Memory competitions have different events; most of them revolve around memorizing digits, cards, faces, etc. over a defined period of time. However, world champions use established techniques like Person-Action-Object and the method of loci, indicating that the information efficiency of remembering narratives constructed out of objects is higher than that of memorizing the digits directly, even taking into account the time-consuming encoding technique.
Encoding digits into objects is a significant mental burden: the one-hour record for memorizing decimal digits is 4620 digits, which works out to 4620 lg(10) = 15347 bits/hr. The record for cards is 2530 cards, which is an information content of only 2530 / 52 * lg(52!) = 10975 bits/hr. This means that it's significantly harder to encode playing cards than words.
What would the limit be if entropy were the bottleneck, rather than encoding? That is, imagine the following memory discipline:
- Competitor is taken into a room with a powerful computer running software of their choice.
- A random sequence of bits, or compressed incomprehensible alien language, is received by the computer, which encodes it into some format. Maybe it's a movie of one's memory palace, or a fictitious oral history, or a written story, or a list of random words.
- Competitor watches the movie or other encoding.
- After 1 hour, competitor goes into a different room and has an unlimited amount of time to input information into a decoder computer, which reconstructs the original information.
The computer is allowed to use voice or holographic interfaces, error-correcting codes, etc. But brain-computer interfaces are considered cheating.
- What type of encoding would maximize the bits of information a human could memorize in an hour? Would it be unique for each person, or relatively universal?
- What is the estimated bandwidth in bits/hr of this encoding? (I'd guess that it's much less than the human conscious bandwidth limit of ~120 bits/s = ~400,000 bits/hr, unless unconscious memory systems are much higher-bandwidth than we think.)