On May 29, 2007, in the case of Columbia Pictures Indus. v. Bunnell et al. (C.D. Cal. May 29, 2007), Judge Jacqueline Chooljian entered an unprecedented Order in which Defendants were ordered to begin capturing and preserving Internet Protocol (“IP”) addresses processed by their computer servers that were temporarily stored in volatile memory and not otherwise written to a permanent log file. Put another way, the court ordered Defendants to begin preserving digital information that resided on their computers for, at most, a few hours, after which it would normally be deleted or purged through an automatic process, and which would be instantly lost if the computer were restarted or turned off. Commentators have expressed considerable alarm at the idea that, for purposes of electronic discovery, the mere existence of information in a computer’s working memory could be considered “information stored in a medium from which it can be retrieved or examined” (Federal Rule of Civil Procedure 34). And obviously, a party’s obligation to preserve potentially discoverable information would become paralyzing if it were routinely extended to all information stored in volatile random access memory (“RAM”) as well as in more traditional file-based storage media.
A closer reading of the opinion, however, suggests that the Bunnell opinion may have less of an impact than any of these worst case scenarios. Indeed, one can make a plausible argument that this opinion is likely to be inherently limited to a fairly narrow set of circumstances.
The information in RAM was uniquely relevant to proving legal claims and could not be obtained by any alternate means.
Recognizing the pioneering nature of its Order, the Court methodically presented the grounds for its ruling. First, the Court emphasized the importance of the requested ephemeral digital information to resolving the dispute. “There can be no serious dispute that the Server Log Data in issue is extremely relevant and may be key to the instant action.” Order at 11. Second, the requested IP address information that the Court had found so highly relevant was not available from any alternate source. Taken together, these two findings of fact strongly suggested that this digital information, which had been specifically requested by the Plaintiff, should be produced even if the burden of collecting it and readying it for production was high. (In the Court’s mind, the relevance of the IP address also trumped certain personal privacy concerns-an entirely separate topic of discussion not covered here.)
The Court’s analysis, however, is extremely fact-specific. In Bunnell, the request for ephemeral digital information was justified through ample supporting evidence. Unstated but inherent in the Court’s analysis is the idea that a boilerplate demand to preserve all digital information, whether or not it specifically mentions information stored in volatile RAM, would likely not give rise to the duty to capture this fleeting digital information. Moreover, using Bunnell as a model, producing parties should have a full opportunity to vigorously oppose such discovery requests by demonstrating the burden (and possible impossibility) of complying with them. The fact that Bunnell permits discovery of ephemeral digital information in no way makes this a mandatory activity.
The information in RAM could be preserved without extreme effort
One significant limitation in working with ephemeral data is that few mechanisms are available to capture this information as it is created or being processed. Information in working RAM is typically stored in its most elemental and machine-friendly form, not as human-friendly data stored in discrete data files. Moreover, information in RAM is typically updated frequently-sometimes many times per second-as memory buffers are constantly cleared or refreshed to make way for new or updated information. Even if it were possible to do so, static snapshots of information stored in volatile memory cannot accurately, much less completely, capture the entire spectrum of information constantly shifting in and out of memory.
Bunnell provided a unique opportunity to request a specific type of information stored in volatile memory. Unlike most data values that are stored temporarily in a computer’s working memory, the requesting party was able to show that the IP addresses at issue were being processed by a program that already included the ability to automatically write this information to permanent log files. Indeed, plaintiffs were able to elicit testimony from the producing party itself that it was aware of the logging functionality of its server software and had chosen to turn the feature off as part of carefully developed business procedures. Given these facts, the Court felt competent to calculate the effort needed to capture this data in a format that could be preserved and analyzed. Relatively few software packages and systems contain similar functionality for the temporary information they create or process, though, and it is not at all certain that the Court would have reached its conclusions if the burden of preserving this information had required exotic and heroic measures.
The Court imposed this obligation only on a forward-looking basis and did not penalize the parties for past RAM-stored information not preserved
Bunnell presented the situation where active data stored in volatile memory was found to be uniquely important in resolving a legal dispute. Litigation largely concerns actions that took place in the past, and it is overwhelmingly likely that any relevant ephemeral electronic information has long since been purged by the time a party incurs a legal obligation to preserve potentially relevant information. While there may be an obligation for a producing party to check whether ephemeral data could be a potential source of relevant information, even the most superficial analysis is likely to determine that this information rarely contains relevant information or information that is not duplicated elsewhere in actual data files.
The Court acknowledged the fact that past ephemeral data could not be retrieved with any existing technology and ruled that the lack of prior effort to retain this information did not rise to affirmative spoliation of discoverable electronically stored information. Under all other available case law prior to the May 29 opinion, data loaded into volatile RAM would almost certainly fall into the “inaccessible” category of electronic information that the amended Federal Rules of Civil Procedure have sought to generally exempt from standing preservation orders. Moreover, the Court found that the producing party appeared to have acted reasonably in preserving other sources of generally-recognized discoverable electronically stored information. Given these circumstances, sanctions would not be appropriate.
Extrapolating from this analysis, Bunnell suggest that ephemeral electronic information will rarely rise to such importance that the producing party must take exotic measures to preserve this information. Indeed, as noted previously, the Court’s analysis is predicated on finding that the information in RAM had a specific and unique value in the litigation and could not be found elsewhere. The absence of sanctions in this case suggests that generic e-discovery document requests that include a demand for digital information stored purely in RAM will not automatically contain sufficient detail that this information should be preserved as a routine matter. Absent a detailed demonstration of why this information is critical for deciding a case, Fed.R.Civ.P. 37 and its state law analogs are likely to support producing parties if they maintain their ongoing operations by necessarily updating, changing, and otherwise altering and purging information in volatile computer memory.
Conclusion
Bunnell will be closely scrutinized by other courts in the months and years to come, particularly its use of a copyright infringement language regarding duplication of existing documents to create an expanded definition of original digital documents for e-discovery purposes. Other courts could very well reach different interpretations in deciding the extent to which non-duplicative loose data temporarily stored outside of a defined file format constitutes discoverable information that must be preserved. However, as this opinion is weighted for its precedential value to the e-discovery canon, practitioners are well advised to understand that yet another form of digital information must be analyzed when creating an inventory of potentially relevant and discoverable information.