I find Wikipedia quite useful, in the proper context, chiefly when I'm trying to find out about (1) something that's not terribly controversial, so the risk of Wikipedia error or spin is likely lower, (2) and — more importantly — something that's not very important to my work, so the cost of possible Wikipedia error is lower. We rely on plenty of less reliable information under those circumstances; for instance, we might rely on our own recollections' of partly forgotten sources, or conversations with friends. There's no reason to categorically disqualify Wikipedia in those situations.

On the other hand, some uses of Wikipedia strike me as unsound. Consider this, from a law review article:

The history of the concept of "responsibility to protect" [FN1] sounds almost like a fairy tale. The International Commission on Intervention and State Sovereignty developed this concept in its 2001 report The Responsibility to Protect.... In December 2004, this idea was taken up in the context of the debate on United Nations reform.... In March 2005, this finding was endorsed by the report of the UN secretary-general .... [More historical details omitted. -EV]

Not long ago, the notion of "responsibility to protect" was added as a key word to the Wikipedia Free Encyclopedia, where it is defined as a recently developed concept in international relations that aims at "provid[ing] a legal and ethical basis for 'humanitarian intervention.'"

The articulation of the concept of responsibility to protect is a remarkable achievement.

The seeming success of the movement to recognize a responsibility to protect may well be a remarkable achievement generally — but having something "added as a key word to the Wikipedia Free Encyclopedia" is not a remarkable achievement: It can be done by one person anywhere in the world, with a modest investment of effort. It's even less of an achievement than being nominated for a Nobel Peace Prize.

Note also how the passive voice (another thing that I've defended, when it's used in the proper context) hides the unsound argument. The notion of responsibility "was added" to Wikipedia, the article says. But who did the adding? Some unknown and unscreened contributor.

Recasting the sentence in the active voice, as "Not long ago, a Wikipedia contributor added the notion of 'responsibility to protect' as a key word to Wikipedia," would have exposed the error. The passive voice conceals the error, especially to readers who don't fully understood how Wikipedia works. (And even if the author himself didn't fully understood how Wikipedia works, trying to recast the sentence in the active voice might have led him to figure the matter out.)

Here are a few pagraphs that I plan to add to my Academic Legal Writing book on this subject; comments are welcome.

Over two hundred student articles cite the online Wikipedia encyclopedia. Unlike with most encyclopedias, anyone is allowed to create Wikipedia entries, and generally to update existing entries. An unorthodox approach for an encycloped, but the theory is that (1) those people who want to spend time writing entries tend to be knowledgeable, and (2) even when they err, the errors end up getting corrected by others.

Surprisingly, the theory works, most of the time. Wikipedia entries tend to be relatively accurate, probably no worse and possibly better than the typical newspaper article. (This is especially so given that many newspaper articles are written by generalist reporters who are relying on hastily assembled materials from others.)

Nonetheless, while Wikipedia may sometimes be a good place to look, I advise you not to stop looking there. Instead, find the original sources that the Wikipedia entry’s author relied on — they’ll often be cited in the entry — and read, quote, and cite them.

First, that’s the standard procedure you should use for intermediate sources (including, as I said before [earlier in the book chapter], newspaper articles). Second, whether or not Wikipedia is more reliable than the typical newspaper article, many readers will assume that it’s less reliable; citing to it may thus decrease your credibility.

UPDATE: I at first noted that "I don't feel the need to mention that Wikipedia's contents may change over time, since I endorse citing to original-source Web pages, while recommending that the author print, save, and possibly even post and link to a copy of the page as of the time the article is written." D'oh! Forgot all about Wikipedia's change tracking system, which will let readers see the page as of any particular date (usin the oldid= feature). Thanks to Dan Lewis and RichardP for pointing this out.

Courts have cited Wikipedia over 300 times, and many of those cites are in my view just fine when the citation is for a tangential and uncontroversial matter. But the Seventh Circuit's use of Wikipedia in Rickher v. Home Depot, Inc., handed down Monday, strikes me as troubling.

The key issue as to one part of the plaintiff's lawsuit was the definition of "wear and tear." The plaintiff cited Webster's II New College Dictionary and Random House Webster's College Dictionary, which defined the term as “Depreciation, damage, or loss resulting from ordinary use or exposure” and “Damage or deterioration resulting from ordinary use; normal depreciation,” But the court disagreed:

Although it is true that dictionary definitions of “wear and tear” often employ the word “damage,” that does not mean that damage and “wear and tear” are synonymous. Wear and tear is a more specific phrase that connotes the expected, often gradual, depreciation of an item. See Wear and Tear, http://en.wikipedia.org/wiki/Wear_and_tear, last visited May 30, 2008.

It is a form of depreciation which is assumed to occur even when an item is used competently and with care and proper maintenance. For example, friction may erode a hammer’s head. In the normal use of a hammer for its designed task erosion is impossible to prevent, and any attempt to eliminate this erosion would make the hammer useless. At the same time, it is expected that the normal use of a hammer will not break it beyond repair until it has gone through a certain amount of use.

A subtle difference, but one the Seventh Circuit thought to be quite important, and that does indeed appear to me important to the course of litigation. (The question in this part of the lawsuit was whether Home Depot's Damage Waiver rental contract provision — for which one had to pay more money — added anything beyond what Home Depot already provided for no extra charge under its Wear and Tear provision. If "wear and tear" was roughly synonymous with "accidental damage" but not through "misuse or abuse," then the Damage Waiver might be seen as pointless, and offering it for money might then be seen as a deceptive business practice under Illinois law. If the "wear and tear" was a narrower term than "accidental damage [without] misuse or abuse," as the court concluded, then the Damage Waiver would give the customer something, and offering it for money wouldn't be a deceptive practice.)

Now I strongly suspect that the judges cited the Wikipedia entry because it fit their preexisting understanding of what the phrase meant (an eminently reasonable mode of procedure for citations generally), so I doubt they relied on Wikipedia's wisdom to form their conclusion. Still, the parties obviously disagreed about the matter. The plaintiff's proposed definition was supported by dictionary entries (albeit shorter ones, which might have omitted important nuances). If the judges wanted to argue based on their experience, based on logic, or based on contrary lexicographic authorities — including, for instance, the use of the phrase in other sources — that's fine, and they did that in some measure. But they cited Wikipedia as the lead authority supporting their conclusion, and as the source for their important and controversial definition; and this strikes me as troubling.

First, there does seem to me to be a serious risk of manipulation by the parties in this sort of situation. The quoted part of the definition was added on Aug. 31, 2005, when the case was in progress at the district court. I have no reason to think that the change was made by anyone associated with the litigants (and the Wikiscanner check reveals nothing tell-tale, even when I check all changes from that IP address), but neither can we be sure, I think, that no such manipulation took place. And while it's important not to overestimate the risk of manipulation here — as I pointed out, the judges are likely relying on Wikipedia to support their preexisting understanding of this quite common term, rather than as an expert source that would provide such an understanding — there does seem to be some danger here. It seems possible that the judges, who after all quoted the definition as authoritative, would indeed be influenced by nuances of the definition even if they already agreed with the definition's main thrust.

And, second, I don't see much reason to see why, even unmanipulated, Wikipedia should be a substantial authority here. We don't know who wrote the definition, so we can't rely on his knowledge. This doesn't seem likely to be the sort of definition that would attract a great deal of attention and review in case of error, so that we can rely on a possible "wisdom of crowds." Dictionaries and encyclopedias aren't perfect, and I know there are arguments that Wikipedia is on balance roughly as accurate as the Encyclopedia Britannica (as well as arguments in response). But it does seem to me that, at least until such rough equivalence of Wikipedia and other sources is further demonstrated, courts should rest their decisions about important and controverted matters on sources — such as dictionaries, technical dictionaries, or encyclopedia entries — that at least have some more indicia of likely expertise.

Again, I should stress that for tangential and uncontroversial matters, Wikipedia may be quite good enough. Federal employees' time isn't unlimited, and tracking down authoritative sources to demonstrate the colorfulness of Polish boxer Andrew Golota (to give an example from another Seventh Circuit case, which cited Wikipedia to support such an assertion) is probably not the best ways to spend that time. But for something like the controversy in this case, I would think that the lead authority should be something other than a Wikipedia entry.

For some earlier thoughts of mine on Wikipedia and court opinions, see here. Thanks to my friend Steve Newman for the pointer to the Rickher case.

UPDATE: Link fixed, with thanks to Ted Frank.

From Badasa v. Mukasey (8th Cir. Aug. 29) (paragraph break added, most citations omitted:

Lamilem Badasa entered the country illegally using a fraudulent Italian passport. She later applied for asylum under 8 U.S.C. § 1158 and for relief under Article III of the Convention Against Torture. The Immigration Judge (IJ) found that Badasa had submitted fraudulent documents designed to establish her identity, and that her claim was not credible. The Board of Immigration Appeals (BIA) initially dismissed her administrative appeal, concluding that Badasa had failed to establish her identity. Badasa moved to reopen her case based on a travel document recently acquired from the Ethiopian government, known as a laissez-passer, which Badasa alleged would establish her identity. Noting that the Department of Homeland Security (DHS) concurred in the motion, the BIA reopened the case and remanded it to the IJ for further consideration.

On remand, the DHS submitted several documents designed to explain the purpose of a laissez-passer, and argued that the document did not establish identity and nationality, but rather was “simply the granting of the authorization for an alien to travel to or from that country.” After considering evidence presented by the parties, including information submitted by the DHS from an Internet website known as Wikipedia, the IJ found that the laissez-passer is a single-use, one-way travel document that is issued based on information provided by the applicant. On this basis, the IJ concluded that the Ethiopian government’s issuance of the travel document did not change her prior decision regarding Badasa’s failure to prove her identity, and therefore denied the application for asylum.

The BIA dismissed Badasa’s appeal, concluding that the IJ’s determination that the laissez-passer travel document was insufficient to establish Badasa’s identity was not clearly erroneous. The BIA stated that it did “not condone or encourage the use of resources such as Wikipedia.com in reaching pivotal decisions in immigration proceedings,” and commented that the IJ’s decision “may have appeared more solid had Wikipedia.com not been referenced.” The BIA declined, however, to find that Badasa was prejudiced, because without considering Wikipedia, the BIA believed the IJ’s conclusion “was supported by enough evidence to find no clear error.”

We conclude that the case must be remanded for further proceedings, because the BIA failed adequately to explain its conclusion that Badasa did not establish her identity. The BIA did not adopt the entirety of the IJ’s reasoning for rejecting Badasa’s claim. Rather, the BIA acknowledged that it was improper for the IJ to consider information from Wikipedia in evaluating Badasa’s submission on remand, and the government does not dispute that conclusion here.

Wikipedia describes itself as “the free encyclopedia that anyone can edit,” urges readers to “[f]ind something that can be improved, whether content, grammar or formatting, and make it better,” and assures them that “[y]ou can’t break Wikipedia,” because “[a]nything can be fixed or improved later.” Wikipedia’s own “overview” explains that “many articles start out by giving one -– perhaps not particularly evenhanded -– view of the subject, and it is after a long process of discussion, debate, and argument that they gradually take on a consensus form.” Other articles, the site acknowledges, “may become caught up in a heavily unbalanced viewpoint and can take some time -– months perhaps -– to regain a better-balanced consensus.” As a consequence, Wikipedia observes, the website’s “radical openness means that any given article may be, at any given moment, in a bad state: for example, it could be in the middle of a large edit or it could have been recently vandalized.” The BIA presumably was concerned that Wikipedia is not a sufficiently reliable source on which to rest the determination that an alien alleging a risk of future persecution is not entitled to asylum. See also Campbell v. Sec’y of Health and Human Servs., 69 Fed. Cl. 775, 781 (Fed. Cl. 2006) (observing that a review of the Wikipedia website “reveals a pervasive and, for our purposes, disturbing set of disclaimers”); R. Jason Richards, Courting Wikipedia, 44 Trial 62 (Apr. 2008) (“Since when did a Web site that any Internet surfer can edit become an authoritative source by which law students could write passing papers, experts could provide credible testimony, lawyers could craft legal arguments, and judges could issue precedents?”).

The BIA did say that Badasa was not prejudiced by the IJ’s reliance on Wikipedia, but it made no independent determination that Badasa failed to establish her identity. Whereas the BIA sometimes applies a “harmless error” standard when an IJ considers improper evidence or makes other procedural error, and thereby evaluates whether the error affected the IJ’s ultimate conclusion, the BIA here determined only that there was sufficient evidence, other than Wikipedia, to establish that the IJ’s finding was not “clear error.” This is the correct scope of review when an IJ has made findings of fact based on proper evidence, but application of the deferential “clear error” standard to this situation leaves us without a determination by the agency as to whether Badasa proved her identity. We know only that the BIA thinks that if, hypothetically, the IJ had not considered Wikipedia and reached the same conclusion, then that conclusion would not be clearly erroneous. But we do not know whether the IJ would have reached the same conclusion without Wikipedia, or whether (and, if so, why) the BIA believes that the IJ’s consideration of Wikipedia was harmless error, in the sense that it did not influence the IJ’s decision. Because the BIA’s ultimate conclusion that Badasa failed to establish her identity is not adequately explained, we must remand for further proceedings.

Many thanks to David Schwartz for the pointer.

From Flores v. Texas (Oct. 23, 2008) (unpublished memorandum):

Appellant’s first three points of error relate to the trial court’s exclusion of evidence requested by appellant. That is, appellant complains about the exclusion of evidence concerning (1) the “John Reid” technique that allegedly results in false confessions, (2) the “circumstances” surrounding appellant’s written confession, and (3) his wife’s conversations with police, which were said to be contained on a compact disc. We hold that appellant has failed to preserve these complaints for appellate review.

In order to preserve a complaint concerning the exclusion of evidence, a defendant generally must make an offer of proof or file a bill of exception to make the substance of the evidence known. See Tex. R. Evid. 103(a)(2); LaHood v. State, 171 S.W.3d 613, 621 (Tex. App.-Houston [14th Dist.] 2005, pet. ref’d). Otherwise, as here, we cannot assess whether the exclusion was erroneous or harmful. LaHood, 171 S.W.3d at 621. However, appellant failed to make an offer of proof, or file a post-trial bill of exception, to preserve his complaint to the trial court’s exclusion of evidence.[3]

[Footnote 3:] Appellant suggests that we may take judicial notice of information posted on a “reliable website.” We decline appellant’s invitation to take judicial notice of the Wikipedia entry for the “John Reid technique.” See James Glerick, Wikipedians Leave Cyberspace, Meet in Egypt, Wall St. J., Aug. 8, 2008, at W1 (“Anyone can edit [a Wikipedia] article, anonymously, hit and run. From the very beginning that has been Wikipedia’s greatest strength and its greatest weakness.”) (emphasis added).

An interesting and reasonable result, which supports the view that Wikipedia shouldn't be relied on for contested questions. For certain uncontroversial matters (such as that the capital of Armenia is sometimes spelled Erevan), citing Wikipedia is probably fine, given that the time of judges, staff attorneys, and law clerks is valuable and best not spent on tracking down The Perfect Source. But when the matter is subject to reasonable dispute, there should either be a hearing -- as with other facts about the details of a case -- or a more elaborate discussion (as with so-called legislative facts that a court uses to determine the meaning of statutory language, develop various common-law rules, and the like).

Thanks to BNA's Internet Law News for the pointer.

from last year, in Gagliardi v. Commissioner (Tax Court=):

Respondent attempted to discredit Dr. Pike by claiming her definition of “gambler’s fallacy” was incorrect. Respondent relies on a definition of “gambler’s fallacy” he obtained from Wikipedia. Respondent did not call any witness, or expert witness, to counter Dr. Pike’s conclusions. Respondent’s reliance on a definition of “gambler’s fallacy” found in Wikipedia[18] is not persuasive. Dr. Pike and Mr. Nicely, a second expert witness whose testimony and opinions are discussed in greater detail infra, credibly explained that there is a difference in the definition of “gambler’s fallacy” depending on the field of study -- e.g., psychology versus mathematics. We find Dr. Pike to be credible and rely on her expert opinion.

[Footnote 18:] Although we conclude that the information respondent obtained from Wikipedia was not wholly reliable and not persuasive in the instant case, we make no findings regarding the reliability, persuasiveness, or use of Wikipedia in general.

So holds an unpublished opinion from the New Jersey Superior Court's Appellate Division:

The common law doctrine of judicial notice is codified in N.J.R.E. 201. Subsection (b)(3) describes the rationale of the rule.

The purpose of judicial notice is to save time and promote judicial economy by precluding the necessity of proving facts that cannot seriously be disputed and are either generally or universally known. Judicial notice cannot be used "to circumvent the rule against hearsay and thereby deprive a party of the right of cross-examination on a contested material issue of fact." Because judicial notice may not be used to deprive a party of cross-examination regarding a contested fact, the doctrine also cannot be used to take notice of the ultimate legal issue in dispute.

... Wikipedia bills itself as the "online encyclopedia that anyone can edit." Anyone with an internet connection can create a Wikipedia account and change any entry in Wikipedia. In fact, Wikipedia warns readers that "[t]he content of any given article may recently have been changed, vandalized or altered by someone whose opinion does not correspond with the state of knowledge in the relevant fields." Thus, it is entirely possible for a party in litigation to alter a Wikipedia article, print the article, and thereafter offer it in court in support of any given position. Such a malleable source of information is inherently unreliable, and clearly not one "whose accuracy cannot be reasonably questioned."

The material that the trial court erroneously judicially noticed was a page "offered ... to establish that Bank One Corporation was purchased by J.P. Morgan & Company in 2004. Against this backdrop, counsel represented to the trial judge that J.P. Morgan sold the accounts, (including defendant's account) to his client Palisades Acquisition." This was relevant to whether the plaintiff actually had a right to sue over the failure to pay on that account.

Thanks to Victor Steinbok for the pointer.

The Nevada Department of Motor Vehicle refused William Junge's request for a personalized license plate that read HOE; the DMV reasoned that "HOE was slang for 'whore.'" (Junge said it was short for Tahoe, and part of his overall Tahoe theme for his car: "Although Junge would have preferred TAHOE for his plate message, he settled on HOE because his first choice was unavailable. For his plate background, Junge initially selected the Lake Tahoe panoramic setting to adorn his 1999 Chevy Tahoe.")

The Nevada Supreme Court, in DMV v. Junge, reversed the denial of the plate, but its reasoning goes beyond the surprisingly substantial but rather frivolous field of License Plate Law:

[B]y its own admission, DMV based its decision solely on the Urban Dictionary. Moreover, DMV revealed a policy of only consulting Urban Dictionary to determine if a word is inappropriate or offensive.

Urban Dictionary is predominantly an online dictionary, although a paper version based on the online content was published in 2005. See http://www.urbandictionary.com/book.php (last visited June 10, 2009). Its definitions are user contributed and are generally anonymous. There is no limit to the number of definitions that a user can contribute.

Since definitions are user contributed, they can be personal to the user and do not always reflect generally accepted definitions for words. See generally http://www.urbandictionary.com/tos.php (last visited June 10, 2009). In fact, Urban Dictionary acknowledges that "[i]ts content is frequently presented in a coarse and direct manner that some may find offensive." See http://www.urbandictionary.com/tos.php (last visited June 10, 2009). Moreover, Urban Dictionary readily admits that it "cannot control all [c]ontent posted by third parties to the [w]ebsite, and does not guarantee the accuracy, integrity or quality of such [c]ontent." Id. Furthermore, Urban Dictionary concedes that it "does not and cannot review all [c]ontent posted to or created by users accessing the [w]ebsite." Id. Thus, Urban Dictionary allows, if not encourages, users to invent new words or attribute new, not generally accepted meanings to existing words.

We acknowledge that the Iowa Supreme Court upheld the use of [Jonathon] Green's Contemporary Dictionary of Slang (1985) to review personalized license plates in McMahon v. Iowa Dept. of Transp., 522 N.W.2d 51, 55-56 (Iowa 1994). Nonetheless, we conclude that this case is distinguishable because Urban Dictionary allows for anonymous, user contributed content. Moreover, without any review of the definitions posted on Urban Dictionary, there is a substantial danger that the definitions will not be generally accepted. Therefore, the DMV's practice risks prohibiting words or phrases based on meanings that are not commonly known or recognized, even as slang terms.

An interesting — and, I think, correct — conclusion, and one that's relevant to other user-generated references such as Wikipedia. As I've noted before, for tangential and uncontroversial matters, Wikipedia may be quite good enough. Government employees' time isn't unlimited, and tracking down authoritative sources to demonstrate the colorfulness of Polish boxer Andrew Golota — to give an example from a Seventh Circuit case that cited Wikipedia to support such an assertion — is probably not the best ways to spend that time. But for something controversial and important, it seems to me that Wikipedia and other reader-generated sources that aren't edited by known and trustworthy authorities should not suffice.

Note that I'm not concerned here about outright lies and manipulations. It seems likely that people who contribute to the Urban Dictionary contribute usages that they themselves have observed. And in fact it's possible that "hoe" (and not just "ho") is seen by some or even many people as slang for "whore," unless some other meaning — say, gardening-related — is visible from context. But all the entry in the Urban Dictionary means (unless the court and I misunderstood the way the Dictionary works) is that one person has claimed that a word has a particular slang meaning, and that the site operators didn't block or remove the submission; it doesn't mean that anyone checked to see whether the definition is in fact common, rare, or even purely idiosyncratic with the submitter and his small circle of friends.