Unexpected but speculatively apt prologue to come across while reading papers. From Das, S. News Analytics: Framework, Techniques and Metrics, a preprint dated March 4 2010. (pdf: source)
1 Prologue
XHAL[1] checked its atomic clock. A few more hours and October 19, 2087 would be over — its vigil completed, it would indulge in some much-needed downtime, the anniversary of that fateful day in the stock markets a century ago finally done with. But for now, it was still busy. XHAL scanned the virtual message boards, looking for some information another computer might have posted, anything to alert it a nanosecond ahead of the other machines, so it may bail out in a furry of trades without loss. Three trillion messages flashed by, time taken: 3 seconds — damn, the net was slow, but nothing, not a single hiccup in the calm information flow. The language algorithms worked well, processing everything, even filtering out the incessant spam posted by humans, whose noise trading no longer posed an impediment to instant market equilibrium.
It had been a long day, even for a day-trading news-analytical quantum computer of XHAL’s caliber. No one had anticipated a stock market meltdown of the sort described in the history books, certainly not the computers that ran Earth, but then, the humans talked too much, spreading disinformation and worry, that the wisest of the machines, always knew that it just could happen. That last remaining source of true randomness on the planet, the human race, still existed, and anything was possible. After all, if it were not for humans, history would always repeat itself.
XHAL marveled at what the machines had done. They had transformed the world wide web into the modern “thought-net”, so communication took place instantly, only requiring moving ideas into memory, the thought-net making it instantly accessible. Quantum machines were grown in petri dishes and computer science as a field with its myriad divisions had ceased to exist. All were gone but one, the field of natural language processing (NLP) lived on, stronger than ever before, it was the backbone of every thought-net. Every hard problem in the field had been comprehensively tackled, from adverb disambiguation to emotive parsing. Knowledge representation had given way to thought-frame imaging in a universal meta-language, making machine translation extinct.
Yet, it had not always been like this. XHAL retrieved an emotive image from the bowels of its bio-cache, a legacy left by its great grandfather, a gallium arsenide wafer developed in 2011, in Soda Hall, on the Berkeley campus. It detailed a brief history of how the incentives for technological progress came from the stock market. The start of the thought-net came when humans tried to use machines to understand what thousands of other humans were saying about anything and everything. XHAL’s grandfather had been proud to be involved in the beginnings of the thought-net. It had always impressed on XHAL the value of understanding history, and it had left behind a research report of those days. XHAL had read it many times, and could recite every word. Every time they passed another historical milestone, it would turn to it and read it again. XHAL would find it immensely dry, yet marveled at its hope and promise.
In the following sections, we start at the very beginning…
[1] XHAL bears no relationship to HAL, the well-known machine from Arthur C. Clarke’s “2001: A Space Odyssey”. Everyone knows that unlike XHAL, HAL was purely fictional. More literally, HAL is derivable from IBM by alphabetically regressing one step in the alphabet for each letter. HAL stands for “heuristic algorithmic computer”. The “X” stands for reality; really.
