(dissenting).
The problem dividing the panel — the admissibility of uncharged offenses under Fed.R.Evid. 404(b) is a recurring one.1 See United States ¶. Henderson Simon, 842 F.2d 552, 556 (1st Cir.1988) (Torruella, J., concurring). More important than its recurrence, however, is the importance of this issue in giving criminal defendants a fair trial for the offenses charged because, as the Tenth Circuit has said, “an obvious truth is that once prior convictions are introduced the trial is, for all practical purposes, completed and the guilty outcome follows as a mere formality.” United States v. Burkhart, 458 F.2d 201, 204 (10th Cir.1972). Such evidence has such a “blinding” effect on a lay jury as to require more hesitancy in validating its use than is displayed by the majority today.
I
The initial problem with the rule’s prohibition against admitting uncharged bad acts to prove the bad character of defendant is that, as the majority points out, it runs contrary to common sense. It is certainly common to everyone’s experience to believe that one who has acted wrongly in the past may act in the same way again. It is precisely because of this “commonsensical” deduction, which may lead (or perhaps more accurately, mislead) the jury away from its duty to consider truly relevant evidence on defendant’s guilt, that the trial court should be required to apply a particularly stringent test to the admission of 404(b) evidence.
This “common sense” syndrome is particularly dangerous because it is scientifically wrong. Even psychologists err two times out of three when predicting future dangerousness from past behavior. See Weis-senberger, Making Sense of Extrinsic Act Evidence: Federal Rule of Evidence 404(b), 70 Iowa L.Rev. 579 (1985) at n. 78, (citing J. Monahan, Predicting Violent Behavior — An Assessment of Clinical Techniques, 46-49 (1981)). In light of conventional wisdom, therefore, it is likely that the jury will vastly overvalue this relatively poor predictor of behavior.
A jury might also conclude that, whatever his guilt or innocence in a particular case, a defendant deserves punishment simply because he is a bad person. In consequence, it may convict a person more because of his or her past life, than because the jurors are convinced by direct evidence of the charged offense. The effect of this is to “jeopardize[ ] the presumption of inno*852cence of the crime currently charged.” Spencer v. Texas, 385 U.S. 554, 575, 87 S.Ct. 648, 659, 17 L.Ed.2d 606 (1967) (Warren, C.J., dissenting in part and concurring in part). Juries who would hesitate long before concluding the guilt of someone who had never before committed a crime, might be less reluctant when dealing with an individual who has been in and out of prison before, feeling that the harm done is not so great.
Furthermore, improper admission of evidence of other bad acts suddenly burdens the defendant with the defense of unindict-ed charges that he may have defended before and for which he may already have paid his debt to society. See United States v. Burkhart, 458 F.2d 201, 204 (10th Cir.1972). Not only is this intrinsically unfair, but it takes up trial time and may distract the jury from the real issue in the case: whether the government can prove beyond a reasonable doubt that the defendant committed the crime for which he is on trial. The defendant is faced with the prospect of justifying and explaining not just the crime he is charged with, but his entire history. Such a result is antithetical to our system of justice:
However depraved in character, and however full of crime their past lives may have been, the defendants were entitled to be tried upon competent evidence and only for the offense charged.
Boyd v. United States, 142 U.S. 450, 458, 12 S.Ct. 292, 295, 35 L.Ed. 1077 (1892).
Finally, one commentator maintains that it is inherently unfair to force a defendant to rebut this kind of evidence:
More than just creating a risk of the jury reacting emotionally and punitively to an accused’s past misdeeds, when confronted with the act-propensity-act inference the accused is placed in the posture of arguing to the jury that its collective common sense should be disregarded. The accused has little basis for rebutting the evidence except by challenging the force of the propensity inference itself, and the greater its probative value, the more difficult it is for him to bring himself within the ever narrowing class of persons in regard to whom the inference is conceivably inapplicable. The accused must try to demonstrate that the very notion of probative value is inapplicable to him: he is forced to prove that his case is the anomaly.
Weissenberger, Making Sense of Extrinsic Act Evidence: Federal Rule of Evidence m(b), 70 Iowa L.Rev. 579 (1985) at n. 87 (footnote omitted).
The importance of this rule lies precisely in that concern for fairness:
The rule which thus forbids the introduction of evidence of other offenses having no reasonable tendency to prove the crime charged, except insofar as they may establish a criminal tendency on the part of the accused, is not a mere technical rule of law. It arises out of the fundamental demand for justice and fairness which lies at the basis of our jurisprudence.
Lovely v. United States, 169 F.2d 386, 389 (4th Cir.1948).
In sum, Rule 404(b) protects a person’s right to break with the past and create a new life, without forcing that person forever to disprove having a criminal nature. All of the above clearly militates in favor of close scrutiny of purported 404(b) evidence. Something which, I am sorry to say, has not taken place in this case.
II
Rule 404(b) reflects a conscious choice to bar sometimes probative evidence of one sort and for one purpose; by the same token, it forbids only the inference from propensity or criminal tendencies to the specific act charged in the indictment. Using other bad acts for purposes other than showing a propensity towards crime does not deny people the chance to rehabilitate themselves; the probative value does not rely on the aphorism “once a criminal, always a criminal.” Comparing the structure of Rule 404(b) with the structure of 404(a) demonstrates how the rule addresses the concerns previously discussed. Part (a) of Rule 404 sets out a general prohibition on the use of character evidence to show *853conduct, and some exceptions to that proscription.
(a) Character evidence generally. Evidence of a person’s character or a trait of character is not admissible for the purpose of proving action in conformity therewith on a occasion, except: [Character of the accused, the victim, or a witness, under special circumstances]. (Emphasis added).
Rule 404(b), however, is different in a significant way. It sets out, not a general rule with some exceptions, but a series of uses for a certain kind of evidence (bad acts), one of which is forbidden, and the rest of which may be permissible.
(b) Other crimes, wrongs, or acts. Evidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show action in conformity therewith. It may, however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident. (Emphasis added).
Motive, opportunity, intent, etc., are not exceptions to the general proscription, they are other uses of this kind of evidence. The consequence of this, of course, is that one may not transgress the first prohibition in attempting to fit within a permitted use of bad acts evidence, as one could if these latter uses were exceptions to the rule.
Stated otherwise, “the issue is whether [the evidence] proves a material issue without requiring any inference to the defendant’s criminal disposition.” 22 Wright & Graham, Fed. Practice & Proc., § 5239, at 432 (Supp.1987).
A better way to state the rule ... is that the rule admits evidence of other crimes whenever it is relevant without using the inference of character anywhere in the chain of inference. Or as the California Supreme Court has put it in negative form, the rule excludes evidence of other crimes in any case in which one of the inferences in the chain of circumstantial evidence is the inference from the act to the defendant’s character or propensity to commit crimes.
Id. at 438 (Supp.1987) (footnotes omitted).2 At times, however, a court circumvents the ban 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.
That is exactly what happened here. The judge instructed the jury that it “may not consider [the prior conviction] as evidence that the defendant ... acted the same way, that is, in conformity with the way that he had acted in the past. However, [it could] consider that evidence only for its tendency ... to show knowledge or intent on the part of the defendant.” When pressed at oral argument to explain how the jury could apply this instruction, however, the government could only say that, since Rubio had intended to distribute before, the jury could believe he was more likely to intend to distribute again. This is precisely the circumstantial chain that is prohibited since all that it proves is that because he had a bad character in the past, he is likely to have committed the present crime. The only reasoning the jury could follow, even with this instruction, is that persons who have knowingly dealt in drugs previously have a character such that they are more likely than others to do it again with the requisite knowledge and intent. The silent but inescapable logical link between the two inferred mental states is Rubio’s propensity towards dealing in drugs. Thus, the evidence of past crimes was inadmissible under Rule 404(b).
Ill
A.
Attempting to cure this deficiency in the instruction, the majority, on its own, sug*854gests some alternate inferences, none of which can be found in the government’s brief or its arguments below. With due respect to my well intentioned colleagues, it is the prosecution, at trial, who is responsible for articulating the permissible inferences claimed from a certain piece of evidence;3 and it is the trial judge who, at trial, is charged with ensuring, through an adequate limiting instruction, that the jury understands that proper use. It is necessary, therefore, that they, and not appellate judges, make explicit every link in the chain of reasoning that makes an uncharged crime or bad act properly relevant to the charge in the indictment.
There is little practical value in proposing for the first time at the appellate level a set of permissible inferential chains that the jury could have used if it had thought of them. By that time the cat is most assuredly out of the bag. The importance of making these logical chains explicit for the jury during trial, and not when the case is on appeal, is that it provides a minimal safeguard against the jury’s improper use of the evidence, which is, as I have explained, all too likely. Rule 404(b), after all, is intended to safeguard against the jury’s improper use of “bad acts” evidence.
I am not proposing a revolutionary idea; Judge Coffin, of this Circuit, urged the same approach ten years ago:
Often the relevance of a piece of evidence is so obvious that a judge need spend no effort in justifying his ruling. But in doubtful cases, the following is prudent advice:
“Where relevancy is not immediately apparent, the judge and counsel should clearly identify the terms of the relevancy relationship in the particular case. That is, they should describe the item of evidence being proffered, the consequential fact to which it is directed, and the hypothesis required to infer the consequential fact from the evidence. Without this analysis it is impossible to decide how the evidence may alter the probability of the existence of the consequential fact.”
United States v. Mann, 590 F.2d 361, 369 (1st Cir.1978) (quoting 1 Weinstein’s Evidence, 11401[08]). See also United States v. Flores Perez, 849 F.2d 1, 7-8 (1st Cir.1988).
B.
In addition, the majority rests some of its inferences on an incorrect reading of the record. My brothers make much of the white powder found on and around the scale, claiming that its description “matches what later expert testimony described as inositol, lactose, or lidocaine (substances often used to dilute cocaine when it is sold).” Ante, at 847. The majority then argues that the prior conviction is relevant because Rubio should have recognized the powder as a substance used to dilute cocaine, ante at 847 and 848. It also, I suppose, provides the foundation for the argument regarding Rubio’s knowledge of how to “cut” cocaine, ante at 847-48, since there is no other evidence in the record suggesting that any “cutting” took place in the defendant’s house.
A brief portion of the trial transcript reveals the lack of any basis from which to draw any conclusion regarding that white powder — other than that it was powdered in texture and white in color. Shortly after the jury began to deliberate, it came back with two questions, the second of which was addressed by the court in the following manner:
With respect to the second question: Was the powder found on the scale, Exhibit 6, and in the box, Exhibit No. 5, tested? The parties have agreed that I may respond as follows: The powder allegedly found on the scale was tested for cocaine and found not to be cocaine. The second part of the question: if so, and that means if it was tested, what was it and is it of the type used to cut cocaine? The response which the parties agree I *855may give to that question is: There is no evidence of what it is or whether or not it was of the type used to cut cocaine.
Trial transcript, at 390 (emphasis added). “No evidence” simply means no evidence. There was no description of that powder other than to say it was white powder and not cocaine; which could make it anything from salt, to baby powder, to refined sugar.4 The expert testimony my brothers allude to is merely a generalized description of how cocaine is diluted for sale. With no evidence regarding that powder, the jury could not properly have concluded anything at all from it. Yet, the majority relies on just such an impermissible conclusion regarding the powder as the basis for the attenuated chain of reasoning it uses to justify the admission of evidence of an unrelated crime.
C.
Finally, I come to the actual inferential chains propounded by the majority. With all due respect, I cannot agree that they bring the prior conviction within the bounds of Rule 404(b), even laying aside the problems with their factual underpinnings. The first of these arguments (there appear to be three) I simply do not understand. The majority correctly points out that Rubio’s prior conviction makes it more likely that he knew that the scales he owned, as well as the ledgers he kept and the powder found on his scales, were of the type often used for dealing in cocaine. Then my brothers conclude that:
a person who knew such drug-related items are used to help sell drugs (which he denies knowing about), is more likely to have known about, and intended to distribute, the drugs, than a person who does not know these drug-related items are used to help sell drugs.
Ante, at 848.
This reasoning obscures, but does not avoid, reliance on the inference from character. The relevance of these items lies in Rubio’s possession of them, and his prior conviction does not make it any more likely, except through the character inference, that he was using them to deal cocaine. According to the government’s theory of the case, the jury had to decide whether (a) the objects belonged to Rubio, and (b) he used them to conduct drug transactions. If Rubio was in fact using them to deal in drugs, then he obviously knew that they could be so used, whether or not he was previously convicted. To suggest, as the majority does in the quoted passage, that the jury may have compared Rubio against some hypothetical person whose mind the jury somehow knew was unclouded with any knowledge of drug dealing, in order to evaluate relative probabilities of guilt, is creative fiction.
In reality, the only way Rubio’s knowledge could make his illegal use of the scales and the ledger at all more likely is his knowledge of how to use them, which my brothers compare to the knowledge required to fly an airplane. Fanciful flight is not at issue here; the knowledge here at issue is this: how to weigh a valuable substance on a scale, how to keep track of who owes how much,5 and how to mix two powders. See ante at 847-48. Can one seriously argue that this knowledge sets Rubio apart from the general population, like the ability to fly a plane sets a pilot apart? The majority is in effect calling for the abrogation of Rule 404(b) insofar as previous similar crimes are concerned. After all, a previous bank robbery or a previous burglary will also give an accused the knowledge required to commit those crimes.
The third inference is also improper. The majority argues that someone with drug-dealing experience (such as Rubio) *856would recognize the items in the closet and realize that cocaine dealing was taking place in the house. Quite simply, I fail to understand how this fits the government’s theory of the crime. If the items belonged to Rubio, and he was using them as cocaine distribution paraphernalia, as the government contends, then his knowledge of the dealing occurring in his house is a foregone conclusion. Under this scenario, the 404(b) evidence adds nothing to the government’s proof. The majority, however, appears to have in mind a case where the government has alleged a conspiracy, in which Rubio denied taking part. If that were so, then his likely knowledge of these items, along with his inferred knowledge of their uses, would be relevant to show his knowledge of, and possible acquiescence in, the conspiracy. But that is not the case before us. That theory was never presented at trial, and the jury could not have convicted Ru-bio on that basis.
In sum, the items themselves are damaging evidence against Rubio. But his knowledge of those items and their uses is not at issue. Only his knowledge of the cocaine and his intent to distribute it are at issue. On its own, without the prior conviction, his ownership of these incriminating objects clearly made those elements of the crime more likely. The conviction itself, however, adds nothing to their probative value and was therefore not admissible under the theories advanced by the majority-
Contrast those theories with the following examples which in my view constitute permissible uses of Rule 404(b) evidence. To show intent to distribute, for instance, one could introduce evidence that Rubio had agreed to sell to a third person the amount of cocaine found in his house. This evidence obviously tends to prove that he held the cocaine in order to distribute it. To show his knowledge of the presence of that cocaine in his house, one could introduce evidence that he kept other illegal merchandise in the same closet in which the cocaine was found. This evidence would trigger the concerns addressed by Rule 404(b), but the fact that Rubio often looked in that closet would increase the likelihood that he knew of the cocaine hidden therein, so it would be admissible. If Rubio’s defense had been that he had seen the glassine bags of cocaine, but did not know the nature of the white powder they contained, then his previous conviction for selling cocaine would have been admissible. It then would demonstrate contact and familiarity with cocaine, thus disproving his claimed ignorance.
Here, there was no such permissible logical inference. As discussed above, bad character, the forbidden inference, provided the only logical link between the prior conviction and the fact to be proved. The evidence therefore fails the Rule 404(b) test, and its introduction was error.
IV
Although the majority reaches a different conclusion, I think it is clear that the inferences proposed by the majority add to the government’s case, if anything at all, so little as to almost certainly fail the requisite balancing of prejudicial effect against probative worth. The government is fond of reminding us that in this context Rule 403 is a “rarely invoked limitation[ ],” that “has not proven an especially fertile source of assistance to criminal defendants.” United States v. Zeuli, 725 F.2d 813, 816-17 (1st Cir.1984). That does not mean, however, that this circuit has or should abdicate its responsibility to review the district court’s balancing under Rule 403. See United States v. Flores Pérez, 849 F.2d 1, 6-7 (1st Cir.1988). Rubio’s defense rested not on the ignorance the majority tries to disprove, but on ignorance of the presence of the cocaine in his house. His previous conviction is singularly unhelpful, absent recourse to an inference from propensity, in proving his knowledge of that presence. The government, therefore, had little need for this evidence. On the other hand, the conviction’s prejudicial effect is clear, as discussed in the first pages of this dissent.
As a result, Rubio did not receive the fair trial to which he was entitled. I, therefore, dissent.
. Fed.R.Evid. 404(b):
Other crimes, wrongs, or acts. Evidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show that he acted in conformity therewith. It may however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.
. Weissenberger says essentially the same thing: "[T]here are no exceptions to the forbidden inference of rule 404(b). It may never be used by a litigant.... The second sentence of rule 404(b) merely contains a suggestive, nonex-haustive list of traditionally applied theories of relevance that do not capitalize upon the prohibited inference designated in the first sentence of the rule.” Weissenberger, ante, at n. 47.
. Compare, for example, Huddleston v. United States, — U.S. -, 108 S.Ct. 1496, 1499, 99 L.Ed.2d 771 (1988) explaining the prosecution’s theory of relevance which was, in any event, conceded to be proper by the petitioner, and United States v. Vest, 842 F.2d 1319, 1324-25 (1st Cir.1988) where the prosecution advances several theories of relevance.
. I would add only the following thought. If this powder had been inositol, lidocaine, or lactose, it would easily have been the most powerful evidence against Rubio in the case, given his admission that he owned the scale. If it could plausibly have been identified as one of these substances, therefore, one would expect the government to have introduced evidence to that effect — especially after it was tested to determine its nature.
. According to the government expert’s testimony, the drug ledger merely consists of pages headed by names, recording amounts owed and amounts paid.