Sunday, July 02, 2023

Modeling Language

The idea of a "word-meaning trajectory" resonates with some of the memes bouncing around in this Tai-Danae Bradley lecture.

I used to write about the word "Pepsi", atypical (i.e. an outlier) in the philosophy of language, generically averse to commercial brands.  

Highlighting Pepsi is shorthand for linking in Edward Bernays style "invisible persuader" propaganda budgets aimed at establishing a "market position" i.e. a place for some brand or meme in the collective mind of a stakeholder public.  That's where "rubber meets road" in terms of "revectoring".

Stakeholders include the potentially and/or actually negatively effected.  These are stakeholders nonetheless.  One may have a stake in opposing one advertising campaign with another. 

Expecting propagandists to just sit idly by while they see an underdog needing their help, is unrealistic.  Onlookers self selectively surrender their innocence (as bystanders) and join some ongoing fray.  They take a side, become partisans.  Pertinent figures of speech:  "getting sucked in"; "falling into a gravity well".

So where does a Pepsi or a Tylenol live in an LLM (large language model) and how do we machine learn to fully automate or at least cue finely-tuned repositioning maneuvers?  Grossly-tuned if needed.  Infants have similar needs for motor skills (and nutrition!) when learning to walk and/or to continue walking.

Keep a brand burnished and polished.  How?  Answer: keep introducing new usage patterns (permute the namespace).  That may be easier said than done of course.  Inertia levels may be high.  Weights pertain.  It's a balancing act, one of maintaining a qualitative as well as quantitative equilibrium, sometimes punctuated.

Other times, the wiser strategy is to just coast.  Why fix what ain't broke?  Why advertise what's taken for granted?  Why make billboards to advertise blue sky?

What are the mechanics of gaining traction and getting work done?  How does a media campaign leverage energy expenditure?

For example, how did Synergetics (Fuller) first inject and the revector the "Jitterbug" meme?  It came in as a dance,  somewhat in the Twist family, and Fuller capitalized on this existing trajectory to signify a geometric transformation.

Picture two tetrahedra on each side of a pinch point, like a bow-tie, left and right.  Then picture a stella octangula or merkaba wherein left and right tetrahedrons are centered and interpenetrating. Packed spheres anchor the context, much as cubes do in XYZ.  The icosahedron to cuboctahedron inter-twisting relationship rounds out this so-called Jitterbug Transformation.

From Tai-Danae Bradley I get a boost in seeing an ad campaign as a statistical challenge, i.e. to revector through "brute force" for example, whatever that means.  Say by means of "blanket coverage".  Omnipresence.  More like a Pfizer or LEGO might roll out.  

That's high budget advertising and one hopes a way of financially supporting a not-hostile relationship i.e. "at least you won't attack me if I'm sponsoring your program".  Funding buys non-negative coverage, is the press, although it's not that simple.  Sometimes a business or politician suffers because of disclosures about who the donors are.

Another strategy is to stay small and hard to find, such that the "surface area" or "perimeter" one needs to defend, is relatively minuscule.  Not every luxury is mass-market, by definition.  Instead of playing the most popular girl in school everybody knows, play at being mostly invisible and within some opaque clique.

Some companies are actually satisfied with their size and discourage sudden burgeoning, which could only mean uncontrolled.  Metamorphosis is sometimes in the cards, but if so that's likely presaged in the company's DNA e.g. some of its antecedents were gigantic, and/or a tycoon-minded heir has come along.  Many a small business deliberately stays small, with no desire to establish remote branches.  The same stakeholder may be involved with companies in different phases of their lifecycle.