"Well, probably a very good time to go lash down everything moveable, and really examine your surroundings for what could shake loose…"
So started the rare between-reports email from Cliff, head "time monk" at Half Past Human, indicating the increased likelihood of a major earthquake that Saturday night. Being new to the world of "predictive linguistics," I made a mental note to keep an eye on the news that weekend, but didn't think much more of it.
Until I got in the car Monday morning and heard the reports about an 8.0 earthquake that had hit the Sichuan province in China overnight.
They may have missed the mark on a few details, but they sure got the big one right. And a few little ones, too, including predictions of media coverage around a "wedding interrupted." This made more sense when AP images started to show up on outlets like CNN and the BBC showing a bride and groom surrounded by rubble after the quake hit in the middle of their ceremony.
From there forward, I started paying more attention.
The solo project began back in 1997, as an attempt to make a little money by predicting stock price movements via speed-reading the Internet looking for certain terms. While collecting pieces of information that referenced Sun Microsystems in an effort to find patterns of people buying or selling, Cliff also detected patterns that made no sense. Beyond the financial analysis, Cliff also noticed a lot of language from gardening forums, ocean navigating forums, mountain climbers, etc. talking about the sun in the sky seeming "whiter." Being an avid bicycler, he'd also made this observation himself. "I realized that if these people were emotionally concerned about this, and it was rising to levels so much more ‘hot' than the language around money, then there might be something here."
Here's the basic idea: just like a rock thrown in pond makes ripples in all directions, events in time cause ripples both forward and backwards. Some people are more inclined to perceive those backwards ripples than others, and now that the Internet provides a tool to actually see what a huge number of people are thinking about like never before, there is a potential medium to tap into those ripples and make predictions.
As Cliff puts it, "Basically, the computer part doesn't matter. Humans are psychic — they just may not know they're psychic. But even those who don't know they're psychic ‘leak out' information. We go and gather up the information that leaks out in their correspondence with Aunt Maggie and analyze it for patterns. We are more privy to the media hysteria after the event than to the event itself." Although he's also quick to note that they only pull from public information sources, and aren't reading people's email.
Unfortunately, the real world is more like dumping a truckload of gravel into the pond all at once. Each stone makes its own set of ripples, with these waves within waves all interacting with each other and generally making a jumbled mess. The art of radical linguistics is in sorting through the mess to find coherent patterns, and keeping track of whether the emotional tension around those patterns is building or releasing over time.
The technology is comprised of a database of roughly 750,000 words and phrases in various languages and alphabets. Each word is linked to a matrix of values that define the word as it relates to different qualities such as intensity, duration, and emotional direction. For example, the phrase "to jack" might rank a higher value for intensity than "to steal" -unless it's constrained within the context of a flat tire, and those constraints are tracked too. And a word like "anger" might have a shorter duration value than a word like "grudge."
"Clear as Sirius on a foggy night?" as the time monks might say. Well, this is only the beginning of the complex analysis.
This entire dictionary is then broken down into groups of increasingly finer granularity. At the top are large entities like "GlobalPopulace" and "Terra." GlobalPop does not include the U.S. populace, a split necessary due to such strong anti-Bush sentiments in the global media post-2001 that the analysis would otherwise be skewed. Terra keeps track of all biosphere-related concepts. Others big buckets include "ThePowersThatBe," "Markets," and my personal favorite, "SpaceGoatFarts." Here we find references to "unknown forces of all kinds from space which are terra/populace affecting."
Out of this entire master lexicon, the time monks hand-select roughly 30 to 40 thousand words and phrases that seem to have the most current predictive value for the future. This list gets tuned between data-gathering runs to better reflect what concepts grew in strength, or started to drop off, in the previous analysis.
"Spiders" then crawl the web searching for those words, much in the same way Google goes about indexing the internet. When a word is found, the spiders grab a chunk of the surrounding text, and copy it into the database, and reduce it down to a set of numbers linking back to the words in the main, larger lexicon.
After roughly three weeks of data gathering, the analysis can begin. The system creates scatter plots of the found words, the "modelspace" of the system, using the structured lexicon and linked emotive values as a guide. Each word represents a single pixel, with color and shading changing depending on whether you're looking at the "duration" layer, the "emotional direction" layer, etc.
This is where the human part of Half Past Human really goes to work. By drilling down through the various layers and searching for shifts in language usage, such as new concepts appearing, existing ones growing stronger, or older concepts suddenly disappearing, they can begin forecast changes in the real world.
Explains Cliff, "In a way this is similar to how astronomers find new thingies in space by rapidly shifting the view of various snapshots of the same spot of space. In both instances, movement, or change is being sought in the minutia of the flickering images. We probably have it easier than the astronomer fellows."
Occasionally, "meta data layers" also appear in the analysis. These are similar concepts that start to appear across all of the entities, pointing toward key things to really keep an eye on. Current meta data trends include restrictions on movement, encounters with scarcity, duality, and resistance/rebellion/revolution.
From the analysis of each data run, the time monks generate a series of six to eight weekly reports that summarize the appearing trends for subscribers. The reports start with near-term predictions and then progress further out in time as the team drills down through the various layers of data.
The reports are offered on a per-run basis, with new subscriptions costing $280 USD. Subscribers can then renew for later runs at $70 USD. The price points, in part, are intentionally chosen so that people who don't have money to lose don't "gamble" their limited resources on a self-described crackpot prophesy. Also, were the audience to grow too large, the likelihood of the reports appearing on the Internet intact increases. And with that comes the increased chance of the snake swallowing its own tail, as the search spiders find their own reports referencing back to their own dictionaries, which results in corruption of the data. (This actually happened with their first attempt to go public with a prediction.) So, the first rule of Half Past Human is: don't talk about Half Past Human.
The small size of the subscriber base (in the low hundreds) is also reflective of their marketing techniques. Their site states, "Most humans will wisely refrain from subscribing to an ALTA (Asymmetric Language Trend Analysis) series." Returning subscribers are thanked for their support, but cautioned that, "really, you should probably seek counseling now."
Hits and Misses
The group readily acknowledges that predictive linguistics is a fledgling field, and the reports themselves frequently summarize propositions with, "Then again, we're probably wrong." If their past track record at latching on to large events ahead of time is any indication, though, the data may be pointing to some interesting times in front of us, particularly as we approach summer 2009, which they've taken to labeling the "summer of hell" already.
And that track record does indeed include some pretty uncanny results. Their first public report came in July of 2001. Cliff approached George Ure of the site Urban Survival because he believed his analysis had crossed a tipping point for a world-changing event within the next 90 days. George primarily covers the world from a financial perspective, and Cliff was seeing a coming event focused around currency markets, the military, and an accident. Concerned about the possible impact, but also concerned for his own privacy, Cliff funneled his rough predictions to the world through George.
Being new to the process, HPH's focus on a mid-July timeframe for the event initially seemed to be a failure; until about six weeks later when the twin towers toppled to the ground in New York. "I instantly went into a state of denial," says Cliff. "I'll never forget the words when George called and said ‘Is this it? Is this the hit you've been waiting for?'"
Following that fated prediction, their timing and descriptive clues began to improve. They saw "an attack on house or assembly" before the anthrax attacks later in 2001; predicted an "attack on a commemorative event" prior to the American 587 crash on Veterans Day; and correctly identified aspects of the D.C. sniper case in 2002.
It's still difficult to make clear descriptions of events when you've reduced whole areas of complex human language down to a series of dots on a screen. In January 2003, the analysis kept coming back with a "maritime disaster over the West." This seemed cryptic, until the Columbia Space Shuttle disaster happened weeks later. Evidently the word "ship" as a pivot point could track sideways to "maritime" rather than "spaceship" as the data analysis plays Plinko with the dictionary.
Later, in July 2003, they had references to an "energy plant" associated with a "vertical wall" near an area where "rivers congregate." They were specifically searching for terrorist targets at the time, and found language around "dead man switches" and "switches/changes," that through "step by step" actions would leave people "blindly deluded" with a "fear of the dark" and feeling "tricked" and "exhausted."
Again, while it's easy to see in retrospect, it would be hard to predict from there that a series of cascading power failures would overload the Niagara power grid and result in the largest blackout in North American history. Still, just being able to accurately say that "something" big is going to happen around a certain time is pretty impressive for a garage-built time machine with no real funding.
Items coming out of the "SpaceGoatFarts" entity are particularly challenging, as these are things which we may well have no language for yet. This makes using the Internet as a tool to determine what the Universe is sending our way a difficult proposition at times.
The chief time monk explains, "The limitations of language surface, as it is probably impossible to use words bound by time to describe what is on the other side of the time horizon. Kind of like a Zen koan intended to shock the rational/monkey mind out of the way so enlightenment can express itself. So, says the rishi to the student, what color is not seen? What sensation is not felt? What sound is not heard? What time is on the other side of human?"
Recent smaller predictions are playing out well, though, indicating that their interpretation skills may be getting more detailed over time. As an example, they predicted a "uniting female personality" who would "emerge from the highlands of South/Central America" just as Ingrid Betencort was released from the jungles of Columbia. Ingrid, who has dual French and Columbian citizenship, went on a media blitz afterwards encouraging people to "bless their enemies."
I've also had "Turmoil in Pakistan?" written on my calendar on August 17 for a couple months now, based on their predictions, missing President Pervez Musharraf's resignation under threat of impeachment, and resultant unraveling of political coalitions there, by only a day.
When these positive hits appear, they're able to use them to fine-tune the engine by more closely matching the predictive language with the language used in actual media coverage of the corresponding event. "Following Ingrid Betencort's release, we got a ton of stuff around her and France, so now I've got a much better emotional base for the French," says Cliff. "It looks like I've got to add about 4,000 French verbs to the system, but we now have a much better idea of what constitutes ‘immediacy' and so on."
As always, though, timing remains a big challenge. In particular, how do you distinguish between a small event happening soon from a large event happening further out? Still, when asked how accurate the system is overall, "I believe we're better than what science calls ‘chance' by at least a factor of two," is Cliff's summation. "We get three to four big hits a year. Mostly we're wrong. Occasionally we're correct. But when we're correct, we're spectacularly so."
As a precursor to their overall label of "Transformation" for the year 2009,the linguistics are currently predicting a period of significant "release language" beginning around October 7, 2008. In their analysis, some words such as "frustration" are associated with building tension. Contrast this with words like "rage" that are associated with the release of tension.
Says Cliff, "The shift into emotional tension release language in October will be the first manifestations of some very large changes in the planetary social order. The limitations of language come into play here, as 'large' just does not cover the real nature of the changes. Monumental means that after the fact, in future generations, 'monuments' will be built to this time. And that may be the case, but for those of us living it, 'monumental' is just not the proper adjective to apply to these coming months/years from October of 2008 through December of 2012."
Life as a time monk is challenging at times when the full weight of what's appearing in the predictions comes into clearer focus-a focus they alone may have. Says Cliff, "It can be extremely depressing. If I had children, I don't know if I'd have the intestinal fortitude to read my own crap. But now and then I get feedback like 'Because I read your crap, my eight-year-old daughter is alive because we knew the floods were coming.' So, there's some small compensation for it."
Pie helps, too. The monks are firm believers that all reports should only be read following a nap, and with a big piece of pie. Cliff suggests marionberry or peach depending on the season, but ultimately the best kind comes down to an individual choice. I can personally attest to these benefits, and regardless of whatever does or doesn't occur in the future, I am thankful to them for reintroducing pie into my life.
With many acts of transcendence, there's often a level of personal discomfort preceding a later phase of advanced awareness. Perhaps what the time monks at Half Past Human have detected is the beginning of our collective purge as we work towards a new level of global enlightenment in the coming years. Bring on the locally-grown, organic blackberry pie.
Image by oddsock, courtesy of Creative Commons license.