dissenting.
This court’s opinion today destroys copyright protection for software that continually monitors computing machine behavior. The opinion also conflates methods used to protect trade secret information with the actual information constituting the trade secret. Because these holdings are contrary to the underlying law, I respectfully dissent.
The safe harbor created by § 117(c) is not a carte blanche license to use any program loaded into a computer’s RAM when a machine is turned on. Section 117(c)(2) specifically precludes a repairman from using copies of programs loaded into RAM upon powering-up that are “not necessary for that machine to be activated.” Maintenance code that continually monitors for faults, as does Storage Tek’s, is loaded into RAM upon powering-up the system but, as CHE admits, the maintenance code can be disabled with no affect on the operating aspects of the system. Of course, disabling the maintenance code eliminates continuous monitoring for faults. Even though Storage Tek has chosen to load the maintenance code upon activation, the maintenance code as such is incidental, not indispensable, to activation. Consequently, CHE’s use of copies of Storage Tek’s maintenance code falls outside the safe harbor created by § 117(c).
This court’s opinion holds that CHE’s copying and use of Storage Tek’s diagnostic “maintenance code” software falls within the protection of § 117(c) because “CHE ‘checks the proper functioning’ of the storage library and ensures that the machine ‘works in accordance with its original specifications.’ ” However, § 117(c) places restrictions upon the use of maintenance software. Therefore, while I agree that “maintenance” includes checking the proper functioning of components, I do not *1322agree that CHE’s use of Storage Tek’s maintenance code falls within the protection of § 117(c).
When using Storage Tek’s maintenance code, which is not code “necessary for the machine to be activated,” CHE does not reboot the storage silos of its clients for the sole purpose of making a new copy “by virtue of the activation of the machine.” If it did, CHE would be within the safe harbor of § 117(c). Although at the time CHE reboots it repairs nothing, adjusts nothing, and checks nothing, CHE subsequently accesses and uses the maintenance code to send data packets that indicate the operation of the system. Therefore, CHE does not fall within the safe harbor of § 117(c)(2).
Alternatively, even if Storage Tek’s maintenance code were so written as to be “necessary for the machine to be activated,” when CHE reboots the silos with that code, it once again repairs nothing, adjusts nothing, and checks nothing. Because CHE does not perform any “maintenance” or “repair” when it reboots the system to manipulate the maintenance code, this copying of the maintenance code does not fall within the protection of § 117(c)(1).
Section 117(c) allows a repairman to turn on a machine and to use the programs necessary to run the machine for the limited period of time the repairman is actually working on the machine, whether the repairman is fixing something that is actually broken, or servicing parts to prevent the machine from breaking in the future. Servicing parts may certainly encompass “putting the machine through the paces” to ensure that all the parts are properly functioning. In other words, the words “maintenance” and “repair” in § 117(c) extend the protection of § 117(c) to cover both the monitoring of function to assure that repair is not needed, and identification of malfunction to facilitate repair. However, the “maintenance and repair” must be of limited duration. Section 117(c)(1) specifies that any copy of the maintenance software must be destroyed “after the maintenance or repair is completed.” CHE, however, does not meet this condition. CHE runs the maintenance software continually to monitor operation. Only when a problem arises during monitoring, does CHE actually work on the silo. Thus, CHE uses the copy of the maintenance code in RAM beyond actual servicing or repairing. This continual use falls outside the scope of § 117(c).
As this court’s opinion states, the policy of § 117(c) is “to ensure that independent service organizations do not inadvertently become liable for copyright infringement merely because they have turned on a machine in order to service its hardware components.... The point of requiring that copies be ‘destroyed immediately after the maintenance or repair is completed’ was not to create artificial restraints on companies engaged in legitimate repair and maintenance activities, but to prevent persons from invoking the protection of § 117 and then later using the copied material for a prohibited purpose.” CHE, however, does not boot the machine in order to service or repair it; it boots to manipulate the maintenance level of the maintenance code so that it may read fault codes. As noted above, this is not “maintenance” or “repair” under § 117(c). CHE also does not “immediately destroy” the copy when the service or repair is completed. CHE uses the maintenance code even while the machine is functioning properly and is in full use by the client. How can CHE’s continual use of Storage Tek’s software during a three-year (or more) contract not be “using the material for a prohibited purpose”? Again, the “immediately destroy” requirement of § 117(c) protects use only during the limited time the repairman is actually working on the computer.
*1323This court also holds that the diagnostic information contained in the data packets created and sent by the maintenance code cannot be a trade secret because “the reason the machine is malfunctioning [ ] cannot possibly be considered a trade secret.” But it is the data packets themselves, not the physical operation (or mis-operation) that they describe, that is the trade secret. A malfunction may be independently discoverable, but that does not preclude the information which describes it from being a trade secret. In my eyes, this court’s analogy to a special “buy price” is flawed. As I see it, the message that the arm of a storage silo is “broken” is analogous to the “buy price” that this court states “cannot possibly be considered a trade secret.” The diagnostic information contained in the data packets, i.e., the information detailing precisely which aspect of the system is broken and how to fix it, is analogous to the information detailing why the “buy price” makes a good buy. That a third party may perform its own diagnostics on the broken system and discover the cause of the malfunction does not remove the proprietary nature of the inforfnation until the third party actually performs the diagnostics and discovers the malfunction. Trade secrets remain trade secrets unless and until another party discovers the information on its own, at which point the information enters the public domain and is no longer protected. The point is that the other party must perform the work to discover the information on its own instead of stealing the information from its competitor.
Finally, this court’s suggestion that the meaning of the fault symptom codes themselves may have been a trade secret if they had not been introduced into the public domain misses the point. The proprietary fault symptom codes are the language used to express the faults, so it is the correlation of fault and code that forms the lexicon by which the trade secret can be deciphered.
For the foregoing reasons, I respectfully dissent.