beech trees,

There is a lot that's troubling about the D.C. police's reliance on polygraphy in the Chandra Levy murder investigation. The latest Washington Post article that you've cited (which is also linked to and excerpted from on the AntiPolygraph.org Polygraph News page) is disturbing in a number of ways. Here are some preliminary ruminations:

Based on the Post article, it appears that no later than September 2001, an inmate in the D.C. jail "told authorities that [Ingmar A.] Guandique had confided in him that he stabbed Levy and left her body in [Rock Creek Park]." This was months before Levy's skeletal remains were discovered in that very park on 22 May 2002, a point that the authors of the Post article fail to note.

But this jailhouse informant, who, according to the article, "didn't try to trade the information for a lighter sentence, saying he came forward because he felt bad for the Levy family," failed a polygraph "test" about his allegations in September 2001.

Guandique, however, passed his "test" (a predictable result if the informant had been polygraphed first and "failed"), so the D.C. police discounted the informant's claim.

Now, some four months after the discovery of Levy's remains, the D.C. police need to discredit the polygraph results. They must have known at the time that an interpreter was used in Guandique's polygraph examination (as well as the informant's). But it is only now, when the polygraph results apparently no longer fit the leading investigative theory, that the use of an interpreter has become a problem. So now the D.C. police offer the public an ad hoc hypothesis about the use of an interpreter having confounded the polygraph results, anonymously whispering in the ear of the Washington Post, which dutifully (and uncritically) parrots this new official line.