[NB: Over at Copyrights & Campaigns, a "debate" has arisen in the comments to this post. This "debate" notes that the Computerworld magazine and website recently published an interview with Lime Group CEO Mark Gorton in which he attacked my most recent paper on inadvertent file sharing . I have thus attached, below, a slightly edited version of the reply that I provided to Computerworld weeks ago.]
I would like to thank Computerworld for asking Mr. Gorton to cite an example of what he called the "very tricky and deceptive" and "completely biased" analysis in my recent paper, Inadvertent File-Sharing Re-Invented: The Dangerous Design of LimeWire 5 ("Dangerous Design"). Predictably, only Mr. Gorton's example was "very tricky and deceptive": he misrepresented what my paper said and how his own program behaved.
Before my paper was published, Mr. Gorton was misleading the public, Members of Congress, and technology journalists with claims like this:
To understand first-hand the level of security we have achieved [in LimeWire 5.1] I encourage any member of the [House Committee on Oversight and Government Reform] to do a default install of LimeWire 5 on any computer and attempt to share a document file type: LimeWire will not permit it.
Mr. Gorton does not deny that Dangerous Design proved that this claim was dangerously wrong: a "default install" of any version of LimeWire 5 on "any computer" can not only "permit" but actually cause sharing of tens of thousands of family photos, scanned documents, home movies, popular music--and perhaps even documents--even if no version of LimeWire is presently installed on that computer.
So when Mr. Gorton said, "Sydnor says he went and installed LimeWire on a computer and that it immediately started sharing (all of the files) on the computer and that it was a security disaster," he admits that this can happen just because someone completed a default installation of LimeWire 5.2 on a home or work computer on which no version of LimeWire was presently installed. Nevertheless, Mr. Gorton argues that it was "tricky and deceptive" for me to refute his inaccurate claims by describing how LimeWire 5 can actually behave:
What [Sydnor] failed to mention is that in order to achieve that result, he had to have taken that computer and installed a previous version of LimeWire on it. He then, really, deliberately had to go and remove all of the security settings on it, ignore countless warnings and consciously share file by file all of the files he was talking about.
Mr. Gorton thus made three claims--two about what my paper allegedly "failed to mention" and a third about how his own program allegedly behaved. All three are false.
First, Mr. Gorton claimed that Dangerous Design "failed to mention" that LimeWire 5 will only share files by default if a prior version of LimeWire either is installed, (or had once been installed and then uninstalled), on a given computer. No honest reader of Dangerous Design could make that claim: Dangerous Design not only "mentioned" this, it discussed this "partial uninstall" behavior in detail and illustrated its real-world consequences. See Dangerous Design at 18-20. My other papers on inadvertent sharing, Filesharing Programs and "Technological Features to Induce Users to Share" and Inadvertent File-Sharing Revisited, did the same.
Next, Mr. Gorton claimed that Dangerous Design "failed to mention" the "countless warnings" that alert readers of 6-point type might see if they enable document-sharing in LimeWire 5.1. But were only two such warnings, and Dangerous Designs analyzed and displayed both of them in a screenshot-illustrated analysis of their many defects. See Dangerous Design at 12-13. Consequently, and once again, no honest reader of Dangerous Design could claim that it "failed to mention" the document-sharing warnings that it had discussed in detail.
Finally, Mr. Gorton claimed that to get a newly-installed copy of LimeWire 5.2 to share files by default, I would have had to have uninstalled a version of LimeWire 5.1 in which I had "consciously share[d] file by file all of the files [I] was talking about."
The quoted claim was (and is) false: LimeWire 5 users have never had to "consciously share file by file" every file that they end up "sharing." Indeed, if LimeWire really did require its users to "consciously share file by file" every file that they "shared," then I would agree that the problem of inadvertent sharing should be resolved for LimeWire users.
But every version of LimeWire 5 has contained dangerous sharing-related "features" that ensure that one common or reasonable mistake can cause mass-sharing of tens of thousands of a user's personal files, including their entire collections of family photos, scanned documents, home movies, and entire music collections, including all legally acquired audio files downloaded from Amazon.com or ripped from purchased CDs.
For example, my written testimony to the House Committee on Oversight and Government reform contained a screenshot showing what could happen if someone completed a "default install" of LimeWire 5.2.8 on a test computer that, like my main home computer, had 16,798 document, image, video, and audio files stored in subfolders of its My Document folder.
I did not create that screenshot by sharing 16,798 files "file by file." All I had to do was (1) complete a default installation of LimeWire 5.1.3, (2) enable document sharing (perhaps because I was a pre-teen who wanted to legally share some short stories), (3) go to an interface in 5.1.3 that seemed to have nothing to do with audio, video, or document files, (and nothing whatsoever to do with entire pre-existing collections of any type of file), (4) accidentally click "share all," and then (5) uninstall LimeWire 5.1.3 to remediate the resulting disaster. Had I skipped step #2, (enabling document sharing) then a subsequent default install of LimeWire 5.2.8 would only have shared 13,212 files by default, including entire collections of family photos, home movies, popular music, and many scanned or faxed documents stored as image files.
Indeed, Mr. Gorton and LimeWire LLC have never denied that LimeWire 5.1 exhibited a knowingly dangerous default behavior (loading all document, image, video, and audio files in My Documents and its subfolders into a "Library" that could be "shared" with one click, even in interfaces that seemed to have nothing to do with a user's pre-existing collections of personal files). Nor can they deny that this knowingly dangerous default behavior exacerbated the inherent risks imposed by a "share all" feature in LimeWire 5.1 that proved to be so indefensible that even Mr. Gorton now rationalizes its behavior as a very dangerous (but very obvious) "bug" that he and the entire "LimeWire team" somehow overlooked for eight months. Nevertheless, even after all these defects were known to both Mr. Gorton and LimeWire, they chose to ensure that LimeWire 5.2 would perpetuate all sharing of document, image, video, and audio files caused by the knowingly dangerous, concededly "buggy" LimeWire 5.1.
Computerworld reporters, editors, and readers can thus rest assured that I am now preparing to present to the House Committee on Oversight and Government Reform a detailed written response to every complaint about the analysis in Dangerous Design presented in Mr. Gorton's written or oral Testimony, or his representations to Computerworld. I will do so not because Mr. Gorton or LimeWire have identified any material defect or omission in my paper. Rather, I will do so to document further episodes indicative of LimeWire's bad faith or recklessness.
For example, Computerworld readers may have been relieved to hear Mr. Gorton make the following claim: "We have gotten right of the entire concept of shared folders, it doesn't exist anymore." After all, Mr. Gorton and LimeWire's trade association, the Distributed Computing Industry Association, spent the spring of 2009 telling federal officials and journalists that inadvertent sharing was caused by file-sharing programs that allowed what they called "recursive sharing" of folders, which their own Voluntary Best Practices define as follows: "'Recursive Sharing' means the automatic sharing of subfolders of any parent folder designated for sharing." Consequently, it would be great if, in the "latest version" of LimeWire 5, "the entire concept of shared folders... doesn't exist anymore."
But in LimeWire 5.2, recursive sharing of folders does exist. After Dangerous Design identified how the knowingly dangerous default librarying behavior of LimeWire 5.1 interacted with its allegedly "buggy" "share-all" feature, LimeWire hastily discontinued these behaviors in LimeWire 5.2--and then it just as hastily re-introduced default recursive sharing of subfolders. My testimony (p. 13) provides a screenshot that shows LimeWire 5.2.8 enabling default recursive sharing of subfolders.
Granted, this screenshot appears to depict multiple violations of LimeWire's own Voluntary Best Practices, (see 3(A), 3(B) and Def. (2)), but it also accurately depicts the still-existing actual behavior of all versions of LimeWire 5.2 that I have studied to date, 5.2.8, 5.2.10, and 5.2.12. Consequently, what Mr. Gorton said of me might be better applied to him: "He manages to parse the facts cleverly enough to give a highly misleading picture of reality."
NB: Unless otherwise noted above, the following terms refer to the following versions of the LimeWire 5 file-sharing program. References to the behavior of "LimeWire 5" refer to behaviors of the non-beta version of LimeWire 5 that I have studied: LimeWire 5.1.1, 5.1.2, 5.1.3, 5.1.4, 5.2.8, 5.2.10, and 5.2.12. References to the behavior of "LimeWire 5.1" refer to 5.1.1, 5.1.2, 5.1.3, and 5.1.4 (which LimeWire is still actively distributing here). References to the behavior of LimeWire 5.2 refer to 5.2.8, 5.2.10, and 5.2.12. My paper, Dangerous Design focused on the behavior of LimeWire 5.1, those knowingly dangerous versions of LimeWire 5 that Mr. Gorton had claimed "put the final nail in the coffin of inadvertent sharing..." and "ensured the complete lockdown of the safety and security of LimeWire users...."