First of all, I'm for truth. Even though I will hold here to my contention that Kuhn's paradigm theory is pretty spot-on. (See The Structure of Scientific Revolutions.*) So if we are going to rescue ourselves from those objective types who can't wait to impale the naysayers--the mealy-mouthed who think life is but a dream--with their wagging fingers, we have to make an attempt to define what the truth is. A dream?! Pah! Here's a quick and dirty account.
The issue is seen clearly if we ask ourselves, as we navigate through the network of 'facts' that comprise our own personal narrative, "What are we going to spend our time learning?" Of course we'd rather learn true things, right? We will see that truth is really about interconnected language.
Let's compare two motifs of theory making, or, depending on our level of confidence, learning: We can learn about black people and white people, about race and about groups--anecdotally and through loose conversation. Or we can learn about melanin, eumelanin, pheomelanin, and sociology. Which is framework true? Is it the second, scientific one? If so, why? Remember that very paradigm and all the science that follows from it is a model waiting to be toppled. What ends up determining the amount of truthiness, then, while often scientific, does not lay prostrate before science. (Although I will say that the god of the 21st century has traded in his white robe for a white labcoat, and, as always, those who most fervently believe in his magic are the ones who least understand him.) No. The most we can say of a statement of fact is that its truth depends upon the degree to which that purported fact is interwoven with others. Is fact A an anecdote? A guess? How insular is it? Is it a groundless myth? Or is it instead something rooted in a broad and deep theory, in other facts?
If it is the case that fact A derives from the last in the above list then I dub it "true," and I do so because something very remarkable happens when we buy in to a protracted, thoughtful exercise in the production of thought; it's the same thing that happens when we labour over symphonies or spend hours each day as Coltrane did learning to improvise well--we do good work. This happens because human beings are not unlike the computers we have made in our image. We can "machine learn**" a great number of things, but for the knowledge to be worthwhile it must be predicated upon and presided over by an engine of logic. The better the logic engine and the longer it is allowed to run the better the machine's output in the end. Lousy facts have a way of weeding themselves out; they won't stick with many other facts. For an example, take the musings of a schizophrenic. She might create a very long string of internally consistent delusions, but eventually that consistency runs out and the narrative crumbles. It is no accident, perhaps, that those who suffer from schizophrenia often invent spontaneous words, the least sticky and most disconnected of all linguistic bodies. What has happened, as we sometimes say to one another, is that she's lost logic. (We would do well to always ground ourselves in fact--in logic--lest our ignorance lead to insanity.)
I argue that human logic is internal, and a well-supported fact is precisely what happens the more we think about something, the more we allow our "program" to run. Our internal logic is the only means by which a thought is ever vetted and its appearance is what makes something ultimately true. Still, truth needs to be fully defined before we're done. The game we are playing has not very much to do with the incremental march from the dawn of our ignorance into the historically inevitable light of truth, which is the notion that Kuhn argued against; it has everything instead to do with instances of a priori synthesis***, or as I would have it, linguistic interconnectedness. The more nexus that become linked by strands of logic, the greater the number of ideas fashioned firmly together, the stronger this lattice of language becomes. And so truth at last is a matter of degree; but it is one that depends (perhaps reassuringly) upon a magnitude of connections in a linguistic web that is threaded with logic--the stuff of all dreams.
* Basically, he argues that every scientific model is only a framework that's "truthiness" is really more about efficacy, since our perspective might change fundamentally, and tomorrow, in support of a totally different theory.
** Machine learning involves teaching a computer to recognize an object based on a preordained set of patterns. And here the analogy breaks down. Though we could try to make comparison between the patterned object and either some type of cultural or semiotic object--something I would tend to to but won't do here--or to neurological pathways. These are two completely different routes for completing the analogy: the first of which I would argue collapses eventually into the second, and the second, in practice at least, has already been preempted by neural networks, a completely different form of AI that attempts to imitate the structure of the human brain.)
*** A shambled two-second definition might be: Kant's word for logic + impressions of the world
