On Knowledge II: Individual Knowledge

In the article "On Knowledge I, " I estimated that the size of our culture's knowledge base is less than 100 gigabytes. To approach the question of individual knowledge, again requires making the distinction between knowledge and relevant data categories, sometimes called information.

First, what is the relationship of knowledge to relevant data categories? Knowledge is what lets us form sentences. Relevant data categories include the body of words we use. Words are clearly relevant data categories not raw data, because the word chair, for example, can be applied to thousands of different shapes that we see during our lives. One chair would be data.

The average person uses less than 105(100,000) (bytes) of mental space for the available words in his or her language in the course of an entire life. The brightest people use less than 106 (a million bytes), while the best dictionaries have less than 107 with all the available words in a language, plus descriptors. Some people know several languages, but the total number of the words they know is still fewer than 106 bytes of all the available words in those languages.

The knowledge necessary to use these words is not great. Imagine a six-year-old with a volume of words of less than 104 (10,000 bytes), the child can speak well and imaginatively. My guess is that the child's knowledge, which is learned at the same rate, month by month, can't be much more, maybe 105 bytes. The dumbest functional person can operate on the level of a six year old and probably get by on less than 104 bytes' worth of words and 104 knowledge of language. I assume that the knowledge for using words, the context for words, can only be one magnitude greater than the words in a child.

When we get to the 105 bytes for storing words considered above for the average person, I think it is reasonable to use at m,ost double that same amount for knowledge of word usage. The average person's language base of words, plus the knowledge to use those words, would still be well under 106 bytes. The dumbest functional person would be less than 105. The smartestwould be under 107. Only two orders of magnitude separate the dumbest people from the smartest people in the use of language.

Let's check this. The best minds studying intensely for four years in a foreign language (In my mind I've chosen rabbinical students, who must memorize the entire Old Testament and commentaries in Hebrew, as my conceptual measure) never achieve more than 107th bytes of permanently stored relevant data categories. That is under ten million bytes.

I would guess that the relevant data categories needed to get a 1,600 score on the SAT, which would take the best minds studying for five years from age 10 to 15, would not require more relevant data categories than 107th bytes, and the knowledge to achieve this 107 would equal a combined total of 2x107 bytes (20 million).

Short-term memory can be measured in a similar manner. The best editor of a 600-page book or a one-hour film can remember nearly all of it for months, but the magnitude of such a memory is probably under 107 bytes. Visual and written data are made comparable in magnitude by using the mechanism for comparison. The best visual editor can work with an hour's worth of final material, drawn from eight hours of raw data, while a book editor can work with 600 pages on a wide-ranging subject drawn from three times that number of pages.

Comparing the best minds to the average mind is not too difficult. I estimate the difference to be one order of magnitude, or about four standard deviations at the level of bytes we are considering. The relationship of the average mind to one that is four standard deviations greater is the same as a comparison of the average 35-year-old who has sex twice a week to the most active 35-year-old person who has sex thirty times a week (for fun, not money). That means that the average person's knowledge is 106 bytes and the relevant data categories the average person has readily at hand is 106 bytes: a total of under 2x106.

Since everyone has a different range of daily experiences that are nonverbal, and probably much of this is nonvisual&emdash;consisting of movement, smell, and sound&emdash;this reservoir of experience exists in the mind in some ratio to the verbal and visual relevant data bases. Whatever it is, it can't be very large. I will arbitrarily consider it to be the same size as the communicable domain of verbal/visual, or106 bytes for the average person.

The nonverbal, nonvisual realm can't be measured satisfactorily, because it has no descriptors; what I offer is a rough estimate. Zen monks and wise hermits who focus intensely on this domain with no descriptors don't seem to translate their own, obviously large magnitude of this nonverbal/nonvisual realm into much verbal or visual material. Basho's famous calligraphic circle is an example of one such translation: Though it is wonderful and evocative, it hardly reflects a mind space of great magnitude.

So the average person with verbal relevant data, knowledge to use it and non-verbal relevant data has less than 3x106 bytes of mental storage space. That is a total that is still less than 3 megabytes for all experience, including verbal and visual. If all the material being stored could be digitized, it would total well under10 megabytes.


We have many average people walking around our streets with knowledge and relevant data categories of less than 3x106 bytes, about five dense books worth. This is based on a common language pool of 105 bytes. The very brightest are only in the 107 range.


Michael Phillips, 1997 (revised 2000)


Home |Public Radio |Clear Glass Press |Resume |MP's Books |Articles |Misc |E-mail|