Since Clif has been occupied with close, urgent family matters, plus prepping for next year’s whatevers, I’ve been thinking on how to get some kind of close-in sense of what it coming.
Coincidentally, one of our readers has come up with an interesting rehash of an old intelligence technique called “Word Frequency Analysis.”
The idea behind it came out of WW II, when it was noticed that certain word use shifted every so often and as a result, some inferences could be made about what Axis nogoodnicks were up to.
So what does WFA tell us about what’s going on right now?
As you can see, there is some advancement on simple word counts: Our contributor who’s a FoxPro jock sorts out the word frequencies into the basic elements of news stories: Who, where, why, action, what, how, and when.
Nowhere near the complexity of Prolog and word aspect/attributes, but better than nothing.
Not that specifics can be hard from this kind of data. The real value is in how it changes over time. Especially the timing notes. It’s asking a lot of the programmer, but the arrival of new words would be incredibly interesting, as well as changes in position. That is likely where the action would be.
Still, it is tempting to construct phrases out of various lines even though they are not specifically meaningful. “America, fearing a Chinese attack on the internet, might consider legislation this month” would be one way to parse it down to something logical.
Populate a Google search with “fear Chinese internet attack” and you get pages to news to ponder which are right along point.
Timing is another loosey-goosey concept. The idea I’m noodling is that the shape of the time-related phrases would change over time, and the change would either weight backwards (problem receding) or forward (problem getting closer).
If the first two time references (this month, tomorrow) had numeric values attached, a month would be 30-days and a day would be 1, so 31 days, so the public mind may be thought to be most preoccupied with something 15 days out.
Or, we could use the top three date references, divide by 3 and come up with a change-state that way.
So with some hints to our latest “see the future” using software contributor, such as reading of Sherman Kent’s work on “Words of Estimative Probability” we shall continue to ponder new ways to slice and dice the future in a nonpartisan, data intensive way. And we shall see what we shall see.
Plenty of existing work in the field (such as Kesselman, 2008) but there are also some very difficult concepts to integrate and that’s where the frontier is.
The basic problem might be reduced to noting that our collective future arrives as a “template” which slowly resolves, ghost-like, out of the future, and then solidifies into a mental picture in a sudden, often-jarring way, as did 9/11 which indelibly burned that “picture” of airplane hits building into that moment in time.
Yet the template for the events of that day stretched back years in advance: An unsuccessful bombing attempt of the building, the rise of military Islam, possible governments (and not necessarily US) having a hand in events, and so forth.
The difficulty of word frequency analysis, therefore, and to some extent predictive linguistics, is not much focus has been placed on the “glue” that binds template words to a particular precognitive image. In other words, at what point did the template for 9/11 begin to coalesce in a separate and distinct way from a number of other more conventional images such as a plane crash (or the routine type) or terrorist bombing (of the routine type)?
Lots of different approaches to be tried, in fact I’ve got a list of ideas I’m starting to collect. It’s just that there’s not much market for such work, especially if one wants to keep the work “clean” – and not focused on a particular set of templates which evolve into pictures, since that’s what the alphabet agencies and governments are concerned about. But what of the general (open source) case?
Like I said, lots to think about, but no one is renting my brain to do this stuff, so it takes a back seat to other things, like eating, for example. Ideas are welcome, though and it’s worth a few moments of reflection since the future’s going to arrive whether we want it to or not. Nice to know which direction it will be blowing in from.
World’s Gone Crackers
A new brand of crackers - Soylent Green – promises to deliver “Just what Grandmas used to taste…”
More tomorrow morning for Peoplenomics subscribers including a self-assessment of how “branded” you are…
Write when you break even…
George Ure (email comments to george@ure.net)
