Quantity vs. quality
On newspaper columnists, unpopular right answers, and a Gordian knot problem
Technology columnist Christopher Mims of the Wall Street Journal observes, "One way I'm different than 10 years ago is, even when I feel I have some expertise on a topic, I'm hesitant to weigh in, because I know how *dangerous* knowing a little (but not the full story) can be", adding that his observation is "a thank-you note to those who are ready with deep domain area expertise and can articulately weigh in when their moment comes".
■ One of the distinguishing characteristics of knowledge in the digital era is the prevalence of quantification. It is as though the old managerial proverb, "If you can't measure it, you can't manage it", has been applied wholesale to how information is evaluated. It started, to some extent, when Google first began scoring websites on the basis of backlinks from other sites as a partial indicator of quality.
■ Quantification spread through all sorts of other mechanisms (upvotes/downvotes, for instance), and has reached its apogee (thus far) in the large language models that are so big and active that they are tilting the scales of electricity demand.
■ The comments from Mims value what we might call "encyclopedia knowledge" -- not what necessarily appears in an encyclopedia, per se, but knowledge that is valued because it has the approval of someone with credible authority. Encyclopedia editors don't put knowledge to a popular vote; they check with subject-matter authorities. An unpopular answer is no more right nor wrong because of its unpopularity. Quantification matters not one bit to the truth.
■ Our contemporary problem is that there is no way to readily reconcile the encyclopedia-knowledge approach with a quantifiable methodology like what drives an LLM.
■ The high-minded approach applauded by Mims is not only incompatible with quantifying or scoring knowledge, it's expressly contradictory. If the people with real knowledge intentionally hold back until a subject is squarely within their domain of competence, then they will usually be lapped by people running a race to be first and loudest. The resulting imbalance of content gives the advantage to quantifiable frequency -- saying things a lot, rather than waiting to say them with authority.
