concurring:
I am pleased to join Judge Reid’s opinion for the court. I add a few observations from a slightly different perspective.
*117To prove that Mitchell was guilty, the prosecutor had to establish that all of the following propositions were true:
1. that Mitchell had heroin on his person, but that the police did not search him thoroughly enough to discover it;
2. that the police searched the car before placing Mitchell in it, and that this search was sufficiently thorough to ensure that the heroin must have been placed in the vehicle after the search;
3. that Mitchell was able to dispose successfully of the heroin with his hands cuffed behind him;
4. that Mitchell “squirmed,” apparently while disposing of the heroin, but that the police did not investigate or take any action; and
5. that although Mitchell’s activities while handcuffed apparently caused some of the white powder to spill out of the zip-lock baggie on to the floor of the police car, no powder was recovered from Mitchell’s slacks or his underwear.
Each of these propositions, considered alone, may be plausible. I do not believe, however, that an impartial trier of fact could fairly conclude, beyond a reasonable doubt, that all five of them were established.
That is not all. The government could have produced photographs of the interior of the police car and of the spilled powder, so that the trier of fact could more readily assess the plausibility of the prosecution’s version of events. It did not do so. The production of weak evidence, where strong evidence would ordinarily be available, warrants an inference against the proponent of such evidence; indeed, in some instances, it will compel such an inference. See, e.g., Murphy v. McCloud, 650 A.2d 202, 215-17 (D.C.1994); cf. Interstate Circuit, Inc. v. United States, 306 U.S. 208, 226, 59 S.Ct. 467, 474, 83 L.Ed. 610 (1939). Moreover, the fact that the police car was left unguarded for some period of time presents a kind of “chain of custody” problem which further weakens the prosecution’s case.1
Finally, one might reasonably have doubts about another aspect of the prosecution’s case. Officer Silva stopped Mitchell in southeast Washington because the ear he was driving had Virginia tags but no Virginia inspection sticker. He testified that he thereafter arrested Mitchell because, when he ran a check on Mitchell’s social security number and driver’s license, he was erroneously advised that there was an outstanding warrant for Mitchell’s arrest. According to the government’s evidence at trial, the warrant was actually for someone else named Mitchell who had a similar date of birth and a similar scar on his left knee. All of this is very mysterious. “Coincidences happen, but an alternative explanation not predicated on happenstance is often the one that has the ring of truth.” Poulnot v. District of Columbia, 608 A.2d 134, 139 (D.C.1992).
This is a close ease. The trial judge was on the scene. He heard the testimony and observed the demeanor .of the witnesses. Mitchell may well have been guilty.2 Never*118theless, on the basis of the unusual assortment of problems and difficulties which .characterized the government’s case, I agree with Judge Reid that, as a matter of law, there was a reasonable doubt of Mitchell’s guilt.
. Personally, however, I do not attach a great deal of significance to this aspect of the case. The possibility that some civilian third party placed heroin in a police car in order to incriminate Mitchell appears remote. It is theoretically possible that the police planted the contraband, but there is no evidence that this occurred.
. I do not want the reader to infer that I believe that Mitchell was framed. I note, albeit from a concededly unorthodox perspective, that there is information in this record — specifically, two pri- or heroin-related convictions — which logically makes it a great deal more likely that Mitchell committed this offense, which also involved heroin. "[A] defendant's past conduct is important evidence — perhaps the most important — in predicting his probable future conduct.” Cruz-Foster v. Foster, 597 A.2d 927, 930 (D.C.1991) (quoting State v. Krol, 68 N.J. 236, 344 A.2d 289, 302 n. 12 (1975)); cf. United States v. Crowder, 318 U.S.App.D.C. 396, 407, 87 F.3d 1405, 1416 (1996) (en banc) (concurring opinion of Silber-man, J., joined by Buckley and Williams, JJ., at 1) ("evidence that a defendant charged with a drug distribution crime has previously committed drug distribution crimes should be admissible to show likelihood (propensity, if you will) that the defendant did it again"). I recognize, of course, that the information about Mitchell’s convictions, albeit obviously probative, was not presented to the court until the prosecutor's recross examination, that it was therefore excluded as untimely, even for impeachment, and that, under current law, it was plainly inadmissible to show that Mitchell "did it” this time. See, e.g., Thompson v. United States, 546 A.2d 414, 418-20 (D.C.1988).