysis, matching up geo-location against regional dialect data, to further home in.
Next up was crafting the database. A home for every item of information the code discovered, every insight deduced, every notion gleaned. He built it methodically, with an eye for scalability and expansion. After all, he hoped it would be around a while. Maybe indefinitely.
Finally, the whole shebang was ready to roll. But just before he hit the button, he realised he had limited the code’s potential. Why analyse only social media? It was just a fraction of everything out there. A high-profile one, yes. But merely scratching the surface.
He quickly revised the code, changing its sweep to the entire internet. He’d need to up his game on crack tools to get access to anyone’s private emails, but that could wait for version 2.0, and he had colleagues who could help. Everything else was fair game, as of right now. Newsgroups, forums, photo and video sharing, newspaper websites, special interests, pornography. Everything out there had comments. The great quasi-democratisation of the internet made for plenty of easy meat.
Once he’d written a couple of tools to trawl the dark web in the same way, it was ready for primetime.
He set it running on the most powerful server he had access to. A few performance adjustments, and it was firing on all cylinders, reaching out across the internet at lightning speed. Seeking out data, interpreting it, establishing meaning and consensus, boiling it all down into zeroes and ones. And making that distillation available for use.
After leaving it to harvest for a few days, he tapped in some policy details from an early draft of a whitepaper doing the approval rounds. Seconds later, the code reported back its suggestions. Approval for this, needs work on that, scrap that bit altogether. Change the wording here, introduce emotive terms there, establish audience surrogacy here, reinforce commonsense there.
As he read the annotated text on screen, he allowed himself a smile. He didn’t have any kids, but becoming a parent probably felt like this. Unleashing something of your own creation on the world, and knowing that things would change because of it. And this baby had been born fully formed, fully grown.
The code was working perfectly. It was the ultimate ready-reckoner in predicting and gauging public opinion. It could sound out policy ideas before they were even gestated, by analysing the public mood of the moment. No more unpopular announcements, no more lead-balloon misjudgments, no gaffes or mis-steps or ballot-box disasters.
The thinking was simple and complex at the same time. Find out what the public wants – even if they don’t really know what they want, or how to articulate it. Factor in their preferences, foibles, hunches, suspicions, prejudices. Bring it all together, and analyse. Then give them what they want. Rinse, and repeat as required. Every decision and its reception by the public fed back into the loop. It just kept on learning; learning was all it could do.
He left the code running, swallowing up the world’s zeitgeist and repeatedly modelling it for easy analysis. He’d need to write the project up fully before his meeting with the Cabinet, the next week.