concurring:
I concur in the judgment of the panel because I believe it is the correct application of this circuit’s precedent. See, e.g., United States v. Machado, 804 F.2d 1537, 1543 (11th Cir.1986) (“Only a decision by this court sitting en bane or by the United States Supreme Court can overrule a prior panel decision.”). I write separately to express my view that the circuit’s doctrine with respect to admission of Rule 404(b) evidence in conspiracy cases has evolved into one that undermines Rule 404(b) itself and represents a perversion of the origins of the circuit’s doctrine in this context.
The evidence of Matthews’s 1991 arrest was admitted for no purpose other than to show propensity to engage in criminal activity: exactly the purpose for which Rule 404(b) prevents evidence to be admitted. Yet, this circuit’s. doctrine permits such evidence to be admitted as probative of intent. The doctrine presumes that the intent involved in a small-scale drug transaction (not to mention personal drug use, see United States v. Butler, 102 F.3d 1191, 1195-96 (11th Cir.1997)) is somehow probative of one’s intention to conspire with others to commit a drug offense. The result is the admission of prior acts, in the name of intent, that would otherwise be inadmissible propensity evidence.1 Accordingly, if I were writing on a clean slate, I would hold that the evidence of Matthews’s 1991 arrest is irrelevant to the intent at issue and therefore could not be *1314admitted as 404(b) evidence probative of intent.
I.
As noted in the court’s opinion, the starting point for all Rule 404(b) analysis in this circuit is our opinion in United, States v. Beechum, 582 F.2d 898 (5th Cir. 1978) (en banc). There we noted:
What [Rule 404(b)] calls for is essentially a two-step test. First it must be determined that the extrinsic offense evidence is relevant to an issue other than the defendant’s character. Second, the evidence must possess probative value that is not substantially outweighed by its undue prejudice and must meet the other requirements of rule 403.
Id. at 911.2 In elaborating on the test, we explained that “[o]nee it is determined that the extrinsic offense requires the same intent as the charged offense and that the jury could find that the defendant committed the extrinsic offense, the evidence satisfies the first step under rule 404(b).” Id. at 913 (emphasis added). With respect to the second step, we noted that “[i]f the defendant’s intent is not contested, then the incremental probative value of the extrinsic evidence is inconsequential when compared to its prejudice; therefore, in this circumstance the evidence is uniformly excluded.” Id. at 914. We stated, however, that “[wjhether a mere plea of not guilty justifies the Government in introducing extrinsic evidence in its case in chief is an open question in this circuit .... We need not now answer it.” Id. at 915.
In United States v. Roberts, 619 F.2d 379 (5th Cir.1980), the court answered that question in the affirmative in the context of conspiracy cases. Focusing heavily on the uniquely difficult task of proving intent in conspiracy cases, the court concluded that “we cannot apply to them the policy suggested in Beechum of uniformly excluding extrinsic offense evidence when the defendant does not actively contest intent.” Id. at 383. “In every conspiracy case, therefore, a not guilty plea renders the defendant’s intent a material issue .... Evidence of such extrinsic evidence as may be probative of a defendant’s state of mind is admissible unless [the defendant] ‘affirmatively take[s] the issue of intent out of the case.’ ” Id. (last alteration in original) (quoting United States v. Williams, 577 F.2d 188 (2d. Cir.1978)).3
*1315It is this latter statement that is now reflexively invoked, frequently without reference to context, or any other analysis for that matter, to admit any and all prior acts (involving drugs) in drug conspiracy cases. See, e.g., Butler, 102 F.3d at 1195-96 (“Intent is always at issue when a defendant pleads not guilty to a conspiracy charge .... [W]e conclude that the logical extension of our current jurisprudence is to admit evidence of prior personal drug use to prove intent in a subsequent prosecution for distribution of narcotics.”); United States v. Costa, 947 F.2d 919, 925 (11th Cir.1991); United States v. Pollock, 926 F.2d 1044, 1049 (11th Cir.1991) (concluding, after citing the Roberts proposition, that “the precedents in this circuit allowing evidence of prior drug-dealing offenses in drug conspiracy prosecutions are ample”). Yet Roberts’s progeny has ultimately lost sight of two important components of the Roberts opinion. First, the Roberts contribution to the Beechum test referred only to the second prong of the analysis: once extrinsic evidence has been deemed relevant to intent, Roberts presumes that, in conspiracy cases, the probative value of the prior act is not substantially outweighed by its prejudicial impact even if the defendant did not actively contest the issue of intent. Roberts did nothing to alter the analysis of whether a prior act is relevant to the issue of intent. See Roberts, 619 F.2d at 382 (noting that “[i]f ‘the extrinsic offense requires the same intent as the charged offense ... ’ then ‘the extrinsic offense evidence is relevant to an issue other than the defendant’s character,’ and satisfies the first requisite of Beec-hum,” before discussing the relevance of a not guilty plea to the second prong of the analysis (citations omitted) (quoting Beec-hum, 582 F.2d at 913 & 911)); Beechum, 582 F.2d at 914 (“The touchstone of the trial judge’s analysis in this context should be whether the Government has proved the extrinsic offense sufficiently to allow the jury to determine that the defendant possessed the same state of mind at the time he committed the extrinsic offense as he allegedly possessed when he committed the charged offense.”) (emphasis added); United States v. Williford, 764 F.2d 1493, 1497 (11th Cir.1985) (“The first requirement [of a Rule 404(b) analysis] is governed by the general relevance provision of Fed.R.Evid. 401 .... [I]f the relevant issue is intent, the acts must require similar states of mind.”). Thus, it must first be established that the prior act is probative of intent — that the prior act required the same state of mind as the charged offense — before Roberts comes into play. Roberts simply holds that, in a conspiracy case, absent an affirmative step by the defendant to remove intent from the case, intent is an issue that may be proven through extrinsic acts relevant to that intent.
This observation ties into the second overlooked aspect of the Roberts opinion: Roberts deals specifically with the intent to conspire to commit an illegal activity. The more capacious standard for admitting 404(b) evidence in conspiracy cases, announced in Roberts, is fundamentally rooted in the special nature of conspiracy cases. “Charges of conspiracy involve considerations not present in other criminal prosecutions.” Roberts, 619 F.2d at 382. As noted in the court’s opinion, ante, the Roberts court reasoned:
Because the prosecution must prove that the defendant knowingly joined a plan to commit a crime, evidence that establishes a defendant’s participation in a criminal act, or evidence establishing his association with co-conspirators, may be *1316insufficient to support the inference that the defendant voluntarily joined a conspiracy to commit a crime ....
Unequivocal evidence that a defendant committed a substantive offense may justify the inference that he intended to do so, but it does not plainly support the conclusion that he agreed and planned with others to commit the crime. Evidence of a defendant’s association and dealings with a group of conspirators, even when he knows they intend to commit a crime, does not alone show that he. himself had the requisite intent to join the conspiracy.
Roberts, 619 F.2d at 383 (citations omitted). In other words, the mental states involved in the substantive drug offenses have very little to do with a defendant’s mental state regarding an agreement to conspire. And it is this later intent, about which it is difficult to draw inferences from underlying substantive offenses, that the Roberts framework addresses.
Given the relative difficulty of establishing conspiratorial intent, the Roberts court created the presumption that such intent is at issue (regardless of whether the defendant actively contests the issue) and evidence of prior acts probative of that intent may therefore be admitted, unless the defendant affirmatively removes the issue of intent from the case.4 But if this relaxed standard for admissibility of Rule 404(b) evidence in conspiracy cases is premised on the incongruity between the intent to conspire and the intent to commit the underlying substantive offense, then the probative prior acts must be something more than another substantive offense. If, as Roberts notes, intent to conspire cannot be inferred from proof of the underlying drug offense, it strains credulity to presume that a prior, unrelated drug offense might be at all probative of one’s current intent to conspire.
This is not to say that evidence of prior substantive offenses is not admissible in conspiracy cases. Before extrinsic acts are admitted, however, it must first be determined what fact(s) they are introduced to prove and whether they are relevant to such fact(s).5 While prior substantive offenses say very little about one’s voluntary participation in a conspiracy, they may be relevant to demonstrate that the defendant had the requisite intent to engage in the criminal object of the conspiracy. See, e.g., United States v. Calderon, 127 F.3d 1314, 1330 (11th Cir.1997) (defense that defendant “was merely present at the scene of the drug activity” puts intent at issue); United States v. Lampley, 68 F.3d 1296, 1299 (11th Cir.1995) (“The district court, recognizing that [defendant] was presenting a ‘mere presence’ defense, stated that the evidence ... ‘shows intent.’ ”); United States v. Diaz-Lizaraza, 981 F.2d 1216, 1225 (11th Cir.1993) (“[Defendant’s] ‘non-participation’ defense focused squarely on intent. The question in this case ... was why [defendant] was at the shopping center during the transaction.”); United States v. Dorsey, 819 F.2d 1055, 1060 (11th Cir.1987) (defendant claimed he “was merely an ‘unwitting’ observer ... and that he had no intent to become involved in any criminal activity”); United States v. Richardson, 764 F.2d *13171514, 1523 (11th Cir.1985) (“[T]he government had reason at the start of the trial to believe that [defendant] would argue lack of intent because of his reliance on the entrapment defense.”). But the intent in these situations is not the difficult-to-prove intent to conspire that prompted the Roberts presumption; it is the intent to engage in a substantive offense. Whether extrinsic evidence can be admitted as proof of intent to pursue the objective of a conspiracy, or an associated substantive offense, is a question analyzed through the Beechum framework, unaffected by the Roberts presumption. Intent to commit a substantive offense does not become more difficult to prove simply because the crime is associated with a conspiracy to commit it. Accordingly, the prosecution cannot benefit from the relaxed standard of admissibility to provide proof of the defendant’s intent to commit a substantive offense.
Rather, in order for the Roberts presumption to come into play, it must first be demonstrated that the extrinsic act is probative of the intent to conspire. In Roberts itself, the court noted: “Proof that Mr. Roberts had intentionally joined in a conspiracy to operate a gambling business four years prior to his present participation in such an operation increases the likelihood that he had conspired with others to establish and operate the gambling business.” 619 F.2d at 383-84. Thus, the Roberts court admitted prior acts demonstrative of an intent to agree with others to carry out an illegal course of action, even without an affirmative challenge to that intent. See also United States v. Delgado, 56 F.3d 1357, 1365 (11th Cir.1995) (prior act: conspiracy and attempt to possess with intent to distribute); Pollock, 926 F.2d at 1047 (prior act: conspiracy to import marijuana); United States v. Na-hoom, 791 F.2d 841, 843 (11th Cir.1986) (prior act: participation in unrelated joint money laundering scheme); United States v. Nabors, 707 F.2d 1294, 1300 (11th Cir. 1983) (“[F]our years before the drug smuggling operation charged, [defendant] had been engaged with [co-conspirator] in a series of virtually identical operations.”); United States v. Holman, 680 F.2d 1340, 1348-50 (11th Cir.1982) (prior acts: knowledge of and participation with alleged co-conspirators in “nearly identical” smuggling attempts).
The upshot of this framework is that in conspiracy cases extrinsic evidence can be admitted for either of two intent-related purposes. First it can be admitted to show the defendant intended to engage in the underlying offense (e.g., he intended to distribute or possess an illegal drug). On this, Roberts has no bearing. Second, pri- or acts may also be introduced as evidence of a defendant’s intent to conspire to commit an illegal activity. For this purpose, Roberts provides the presumption that intent is a material issue unless the defendant affirmatively removes it from dispute. Yet in these contexts, the extrinsic acts must at least be more relevant to the intent to conspire than the substantive acts of the charged conspiracy, the irrelevancy of which led to the Roberts presumption in the first place.
Our precedent, however, has evolved into a doctrine foreign to the analytical underpinnings of the Roberts opinion, and antithetical to the purposes of Rule 404(b). We have approved of the admission of prior substantive offenses for the purpose of establishing an intention to accomplish the object of the conspiracy simply because the defendant pled not guilty. See, e.g., Costa, 947 F.2d at 925. In United States v. Maxwell, we admitted extrinsic evidence “as relevant to [defendant’s] intent to distribute controlled substances,” where defendant “placed [intent] at issue by pleading not guilty to the charge against him.” 34 F.3d 1006, 1009 (11th *1318Cir.1994). And in Butler, we admitted evidence of a prior conviction for possession of cocaine, “to establish the intent to conspire to distribute narcotics,” after reciting the familiar refrain that “[i]ntent is always at issue when a defendant pleads not guilty to a conspiracy charge.” 102 F.3d at 1195. Compare United States v. Zelinka, 862 F.2d 92, 99 (6th Cir.1988) (“The district court abused its discretion by admitting the evidence in question without requiring the prosecutor to show its relevance and the necessity for its admission. Even if the evidence indicated that [defendant] possessed cocaine for the purpose of distribution ..., it was not admissible to prove that he had participated in a conspiracy to distribute cocaine ....” (citation omitted)). In other words, due to the Roberts doctrine, unless a defendant affirmatively removes intent from issue, all prior drug offenses, whether probative of the intent to conspire or not, are admissible so long as a conspiracy has been charged and Rule 403 balancing is satisfied.
As a result, the Roberts presumption has morphed into a categorical-relevancy doctrine that presumes that virtually all prior drug offenses are relevant and almost automatically admissible in all drug conspiracy cases (subject only to Rule 403).6 This is true regardless of the theory of defense or whether the prior offense involved the same state of mind that is at issue (either by default or affirmatively) in the charged conspiracy. This fuzzy math is not only problematic from a theoretical perspective,7 but it serves to admit propensity evidence in the name of intent.8 The intent necessary to possess an illegal drug is no more relevant to the intent to either conspire to distribute illegal drugs or to distribute them than any other criminal act. Cf. United States v. Baldarrama, 566 F.2d 560, 568 (5th Cir.1978) (“However, there is no real danger of impropriety here, since the prior conviction did not involve a conspiracy, and therefore could not properly have been used by the jury to indicate an intent to conspire.”). Its only relevance is sheer propensity: the theory being that the defendant acted illegally then, and is likely to be acting illegally now. This is precisely the inference the law does not allow. Cf. People v. Zackow-itz, 254 N.Y. 192, 172 N.E. 466, 468 (1930) (Cardozo, C.J.) (“If a murderous propensity may be proved against a defendant as one of the tokens of his guilt, a rule of *1319criminal evidence, long believed to be of fundamental importance for the protection of the innocent, must be first declared away.”).
Thus, this circuit’s doctrine has turned Rule 404(b) on its head, in so far as conspiracy cases are concerned. The second sentence of Rule 404(b)9 is an exception to the general rule that “[ejvidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show action in conformity therewith.” Fed.R.Evid. 404(b). There is nothing in the Rule or in logic that would treat conspiracies differently in this regard. Yet we have bypassed the strictures of Rule 404(b) by presumptively assuming that intent is always an issue in conspiracy cases and that all prior substantively-related acts are relevant to that intent. This is nothing more than propensity by any other name.
II.
The case at bar squarely demonstrates the problems created by our precedents. The Government rightly states, that “Matthews’s not-guilty plea placed directly at issue his intent to join the charged conspiracy to distribute cocaine.” It then seeks admission of Matthews’s 1991 arrest for trafficking cocaine as evidence of his intent to engage in the alleged conspiracy. Neither Beechum, Roberts, nor a straightforward relevance analysis would permit the evidence to be admitted.
There is no question in this case that if the jury were to believe the alleged co-conspirators’ testimony, it must believe that Matthews intended to possess and deliver cocaine. Moreover, Matthews did not put his intent to distribute at issue merely by pleading not guilty. He did not contend that he engaged in the alleged activities but did not know he was distributing cocaine. He did not contend that he was merely present at the alleged transactions. Rather, Matthews claimed that the witnesses were not telling the truth at all. See Brief of Appellant at 19. The open issue, per Roberts, is whether Matthews had the requisite state of mind to be convicted of a conspiracy to possess and deliver cocaine, viz. whether Matthews voluntarily joined an agreement to engage in such activity. On this issue, the 1991 incident sheds no light. To the extent the prior arrest was at all relevant it was to show that Matthews was dealing drugs 1991 and was still dealing drugs in 1999, 2000, and 200Í, which somehow makes it more likely that he was also conspiring to deal drugs during that time.10 In other words, Matthews was bad in 1991 and he’s bad now. Propensity does not become intent simply because the Government says as much.
It is clear to me that this evidence should not have been admitted in this case. Yet our precedent makes it impossible for me to so hold. As indicated in Part I, *1320cases such as Costa and Butler have expanded the Roberts doctrine beyond anything that resembles the original framework.11 In this circuit, in any conspiracy case where the defendant has pleaded not guilty and has not stipulated to intent, any prior act that has anything to do with the object of the conspiracy is deemed probative of the intent to engage in the conspiracy. In this case, Matthews was charged with conspiracy to distribute five or more kilograms of cocaine. He pled not guilty and did not stipulate to intent. He was previously arrested for a crime involving drugs. In this circuit, the prior incident is unequivocally admissible subject only to Rule 403 balancing. I strongly encourage the court to revisit the doctrine leading to this result.
. I concede that the line between evidence admitted to demonstrate intent and evidence admitted to demonstrate propensity is hardly clear. It is difficult to argue that a person had an intention to do something on a particular occasion because he or she demonstrated that intention previously without implicitly suggesting that the person has a proclivity towards that intent. Indeed, some contend that a distinction is impossible: "[E]vidence of an unconnected prior crime is always evidence of propensity and never evidence of a specific intent to commit the crime charged. Though an inference of general intent from the prior crime to the offense charged can be made, such an inference is based upon propensity .... ” Abraham P. Ordover, Balancing the Presumptions of Guilt and Innocence: Rules 404(b), 608(b) and 609(a), 38 Emory L.J. 135, 157 (1989). Regardless, the rules distinguish between the two and so must we. In doing so, however, we need to be mindful of the potential for the exception to swallow the rule. See Thompson v. United States, 546 A.2d 414, 420-21 (D.C.1988) ("[T]he intent exception has the capacity to emasculate the other crimes rule. This is so because, in many cases, it is difficult or impossible to differentiate between the intent to do an act and the predisposition to do it. For obvious reasons, therefore, courts must be vigilant to ensure that poisonous predisposition evidence is not brought before the jury in more attractive wrapping and under a more enticing sobriquet .... If the prime purpose or effect of the introduction of other crimes evidence is to show that the defendant is a bad man (and, in this case, a drug seller), so infelicitous a project cannot be salvaged by characterizing the venture as being designed to prove his intent to sell drugs rather than his predisposition to do solicitation omitted)). At the very least, where the evidence sought to be admitted demonstrates nothing more than a criminal intent (or even, more specifically, a prior intention to violate drug laws) it must be excluded as inadmissible propensity evidence. If the inferential chain must run through the defendant’s character — and his or her predisposition towards a criminal intent — the evidence is squarely on the propensity side of the elusive line. Where, on the other hand, an inference can be drawn that says nothing about the defendant's character — for example, based on the "improbability of coincidence," see Clifford S. Fishman, Jones on Evidence: Civil and Criminal § 17:48 (7th ed.1992) — the evidence is more appropriately admissible for non-propensity purposes.
. Rule 403 provides:
Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.
Fed.R.Evid. 403.
. While it may be too late in the day to question the wisdom of the Roberts presumption, I do find the resulting Hobson's choice particularly troublesome: The defendant must either stipulate to the intent to conspire (which, in a conspiracy charge, would virtually guarantee a conviction) or have the court presume that the defendant has testified and specifically challenged his or her intent to conspire, which the prosecution is then free to rebut with extrinsic evidence. Moreover, at least one commentator has pointed out the logical fallacy of creating "material issues” due to the difficulty of proving a particular crime:
Courts have accepted the premise that intent is difficult to prove in conspiracy cases. After all, some of the actions of the conspirators standing on their own may be lawful. Since it is difficult to prove intent, and since the jury might infer from a defendant's lawful actions that he did not intend any illegal activity, evidence of unconnected, though similar, crimes will be admitted. [In other words]: when the government has a weak case, it is “necessary” to "materialize” a nonmaterial issue and prove it with propensity evidence. Nothing in the conspiracy cases logically makes intent "a fact of consequence” other than the "need” to get a conviction .... Ultimately, the question of whether intent is a material issue should be in the hands of the defendant:
*1315the nature of the defense should determine whether intent is material.
Ordover, supra note 1, 154-55 (footnotes omitted).
. Even an explicit stipulation as to intent may not be sufficient to bar admission of extrinsic evidence. See United States v. Williford, 764 F.2d 1493, 1498 (11th Cir.1985) ("This circuit has refused to adopt a per se rule either for or against admission of evidence when that evidence is relevant to an issue to which the defendant offers to stipulate.”).
. The elements of conspiracy include: (1) the existence of an agreement among two or more persons; (2) knowledge of the general purpose of the agreement; and (3) knowing and voluntary participation in the agreement. United States v. Simpson, 228 F.3d 1294, 1298 (11th Cir.2000).
. We have sensibly limited admission of prior acts to those that at least arguably deal with the same underlying substantive offense of the conspiracy (e.g., prior drug offenses in drug conspiracy cases). See United States v. Young, 39 F.3d 1561, 1573 (11th Cir.1994) ("Alcohol is not a controlled substance, and the illegality of its production is distinct in both fact and law from that involved in growing and selling marijuana. Evidence that the [defendants] made alcohol thus was not probative of their intent to engage in a conspiracy to possess and distribute marijuana, and any inference that could be drawn from the introduction of this evidence was precisely that which Rule 404(b) was designed to prohibit.”).
. Relevance is inherently a case-specific determination, see Fed.R.Evid. 401 advisory committee’s note ("Relevancy is not an inherent characteristic of any item of evidence but exists only as a relation between an item of evidence and a matter properly provable in the case.”), and this court should not be so bold as to announce a one-size-fits-all rule in a context where one size cannot fit all.
.As Judge Torreulla has astutely noted, "[a]t times ... a court circumvents the ban [on the use of propensity evidence] by allowing other crimes into evidence whenever intent or knowledge are arguably at issue, without making explicit the specific logical progression that makes either intent or knowledge more likely in light of the prior crime.” United States v. Rubio-Estrada, 857 F.2d 845, 853 (1st Cir.1988) (Torruella, J., dissenting). This observation rings particularly true with respect to this circuit’s conspiracy case law.
. ''[Evidence of other crimes, wrongs or acts] may, however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident ....” Fed. R.Evid. 404(b).
. As noted above, in its closing argument, the Government argued that
[t]he intent in 1991 that the defendant had was to distribute cocaine. He had those little baggies of cocaine in the trunk of his car and he was out there distributing it back in 1991. And by the time of this charged conspiracy in 1999 through June of 2001, the defendant had that same intent. He had the same intent to distribute cocaine, only now he was a bigger dealer. He’s deeding in kilograms of cocaine/ not little baggies of cocaine any longer.
Vol. 14, at 22. Matthews, however, was not being charged with distribution of cocaine, nor was his intent to further the conspiracy by distributing cocaine put at issue by his not guilty plea.
. Under the prior panel rule, this panel would be obliged to disregard any panel decisions that are expressly or implicitly inconsistent with earlier circuit precedent such as Beechum or Roberts. See Swann v. Southern Health Partners, Inc., 388 F.3d 834, 837 (11th Cir.2004) ("Under the prior panel rule, we are bound by the holdings of earlier panels unless and until they are clearly overruled en banc or by the Supreme Court."). I do not believe, however, that the post-Roberts case law can be interpreted as overturning Roberts, or is irreconcilable with prior 404(b) doctrine. Rather, I simply believe that Roberts’s progeny has, over time, come untethered from the reasoning of the Roberts court and has resulted in an unprincipled and unwieldy doctrine that ought to be tamed. Accordingly, I view this panel as bound by Roberts's progeny and implore the circuit to rethink the direction the doctrine has taken.