Re: what does a patent restrict?

On Wed, 13 Oct 1999, Mark Lemley <mlemley[_at_]mail.law.utexas.edu> wrote:
>
> Actually, the law says "fixed in a tangible medium," which section
> 101 defines as "sufficiently permanent or stable to permit it to be
> perceived, reproduced or otherwise communicated *for a period of
> more than transitory duration."* The court ignored this last phrase
> entirely. The House Report explaining this provision says the
> following: "The definition of fixation would exclude from the concept
> purely evanescent or transient reproductions such as those projected
> briefly on a screen, shown electronically on a television or other
> cathode ray tube, *or captured momentarily in the memory of a
> computer* H.R. Rep. No. 94-1476, at 52-53 (1976).

Playing devil's advocate, 17 USC 101 states that:

    A work is ''fixed'' in a tangible medium of expression when its     embodiment in a copy or phonorecord, by or under the authority of     the author, is sufficiently permanent or stable to permit it to     be perceived, reproduced, or otherwise communicated for a period     of more than transitory duration. ....

The problem is that though DRAM operates in a somewhat transitory manner, the fact that it feeds back upon itself (i.e. is refreshed by being rewritten), data is often present for more than a transitory duration in DRAM. Of course, in the case of other memories such as SRAM (mostly used in cache memories), there is not the destructive refresh, and thus the presence of the data is in many instances definitely not transitory.

So, if you abstract a level, and look at it as a black box, ignoring the physics inside, if you ask can it be perceived for a period of more than transitory duration, the typically yes for even DRAM. Code on some of my computers has been resident and visible in the DRAM of those computers using the right tools for at least 6 months now. Even in my heavily used personal computers, data is perceivable, capable of being reproduced, capable of being communicated from my DRAM for hours, if not days on end. Hardly of transitory duration.

The following is a quick course in basic memory technology. If you understand the differences between different types of memories, you can skip this.

As a quick lesson in computer architectures, when you talk about memory in a computer, you are really talking about three different things: Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), and Read Only Memory (ROM). There are also some variations on these themes, such as EPROMs, EEPROMs, etc., but for our purposes here, they can be consider ROMs.

ROMs are typically fixed at the factory. EPROMs, EEPROMs, etc. are typically fixed a bit later, but fixed they are. There is no question about transitory duration here - they do not change until reprogrammed (if ever).

RAM is typically considered to include DRAM and SRAM. Dynamic Random Access Memory (DRAM) constitutes the bulk of the memory in most computer system. If you purchase a system today with 64 mb or 128 mb RAM, you are talking some variant of DRAM. The problem from a copyright point of view with DRAM memory is that it typically has a destructive read, and must be refreshed. Both require that a DRAM cell be destructively read, then rewritten. The reason for the refresh requirement is that DRAM cells are essentially small capacitors which discharge over time. The trick is that they are refreshed (requiring a destructive read and subsequent rewrite) before they can discharge sufficiently not to be read. If you wait too long to do this, then you cannot distinguish between a voltage high or low state (or a charged and uncharged state).

Static Random Access Memory (SRAM) is typically utilized as the cache memory in most computer systems. Very high performance systems may utilize SRAM instead of DRAM as their main memory due to the much higher speeds inherent in the technology. SRAM can typically operate at or near the speed of a processor, and indeed higher performance processors these days typically include embedded SRAM. SRAM this by being composed of gates (transistors) just like processors. SRAM cells are akin to latches and flip-flops in that they provide a selectably stable output signal regardless of input signal, until they receive a write signal, at which time they latch typically the value of their input signal. They may provide the same output signal for one machine cycle, or for months, depending on whether or not their clock (or write) signal is asserted (i.e., whether or not they are rewritten).

The reason that DRAM is utilized in most computer systems as opposed to SRAM is that each cell is typically much smaller, resulting in a much lower per cell (or per mb) cost. SRAM is utilized for its speed, since it typically operates at or near the speed of the processor, which in many cases is typically several to many times the speed of DRAM.

But as noted above, SRAM typically is utilized as cache memory. Cache memory is typically utilized to contain a fast access version of a fairly small subset of slower, less expensive memory such as DRAM. Thus, the contents of DRAM that are being read, or more importantly being changed, will very often also reside in one or more cache memories.

As a note to the above, DRAM and SRAM cells roughly correspond to bits, with typically 8 bits to a byte, etc. The reason that I say roughly is that some cells are utilized for error recovery, and some provide redundancy, and thus do not provide usable bits of memory to users.

-- 
--------------------------------------------------------------------
The preceding was not a legal opinion, and is not my employer's.
Original portions Copyright 1999 Bruce E. Hayden,all rights reserved
My work may be copied in whole or part, with proper attribution,
as long as the copying is not for commercial gain.
--------------------------------------------------------------------
Bruce E. Hayden                      bhayden[_at_]acm.org
Phoenix, Arizona                     bhayden[_at_]ieee.org
                                     bhayden[_at_]copatlaw.com