Dear Lick:

	I have two thoughts on the large memory problem, one hard one
soft.  Intellectually, however, the hardware thought is softer than
the software thought.

	1. I have thought about how to store information in chain molecules
and how to retrieve it from such a store.  The memory is simply a bottle
that contains the molecules in solution.  The radicals in each molecule
encode both the address and the data.  Unlike crystal lattice schemes,
a regular array does not have to be maintained.

	Information would be stored in the memory by introducing a starter
molecule and sequencing its uptake of radicals from the substrate by
a macroscopic condition such as the presence of a given frequency of
microwave radiation.  Reading is accomplished by growing a side chain
whose sequence of radiation stimuli is such that it can grow to completion
only alogn the molecule with given address.  Then a multiplicative reproduction
proceess is started that will reproduce only the molecule with the parallel
chain attached.  When there are enough of them, the information can be
read by a microwave response.

	This is very sketch and I can do better if I think about it again.
However, what we really need is a chemist or chemical organization that
can do made-to-order chemistry of the quality that resulted in SX-70 film.
If you are interested, we can try to find such people.

	Since one unit (not necessarily just a bit) of information is
stored per molecule, a mole of the memory would store 6 x 10↑23 units.


	2. All the work I know about on large data bases has been based on
the idea of determining the structure of the data base and the programs
for manipulating the information.  However, there are already thousands