In the still active Knowledge thread, Mung asks at 224: “Do we have a sufficient number of examples yet to decide what knowledge is not?”
This is sufficiently important to headline the response made at 225:
KF: >>[W]e have both criteria anchored in experience and insight to define knowledge in a weaker and by extension a stronger sense. Given how science is a major cultural enterprise, we see the importance of the weaker sense: knowledge is warranted, credibly true (and — for emphasis — reliable) belief.
Knowing is a key function of knowers, who must believe . . . accept sufficiently to rely on . . . what they know. But beliefs may be false or irrational. We need a particular subset. That subset has to be such that an individual may know per accepting credible authority (e.g. a dictionary, a qualified teacher, an expert) on perhaps simplified explanation or demonstration but authorities are no better than their facts, reasoning and assumptions.
Thus, collectively at cultural level there must be adequate warrant of credible truth (and so also, reliability). This is fallibilist, and progressive: if warrant fails sufficiently for some X, it should be corrected and surrendered as claimed knowledge even as knowledge on the subject advances to include that X is not known. Some replacement, Y may or may not succeed X.
In a strong sense, some few things may be incorrigibly, undeniably or more broadly self-evidently known, e.g. if one is conscious and so self-aware, that generic direct fact is infallibly known. Similarly, that error exists is undeniable. Something like 2 + 3 = 5 is self evident, known on understanding the claim i/l/o adequate experience of reality, known to be so necessarily, and that necessity is so on pain of patent, instant absurdity on the attempted denial. First principles of right reason pivoting on distinct identity [A vs ~A] are self-evident and foundational to reason, starting with LOI, LNC, LEM. Other more abstruse claims may be shown to be necessarily true on adequate experience, observation and logical, insightful reasoning per grounds and entailed consequents, e.g. much of mathematics short of imposing comprehensive axiomatic systems which are of complexity sufficient for Godel’s findings to apply: inherently incomplete and prone to possible incoherence.
Where, on even the fallibilist model, facts of empirical observation in science hold superior warrant to integrative theoretical constructs in general, which are prone to correction on fresh observation, so should generally be seen as aiming at being empirically reliable and possibly true models of some aspects of reality.
In some cases, theoretical entities not subject to direct observation (e.g. the electron) may be sufficiently warranted from multiple converging but essentially independent lines of evidence that they are taken as real. Thus, degree of warrant, reliability and fallibility must be reckoned with in pondering degree of certainty regarding truth of knowledge claims.
Reliability mainly comes from a sufficiently strong track record of successful prediction and/or fresh explanatory integration that moral certainty is achieved that one can and should trust the claim enough to confidently act on it in momentous circumstances.
Mere claimed consensus of community or of experts is not sufficient to warrant something as credible truth.
Nor are computer simulations to be regarded as though they were or are empirical observations of fact.
Likewise, there is no one THE scientific method that uniquely qualifies Science as a superior knowledge claim, as good methods vary across and beyond the sciences as conventionally so labelled.
Inductive reasoning per adequate support of a conclusion, often on experience, observation, record and testimony, can in many cases deliver knowledge up to moral certainty, but in general it is necessarily fallibilist and provisional. Hence the rise of sciences and other disciplines as collective cultural enterprises targetting credible, reliable, growing bodies of knowledge and best praxis; thus providing expertise. However, all such enterprises fall under the aegis of philosophical analysis, especially logic, epistemology, ontology, metaphysics and ethics. Indeed, reason inextricably involves responsibilities towards truth and right and therefore so also — it involves rational, responsible warrant — epistemology.
Truth, we can take per Ari: saying of what is, that it is; and of what is not, that it is not.
Things that fail of adequate warrant to give responsible, reasonable confidence of credible truth (and so, reliability) will fail of being knowledge.>>
More food for thought. And yes, the design debates have come down to this in part, as we clearly need to examine what knowledge is and is not, towards being able to decide what is and is not a proper knowledge claim. END
PS: One point that is worth a note is a claimed definition of knowledge that would eliminate knowers commonly advanced by frequent objector CR. So, let me also put up a footnote from 199 in the said thread:
Knowledge is information that, when embedding in a storage medium, plays a causal role in being retained.
This needs to be addressed on points:
>>Knowledge is information>>
1 –> Only when knowledge is specifically coded will it be information in the sense of a string of characters in a readable frame.
2 –> Often knowledge is intuitive, an implicit pattern that is recognised and used without being articulated in a coded form, e.g. knowing someone’s peculiar voice.
3 –> Similarly, many perceptions or beliefs are knowledge, e.g. of the threat posed by a car hurtling down a road towards you as you try to cross.
>> that, when embedding in a storage medium,>>
4 –> Code/representation centric again, and codes are themselves sophisticated expressions of knowledge, i.e. perilously close to question-begging.
>> plays a causal role>>
5 –> Stored information is passive, there is need to have active machinery for reading and using it, driving effectors.
6 –> this begs the questions UB has put on the table again and again regarding the architecture of cybernetic systems.
>>in being retained.>>
7 –> this again ducks the points that are pivotal.
8 –> Information as coded or modulated and stored etc is passive, and it is a wider system of processing that gives it effect.
9 –> When information in a system contributes to the system’s success, it is likely to be retained, but that is utterly different from that information being true or well warranted as credibly true. This is a key to modelling theory, models are successful simplifications of or metaphors for reality that may foster easier decisions and actions that can be successful without being true or even approximately true. think of electronic circuit models or economic models or the use of gear trains in a planetarium.
10 –> If stored information is somehow functional as processed and applied and so contributes to system success, this points to reliability, not to truth or to warrant or to knowledge.
11 –> System effectiveness is not truth or warrant as credibly true.
12 –> But CR will dismiss such a reference, so why do I make it?
13 –> Because of what he is diverting attention from. Knowledge is a term of language, and refers to a particular phenomenon that is commonly experienced and observed. Thus, there is a question of basic accuracy of concept.
14 –> Namely, knowledge is associated with knowers, who claim to know things they believe . . . perceive, strongly accept, firmly opine . . . are true or are at least credibly true and reliable enough to bank on when something of high value is at risk, up to and including life.
15 –> But belief is not knowledge, it is a component of it.
16 –> For belief to move into knowledge territory, it has to be adequately warranted as being credibly true and reliable, or even in the strong form as utterly certainly true and reliable.
17 –> Such warrant needs not be fully held by the knower, nor must s/he — it is conscious agents who know — hold the full argument or demonstration that warrants, it is enough that s/he holds on good authority that in turn is well founded in its claims.
18 –> We do this when we routinely look up words in dictionaries etc.
19 –> Founded is another trigger word, I simply point out that warrant comes in chains that for us finite knowers must be finite, and must not be circular, i.e. we start from finitely remote good first principles, observations, facts etc.
20 –> These must be able to hold their own on factual adequacy, coherence, reliability, ability to predict, coherence, explanatory power etc. Try mathematics out for size.
21 –> Such knowledge may well be reduced to coded form and embedded into systems that then can effect actions using that information and will perhaps be reliably successful by some measure. Thus there will be little reason to make changes in that information as such.
22 –> But this is all engineering, obviously downstream of what makes it knowledge.
23 –> In attempting to create what looks like an operational definition, we have a failure due to question-begging and distraction from the actual core meaning of the matter. Sad but unsurprising.>>