concurring.
I concur in the judgment and join in the majority opinion except for its discussion of the sufficiency of the evidence against Hauser and Cotton on the conspiracy charge and its discussion of imposition of a mandatory life sentence on McGIory. Unlike the majority, I would not place reliance in assessing the sufficiency of the evidence to support Hauser and Cotton’s convictions on allegedly “coded” conversations between Cotton and McGIory and between Hauser and McGIory. Absent the testimony of an expert, who might have decoded *352the cryptic conversations between these putative coconspirators and provided a link between the conversations and the conspiracy, I do not believe that such evidence counts in the calculus of determining guilt beyond a reasonable doubt. Because I believe that there was sufficient additional evidence to convict both defendants on the conspiracy count, however, I concur rather than dissent.
Further, while I agree with, the majority’s affirmance of the imposition of a mandatory life sentence on McGlory under 21 U.S.C. § 841(b)(1)(A), I also explain my reasons for concurring in the majority’s judgment on that issue because it is so important, and because I find it necessary to consider a number of additional factors in interpreting the statute.
I. SUFFICIENCY OF THE EVIDENCE AGAINST HAUSER
In discussing the sufficiency of the evidence against Hauser, the majority opinion makes much of five conversations between Hauser and McGlory which took place in 1988 and 1989 and which involved the use of slang and the recitation of various numbers. The majority concludes that all of these could reasonably be understood as code for drug transactions. These conversations included references to “slippers” and the recitation of various numbers, including “44,” “19-29-75,” and “627, 75, 1000, and 302.”
The majority relies in this regard on United States v. Theodoropoulos, 866 F.2d 587 (3d Cir.1989). In Theodoropoulos, we sustained a conviction based on extensive testimony about coded communications regarding “televisions” between coeonspira-tors which concealed a conspiracy to distribute drugs. Id. at 591-92. Theodoropoulos is easily distinguishable from this case, however, because the jury there heard the testimony of an expert who had decoded the conversations and linked them to a drug conspiracy. Id. at 590-92.
Here, however, the jury heard no such expert testimony. In presenting its case, the government merely emphasized snippets of conversations from its surveillance of McGlory and Hauser, which could possibly constitute code. But no expert ever linked that code to the sale of drugs despite the government’s extensive surveillance of these defendants and the government’s sophistication in cryptanalysis. Compare, for example, United States v. Rollins, 862 F.2d 1282, 1293 (7th Cir.1988) (government expert decoding use of “t-shirts” in conversations between co-conspirators as a reference to drugs). Nor does the majority present any analysis, except for a concluso-ry assertion, which would explain the link between these particular numbers and heroin or between “slippers” and heroin. To my mind, there is no supportable link between any of these words and the distribution of heroin.1
Further, I believe that reliance on such evidence to sustain a conviction represents a dangerous precedent. As I have suggested, these words and numbers are merely pieces of larger conversations. Given that the defendants were charged with conspiracy to distribute heroin, these words and numbers no doubt tended to look very suspicious in the eyes of the jury, as they now do when the government urges them as the basis for affirming the conviction on appeal. We have, however, previously noted the dangers of relying on such evidence to sustain convictions:
There is some tendency in conspiracy cases for finders-of-fact to believe that a defendant must have been involved in the *353conspiracy, once evidence has been presented of some questionable acts the prosecution contends are but an extension of the larger conspiracy. Under our system of law, however, guilt must remain personal and individual, and a conviction, especially on charges relating to a conspiracy, must rest on individual guilt proven beyond a reasonable doubt.
United States v. Samuels, 741 F.2d 570, 574-75 (3d Cir.1984) (citation omitted).
For these reasons, I would not convert Hauser’s suspicious statements into a basis for guilt. Instead, I would uphold the conviction based upon the other evidence detailed by the majority. Most notably, the jury was free to infer that the various references to “Mel” on the “owe sheets” concerned Hauser, and that would implicate him in the ongoing conspiracy to distribute heroin. Further, Hauser sold heroin shortly after visiting McGlory, who was undoubtedly a heroin distributor. Because I am therefore satisfied that the jury was entitled to conclude that Hauser was involved in the conspiracy, I join in the judgment of the court affirming Hauser’s conviction on the conspiracy count.
II. SUFFICIENCY OF THE EVIDENCE AGAINST COTTON
I believe that the majority’s reliance on equally cryptic conversations to sustain the conspiracy conviction of Cotton is also misplaced. The conversations between Cotton and McGlory contain references to a “restaurant” and “pizzas.” The majority concludes that a jury could reasonably infer from those references that Cotton and McGlory were co-conspirators plotting the distribution of heroin. The majority finds this inference permissible despite the lack of evidence suggesting that “pizzas” and “restaurants” referred to heroin either in street slang or in some code devised by McGlory and Cotton.
Under these circumstances, I find reliance on such evidence to sustain the conspiracy conviction troubling. I suppose it is possible that “pizzas” and “restaurants” are references to drugs, but I know of no such usage. Further, in the absence of expert testimony decoding the communications between McGlory and Cotton, I am unwilling to say that it is “reasonable” to infer that they were discussing drugs.2
As with Hauser, I do not dispute that there was sufficient evidence to convict Cotton of conspiracy. The majority demonstrates that McGlory and Cotton’s suspicious meetings could give rise to an inference that they were involved in drug activity. Cotton sold heroin shortly after his meetings with McGlory. Further, Cotton told one of his buyers that his source was being investigated at precisely the time that McGlory was being investigated. Also, a jury might have inferred that Cotton’s name, like Hauser’s name, appeared on the drug “owe sheets.” I therefore conclude that the evidence is sufficient for the jury to conclude that Cotton was involved in the drug conspiracy. My disagreement with the majority is solely limited to its reliance on evidence which cannot be the basis for a reasonable inference of guilt.
III. McGLORYS MANDATORY LIFE SENTENCE
I believe that the majority ultimately reaches the correct result in construing 21 U.S.C. § 841(b)(1)(A) with regard to McGlo-ry’s sentence. However, I believe the question is much closer than the majority opinion allows, and that it therefore warrants more attention, particularly because McGlory will serve a life sentence based on our construction of the statute.
I note at the outset the intuitive appeal of McGlory’s argument. The statute requires that a defendant, to be sentenced to life imprisonment, must have committed *354two prior drug felonies. McGlory’s sentence, however, was increased to mandatory life imprisonment based in substantial part on his conviction for conduct that is no longer a felony under Pennsylvania law. Congress, in enacting the mandatory life sentence of section 841 as part of the Anti-Drug Abuse Act of 1988, no doubt wanted to impose the severe penalty of life imprisonment on individuals who repeatedly violated those state and federal laws proscribing the most dangerous and serious forms of drug involvement, and hence limited the penalty to those who previously had committed drug felonies. It therefore appears anomalous, if not harsh, to sentence McGlo-ry to life imprisonment based on conduct that the Pennsylvania General Assembly no longer considers a serious drug offense.
The statute defines a prior felony drug offense as an offense “that ... is a felony under any law of a state.” See 21 U.S.C. § 841(b)(1)(A) (emphasis added). I believe that this definition is hopelessly ambiguous. As I read it, it has three possible meanings. First, as the majority opinion concludes, it may refer to all offenses that were felonies at the time they were committed. Second, as McGlory argues, the definition arguably could refer to those offenses that were felonies at the time of the 1988 amendment to section 841. The statute defines felonies in the present tense, and under this reading, prior convictions for felonies that had been reclassified as misdemeanors by November 1988 could not be considered prior felony convictions for purposes of the statute.3 On the other hand, a conviction for a misdemeanor that was subsequently reclassified as a felony before November 1988 could count as a prior felony conviction.4 A third plausible reading of the statute would require that those convictions for conduct considered a felony at the time a defendant violated section 841 would be treated as prior felony convictions. Because the definition reads in the present tense, the statute could reasonably be read as defining prior felony convictions in the present: what matters is arguably not how an offense was classified when the prior crime was committed, but rather how that conduct is classified when the defendant subsequently violates section 841.
Because all three of these possibilities appear plausible to me under a “plain” reading of the statute, I cannot agree with the majority to the extent that its decision relies on the text of the statute.5 Therefore, I turn first to the relevant legislative history to attempt to discern congressional intent in this regard. As the majority points out, the history is unhelpful on this specific question. If Congress considered the problem of reclassification of offenses from misdemeanors to felonies and vice versa, it did not address it either in the text of the statute or in the abbreviated legislative history. See 135 Cong Rec S17,360-61 (daily ed November 10,1988) (section analysis of the Anti-Drug Abuse Act of 1988). Because Congress may not even have con*355sidered the possibility of reclassification, I would therefore next determine whether we can interpret the statute based on the overall structure and purposes of the Act. See generally Community for Creative Non-Violence v. Reid, 490 U.S. 730, 109 S.Ct. 2166, 2172-73, 104 L.Ed.2d 811 (1989) (relying on structure of statute to construe its meaning).
This statute appears, however, to have had many purposes, and it is difficult to extract a single legislative “purpose” from it. The statute’s purpose may fairly be read as punishing those individuals who, even though they had prior convictions for felonies, were willing to repeat felonious conduct; under this theory, such individuals are dangerous and undeterred by law, and Congress wanted to imprison them for life as a method of incapacitating them. Viewed from this perspective, the subsequent reclassification of an offense is of little significance. Rather, the critical question is whether the individual is so inveterate in his or her criminality that he or she is willing to commit drug felonies despite having already been sanctioned twice for drug felonies by the criminal justice system. Such conduct reflects an individual’s inability to conform his or her conduct to society’s norms at the time, whatever the subsequent classification of that conduct. Congress may simply have wanted to remove such recidivists from society.
We cannot, however, interpret the statute’s purpose as always favoring maximum incapacitation of prior drug offenders. If that were the case, the statute would not limit the mandatory life imprisonment penalty to those previously convicted of felonies. Individuals convicted of two drug misdemeanors also demonstrate recidivism but are plainly not punished with life imprisonment under the statute. An implicit goal of the act is to provide a rational sentencing policy for drug offenders.6 Under such a policy, the penalty of life imprisonment is limited to those individuals who have repeatedly committed drug offenses considered to be of the most serious nature — in other words, those offenses presently considered felonies. It makes little sense to sentence someone on the basis of conduct no longer considered felonious as if that conduct remained among the most serious of offenses. Therefore, if the “purpose” of the act were exclusively to provide a rational sentencing system for drug offenders, it would make little sense to enhance sentences based on conviction for a crime no longer considered a felony.
From all of this, I conclude that there is no clear policy anchor in the statute that allows us to decide this difficult question. Legislative “purposes” that might be culled from the statute are at odds with each other. Therefore, as with the text and the history of the statute, the legislative “purpose” of the act does not favor one result over the other.
When the construction of a statute becomes as difficult as it is in this case, we frequently turn to other principles to construe the statute. For example, courts, in interpreting ambiguous criminal statutes, will resolve the ambiguity in favor of lenity towards criminal defendants. See United States v. R.L.C., - U.S. -, 112 S.Ct. 1329, 1338, 117 L.Ed.2d 559 (1992); Whalen v. United States, 445 U.S. 684, 100 S.Ct. 1432, 1439, n. 10, 63 L.Ed.2d 715 (1980) (“[Ajmbiguity concerning the ambit of criminal statutes should be resolved in favor of lenity.”) (quoting United States v. Bass, 404 U.S. 336, 92 S.Ct. 515, 522, 30 L.Ed.2d 488). That principle is unhelpful here, however. The interpretation that McGlory urges, while more lenient towards him, is not necessarily more lenient to the entire class of affected criminal defendants, which is the relevant consideration. See McNally v. United States, 483 U.S. 350, 107 S.Ct. 2875, 2881, 97 L.Ed.2d 292 (1987). In the situation obverse to that of McGlory’s, a criminal defendant convicted of a misdemeanor that has subsequently been reclassified as a felony would be sub*356ject to having that misdemeanor considered a felony under the interpretation that McGlory urges.7 That same defendant, however, would not be subject to life imprisonment under the interpretation proffered by the majority opinion. Therefore, none of the interpretations is inherently more lenient.
Another important consideration, relied on in part by the majority, is the facility with which courts can administer each of the three possible rules. The rule adopted by the majority opinion has the advantage of easy application. If a defendant has been convicted of a drug-related felony, the sentencing court simply treats it as a felony with no further inquiry. The other two possible constructions require an examination into whether the conviction would still be a felony either in November 1988 (the time of the enactment of the Act) or, alternatively, at the time when the defendant violated section 841.
In certain cases, such as this one, determining whether the offense for which defendant was convicted would be a felony in 1988 or at the time of violation is easy. Where an offense has simply been reclassified from a felony to a misdemeanor, a court only has to note the reclassification. The change in Pennsylvania law involved in this case appears to have been nothing more than a lessening of the criminal penalty and a titular change of the offense from felony to misdemeanor. Often, however, reclassifications involve subtle changes in the nature of the crime that make it difficult to ascertain whether the conduct for which defendant was convicted would in 1988 or at the time of violation be a felony.
For example, changes in the amount of narcotics required to make possession a felony might create confusion. A defendant convicted of a misdemeanor in 1980 for possessing a certain amount of narcotics might potentially have been guilty of a felony in 1988 for possessing the same amount based on a decrease in the amount required to make a possession offense a felony. To discern whether or not he or she was guilty of a “1988 felony,” the court would be required to determine the actual amount involved in the prior conviction, a process that would require a mini-trial on the amount involved in the prior conviction. The events may have occurred in a far-distant state and the critical witnesses may be dead or unavailable. Although, in some cases, a court might only have to refer to the pre-sentence report, in others, where the amount of narcotics involved remains uncertain, it would be impossible for a court to determine whether the conviction would be a felony offense in 1988 or at the time of violation.8 At all events, it is always simpler to rely on what the defendant was actually convicted of, a fact that strongly militates in favor of the rule the majority adopts for purposes of sentencing under section 841.
Thus, I join in the judgment with respect to this sentencing issue on the basis of this administrability concern alone.
. Indeed, our holding in Theodoropoulos demonstrates the difficulty with the majority’s approach in this case. In Theodoropoulos, we upheld the admission of expert testimony to decode communications between defendants "because the code words [were] interwoven throughout the conversation, often in confusing contexts....” Id. at 592. Accord United States v. Rollins, 862 F.2d 1282, 1292 (7th Cir.1988) ("In our view, the district court properly con-eluded that the meaning of narcotics code words and phrases is not within the common understanding of most jurors.”). By finding expert testimony "helpful” in decoding narcotics code words under FRE 702, we conceded that lay jurors often have difficulty discerning the meaning of drug slang and code. If that is true, I find it problematic to base a conviction, even in part, on code and slang as cryptic as the kind involved in this case.
. In contrast, as the majority points out, there was testimony, given by a witness who was knowledgeable about the code used by these individuals, which identified "eating" as code for drug use and sales. I do not object to the majority’s reliance on that evidence because, from all of the evidence, the jury could reasonably have relied on that term in its determination of guilt.
. Such reclassification is not unusual or unlikely with respect to drug offenses. The offense at issue in McGlory’s case, mere possession of cocaine, was subsequently determined to be a non-serious offense and hence downgraded to a misdemeanor. Conversely, possession of certain controlled substances (for example, crack cocaine, see note 7), has sometimes been upgraded from a misdemeanor to a felony. Moreover, there has been a trend towards classifying offenses based upon the amount of drugs involved so that, where the amount required to classify an offense as a felony has been lessened, an offense might be a felony where it had previously only been a misdemeanor.
. If we are to interpret the statute sensibly, we must consider upward classifications of offenses from misdemeanors to felonies as well as downward classifications of offenses from felonies to misdemeanors, see note 3. I know of no principled basis for treating the two situations differently.
. Because I believe the statute is ambiguous, I think that the majority is incorrect in asserting that McGlory’s proposed construction of Section 841 would require the addition of words to the statute. See Majority Opinion at 351. McGlo-ry’s position requires that the statute be read as it is written — in the present tense. The majority position, in contrast, arguably requires the addition of words to the statute. Because the majority allows imposition of sentence on the basis of convictions that either are or were felonies under state law, its interpretation requires the statute to be rewritten to say a prior felony conviction is an offense “that is [or was] ... a felony under any law of a state."
. This fact is made clear by the scale of penalties established in 21 U.S.C. § 841. Individuals convicted of one prior offense who violate the relevant portions of section 841 are sentenced to a mandatory minimum sentence of twenty years. The penalty for those with two prior convictions is escalated to life imprisonment.
. A classic example of reclassification that would hurt criminal defendants under the rule McGlory urges can be found in the federal laws governing the possession of crack cocaine. 21 U.S.C. § 844(a), as amended in November 1988 (the same time that section 841 was amended to add a mandatory life sentence), treats possession of crack cocaine (a mixture or substance that contains cocaine base) as a felony. See Anti-Drug Abuse Act of 1988, Pub.L. No. 100-690, 102 Stat. 4370 (November 18, 1988). Individuals convicted under the old section 844(a) for possession of crack cocaine were liable only for a fine of up to $1,000 and for imprisonment only up to one year, making the crime a misdemeanor. After the amendment, defendants now face at least five years’ imprisonment, which categorizes the crime as a felony. Under McGlory’s interpretation, a person convicted of misdemeanor possession under the old section 844 would have that conviction converted to a felony for purposes of enhancing the sentence upon a new violation of section 841. Obviously, the rule in that instance is harsher towards criminal defendants.
. I note in this regard that I do not see how we can carve out a case-by-case analysis for determining what was a felony in 1988 in order to except the easy cases such as this one. Nothing in the statute indicates that Congress intended such a complex inquiry.