Novelty, Inc. v. Drug Enforcement Administration

BROWN, Circuit Judge,

dissenting:

Tellingly called poor man’s crack (especially telling, as crack is already poor man’s cocaine), methamphetamine— meth — is a national scourge. Behind only “alcohol and marijuana as the drug used most frequently in many Western and Midwestern states,” Methamphetamine, http://www.usdoj.gov/dea/concern/meth. html (last visited May 19, 2009), meth addiction can cause “paranoia, auditory hallucinations, mood disturbances,” and even “homicidal or suicidal thoughts,” Meth Awareness, http://www.usdoj.gov/ methawareness/ (last visited May 19, 2009). “A fairly common hallucination experienced by meth users is the so-called crank bug” where a “user gets the sensation that there are insects creeping on top of, or underneath, her skin,” causing her to “pick at or scratch her skin trying to get rid of the imaginary bugs,” “open[ing] sores that may become infected.” Id. Meth also “reduces the amount of protective saliva around the teeth,” which, along with the fact that “users also consume excess sugared, carbonated soft drinks, tend to neglect personal hygiene, grind their teeth and clench their jaws,” causes “what is commonly called ‘meth mouth,’ ” as addicts’ teeth “fall out” “even as they do simple things like eating a sandwich.” Id. Meth is a very bad thing. A monster.

But we don’t toss the law aside in our zeal to eradicate even an obvious menace. *1171The Deputy Administrator (DA) here has done just that,1 in the process crippling a successful enterprise and costing many employees their jobs. In rejecting the ALJ’s recommendation that Novelty be allowed to retain its registration, the DA transforms a trivial violation of Novelty’s own rules into an imminent danger to the public health and shifts the burden to Novelty to explain why some of the thousands of convenience stores it services are busier than others. Subject to this perverse alchemy, ordinary business practices somehow provide proof of rampant lawlessness. Thus, the DA (1) misreads a statute so any place used for distribution is a “principal place of business,” even a padlocked storage shed; (2) uses the average sales of every convenience store serviced by Novelty as a proxy for legitimate demand at each location without regard to any individual characteristics; and (3) finds Novelty’s efforts to stay in business after its registration was suspended to be proof of villainy.2 Because “[t]he war on drugs is not an excuse to violate the norms of fair play and evenhandedness,” United States v. Cuellar, 478 F.3d 282, 307 (5th Cir.2007) (en banc) (Smith, J., dissenting), rev’d — U.S.-, 128 S.Ct. 1994, 170 L.Ed.2d 942 (2008), I respectfully dissent.

I.

To appreciate how poorly supported the DA’s decision really is, a little history is helpful. This case first came before us when Novelty challenged the DA’s suspension order, but the DA mooted that initial challenge by revoking Novelty’s registration before oral argument. To justify that initial suspension, the DA gave five reasons why Novelty’s continued operations endangered the public. First, Novelty violated 21 U.S.C. § 822(e) by distributing from unregistered locations. In re Novelty Distrib., Order to Show Cause & Immediate Suspension of Registration, slip op. at 1 (Jan. 17, 2008). Second, Novelty sold products to stores in amounts that “far exceeded what those retail outlets could be expected to sell for legitimate, therapeutic purposes,” and some drugs were diverted to a Connecticut meth lab in 2002. Id. at 2. Third, Novelty “could not account for more than 60,000 dosage units of two ephedrine products,” and there were other gaps in the company’s records. Id. Fourth, “[f]rom January 1, 2007, through July 9, 2007, Novelty distributed scheduled listed chemical products on at least 284 occasions to 35 retail outlets that were not self-certified under” federal law. Id. Finally, Novelty distributed products in unlawful containers and in forms that violated state laws. Id. at 2-3.

The case went before an ALJ, who rejected most of these allegations as factually unsupportable. For instance, it’s just not true that Novelty distributed drugs to stores that were not self-certified. ALJ Ruling, slip op. at 91. Nor did Novelty *1172sell drugs in forms or packages that violated any law. See id. at 37-39. And the DEA’s audit failed to consider all the data it received from Novelty: “Specifically, the DEA had failed to take into account dosage units returned to vendors, data that had been removed from the inventory because of expired or obsolete inventory, and other unspecified documents reflecting miscellaneous transfers or adjustments.” Id. at 87-88. Based on the evidence, the most that could be said about those “60,000 dosage units” was that Novelty’s record-keeping was inadequate. Id. at 88. In her revocation decision, the DA conceded the ALJ was correct on these points. See Novelty Distrib., Inc., 73 Fed.Reg. 52689, 52695 (Sept. 10, 2008) (Revocation of Registration) (“At most, the evidence suggests that [Novelty] made a single distribution to a single store before it obtained its certification.”); id. at 52695-96 (form and packaging); id. at 52696-97 (audit).

The ALJ also debunked the DEA’s argument that Novelty was selling over-the-counter drugs in amounts that exceeded legitimate demand. Jonathan Robbin, the DEA’s expert, concluded “a convenience store is expected to sell no more than $14.39 worth of ... products per month,” even though one 24-count package could cost up to $14.99. ALJ Ruling, slip op. at 96 & n.40. In other words, selling even one package per month would be suspicious. The ALJ set forth the flaws in this analysis. Robbin, for example, used the wrong denominator, thereby “forcing a lower expected sales value.” Id. at 55. He did “not provide a detailed breakdown” of how he determined important inputs in his formula, “other than providing a narrative of his approach,” and he offered data about the entire retail industry and just not convenience stores. Id. He ignored that much of his data relied on self-reporting, and instead of reporting sales as “cold and cough products,” some stores used the “general merchandise” category. Id. at 56-57 n.28. He also did not do a multivariate analysis to consider “[f]aetors such as store hours, store location, store size, advertising expenditures, and management practices,” or other “seasonal and environmental variables.” Id. at 58. Unsurprisingly, the DA did not defend Robbin’s analysis. 73 Fed.Reg. at 52693-94.3

Thus, all that lingered of the DA’s aggressive allegations was that Novelty broke the law by using unregistered storage units; that Novelty’s records were deficient in certain respects; and that a small amount of product had been diverted. In revoking Novelty’s registration, the DA also emphasized Novelty’s letter to *1173one of its suppliers that supposedly manifested an intent to circumvent the suspension order, and that Novelty did not comply consistently with its own anti-diversion rules. Finally, because Robbin’s statistics were useless, the DA invented, seemingly sua sponte, a new statistical analysis, placing the burden on Novelty to rebut it. These grounds, though much less compelling than the grand claims made in the suspension order, were enough for the DA to shut down Novelty’s list I chemicals operations for good.

To say the case for revocation is razor thin greatly exaggerates its sufficiency. As the statute is written, the storage units are not illegal. The DA’s newly-concocted statistics are just as bad as Robbin’s, and the “smoking gun” letter is anything but. Moreover, though Novelty did not always live up to its own policies, mistakes were rare, and the DA failed to even consider the relevant denominator. True, company records were somewhat deficient, but no DEA agent had ever before complained of Novelty’s “thorough” records, despite multiple inspections, ALJ Ruling, slip op. at 84, 93, and, in any event, the DA said if the only problem were recordkeeping, “imposing compliance conditions might well protect the public interest.” 73 Fed.Reg. at 52703. If “arbitrary and capricious” is to mean anything in the DEA context, this decision cannot be allowed to stand.

II.

The heart of the DA’s analysis — that unregistered storage units always violate federal law — contradicts the statute, Chevron notwithstanding. Under no reasonable reading does 21 U.S.C. § 822(e) require each link in the distribution chain to be registered. Nonetheless, in revoking its registration, the DA gave top billing to Novelty’s supposed lawbreaking: “First, for more than three and a half years, Respondent disregarded a DEA letter specifically warning it that its use of 150-180 self-storage units ... violated Federal law.” 73 Fed.Reg. at 52703. In fact, not only did the DA lean on this finding for purposes of 21 U.S.C. § 823(h)(2) (compliance with law), she double-dipped and used again it for § 823(h)(4) (past experience). See id. at 52702. But she was wrong; it was not Novelty that bent the law.

Section 822(e) says a “separate registration shall be required at each principal place of business or professional practice where the applicant manufactures, distributes, or dispenses controlled substances or list I chemicals.” 21 U.S.C. § 822(e). Novelty argued its scores of run-of-the-mill self-storage units were not “principal places of business,” and thus did not need to be registered. The DA rejected this position, determining that federal law “requires that a separate registration be obtained at each location at which List I chemicals are distributed.” 73 Fed.Reg. at 52700.

Hogwash. Even with the winds of deference at its back, this interpretation of § 822(e) is a nonstarter. It reads the word “principal” out of a statute whose plain language requires two things: (1) “distribution]” from (2) a “principal place of business.” The question is simple: if all locations where distribution occurs must be registered, why would Congress specifically refer to “each principal place of business”? To make the word “principal” disappear, the DA resorted to magic tricks. She determined each “location” where “List I chemicals are ... distributed from” is a “principal” one because it “plays a ‘consequential’ part in the registrant’s activity of distributing.” Id. at 52701 (quoting Webster’s Third New Int’l Dictionary 1802 (1976)).4 Hence “princi*1174pal place of business where a registrant distributes list I chemicals” = “consequential” place of business = “each location where list I chemicals are distributed.” (abracadabra omitted). But after the smoke fades away, the DA’s approach to the word “principal” causes § 822(e) to read like this: “A separate registration shall be required at each location where list I chemicals are distributed where the applicant distributes list I chemicals.” That is a silly way to read a statute. E.g., Reiter v. Sonotone Corp., 442 U.S. 330, 339, 99 S.Ct. 2326, 60 L.Ed.2d 931 (1979) (“In construing a statute we are obliged to give effect, if possible, to every word Congress used.”).

To be sure, § 822(e) contains ambiguity, as it invokes a common term of art in a way that is not consistent with its ordinary usage. “Principal place of business” is most prominently used in 28 U.S.C. § 1332(c), relating to corporate diversity, and it is settled that for diversity jurisdiction there can only be one “principal place of business.” E.g., Capitol Indem. Corp. v. Russellville Steel Co., 367 F.3d 831, 835 (8th Cir.2004). Section 822(e), by use of the word “each”, contemplates multiple such places. But just because a statute is ambiguous does not mean any interpretation of it is reasonable. See, e.g., Int’l Alliance of Theatrical and Stage Employees v. NLRB, 334 F.3d 27, 34-35 (D.C.Cir. 2003). Though the language here cannot mean the exact same thing as in § 1332(c), it is unreasonable to find that any place where distribution occurs is per se “principal.” No matter how defined, “principal place of business” surely means more than that. If every place is “principal,” no place is.

The DA mused, however, that any other reading than hers “would clearly frustrate the Congressional purpose,” which was to prevent “diversion by requiring that those persons who propose to engage in the legitimate distribution of ... listed chemicals apply for a registration, notify this Agency of the proposed location of their activity, and submit the facility for inspection ... to ensure that it has adequate security controls and procedures.” 73 Fed.Reg. at 52701. For this proposition, she cited 21 U.S.C. 822®, which reads: “The Attorney General is authorized to inspect the establishment of a registrant or applicant for registration in accordance with the rules and regulations promulgated by him.” This is not at all sufficient to support her construction of § 822(e) as “establishment” — used in § 822(f) — is not the same as “principal place of business.” The surest Rosetta Stone into “Congressional purpose” is the language Congress actually uses, but the DA’s construction of § 822(e) effectively deletes Congress’s word “principal” from the U.S. Code.

The DA then explained that accepting Novelty’s view “would also create a perverse incentive” for registrants to “keep adding warehouses or storage facilities so that it could eventually claim that its warehouses were no longer principal places of business and thus were not subject to the registration requirement.” 73 Fed.Reg. at 52701. Perhaps, but if records are kept at these hypothetical warehouses, if employees work there full-time, if they are listed in the phone book, if drugs are kept there for an indefinite duration, and so on, then there may be an argument that these warehouses are “principal places of business.” The DA did not even attempt to offer such a nuanced test. Instead, she *1175found all storage units must, as a matter of law, be registered. How is a shed, locked with a padlock, a “principal place of business” for a company like Novelty with hundreds of employees and thousands of customers across the nation?5

What is most aggravating is that reading the statute correctly would not create the dire dichotomy painted by the DA, one in which any other view but hers would “render the Act a nullity.” Id. The DEA still has recourse to the first factor under 21 U.S.C. § 823(h), relating to the risk of diversion. If a registrant’s distribution practices are not safe, that problem can be addressed by the statutory scheme as written. There is no need to mangle the text to make unsafe distribution an independent violation of federal law. However, instead of simply relying on that first § 823(h) factor (perhaps because there is almost zero evidence of actual diversion), the DA adopted an atextual claim of lawbreaking to bolster the argument in favor of revocation.

III.

The DA also botched her diversion analysis. Let’s be clear: there is remarkably little evidence of actual diversion in this record. The best the DA can muster is an event from 2002 when “twenty-two [sixty-count bottles of an ephedrine-based product] were found at an illicit methamphetamine laboratory in Thompson, Connecticut.” 73 Fed.Reg. at 52694. Novelty, however, did not produce those bottles, and it is uncertain whether the company even distributed them. See ALJ Ruling, slip op. at 74. The DEA, it seems, also never formally warned Novelty about the diversion until it took action against the company in 2008. 73 Fed.Reg. at 52702. But even given the benefit of the doubt, the DA can only point to one instance of diversion, despite Novelty’s hundreds of millions of doses sold.6 See, e.g., Pet’r’s Br. at 15 (Novelty has sold between three and five hundred million doses).

Because proof of actual diversion was paltry, revocation rested on Novelty’s theoretically ineffective controls against diversion. There is much in the DA’s reasoning that I sympathize with; I too wonder whether it is prudent to keep drugs in storage sheds or on trucks for over a week.7 Likewise, though her concern about Novelty’s recordkeeping was overstated, I agree it is important to keep complete records, and if Novelty had concerns about a particular store, it should have been especially vigilant. At the same time, as the ALJ found, Novelty took “a proactive role in ensuring its customers are self-certified,” made sure its customers used “lockable, plexiglass display cases for the display and sale of’ drugs, and “when the DEA brought the *1176recordkeeping errors” to Novelty’s attention, Novelty “identified a reasonable reason for the errors, and acted to correct those errors by upgrading its computer system.” ALJ Ruling, slip op. at 98-99. Novelty also credibly offered to modify its distribution network to address safety concerns. Id. at 100-01. (The DA wrongfully rejected the ALJ’s credibility finding based on her erroneous belief that the storage units were illegal, 73 Fed. Reg. at 52703). Given the large number of transactions, moreover, some mistakes will happen; what human endeavor is done perfectly millions of times in a row? Nonetheless, there is some reason to conclude Novelty’s protections against diversion — even if factually “effective,” as the statute requires, 21 U.S.C. § 823(h)(1)— were imperfect, though such oversights should be addressed by compliance conditions far short of the corporate death penalty that is revocation.

But instead of just relying on this theoretical diversion, the DA went further and reasoned Novelty’s own statistics “showing its sale of [drugs] in the three months prior to the issuance of the suspension order” demonstrated that Novelty did not maintain effective controls against diversion. 73 Fed.Reg. at 52699. The support for that claim is embarrassingly weak. Despite the almost self-evident fact that not all convenience stores are created equal (for instance, the AM/PM next to a major freeway may be a lot busier than Ma’s Gas-N-Go in Middle-of-Nowhere Alabama), the DA conjured a very blunt instrument to derive legitimate market demand: Novelty’s average sales data. Instead of relying on a multivariate statistical study that could have accounted for factors affecting sales at individual stores, she operated on an unstated assumption that every store in the country should request an amount close to the average, meaning any request above the average suggests a likelihood of diversion. But what a store should request is based on legitimate, non-diversionary demand for the product at its location. Just as a store selling below the national average might still show a risk of diversion if it requests many more drugs than non-diversionary demand would warrant at its location, a store selling above the average does not show a diversion risk if it requests the amount which would be expected at its location. The DA’s approach ignored this basic and obvious point.

Similarly, as Novelty’s expert observed, “if an average legitimate sales figure was derived from data from both high-sales revenues stores and low-sales revenues stores, but then that average legitimate sales figure was used to predict sales in only the high-sales revenues stores, the prediction would suffer from aggregation bias.” Instead of accepting this unassailable logic, the DA said “proof that a distributor is selling quantities in excess of the national monthly average of sales of [drugs] by convenience stores would be highly probative of the distributor’s lack of effective controls against diversion.” 73 Fed.Reg. at 52700. This tells us very little about anything; after all, 50% of all convenience stores should be expected to have sales greater than the average.

The DA then said “[m]ore specifically, a registrant cannot ignore evidence that some of its customers are purchasing quantities that greatly exceed what its typical customer buys from it.” Id. This statement, of course, seems a little more persuasive (but only a little); if one assumes all convenience stores are similar, and if a store is receiving many more drugs than others, that should raise a red flag — but, of course, not all stores are similar. The DA concluded by stating Novelty “offered no explanation that was specific to any store for why it was selling in such quantities.” Id. But the burden of proof is *1177on the government before it can take away someone’s livelihood, see 21 C.F.R. § 1309.54(b), and even if it were otherwise, this is an unfair obligation to spring on Novelty as the company was responding to the statistics of DEA’s so-called expert-statistics that even the DA does not defend, 73 Fed.Reg. at 52693-94. In any event, there is no basis to shift the burden to Novelty because we should expect statistical outliers. See id. at 52700 n.45. Of the thousands of stores serviced by Novelty across the country in countless different climes and circumstances, is it really unthinkable that a small subset of them will be much busier than the others? This statistical inevitability is not something that a company should have to explain, nor should the DEA be able to escape its burden of proof so effortlessly.8

The DA’s problems with large numbers do not end there. Ignoring that no complex system can be 100% foolproof, the DA applied an unreasonable zero-tolerance standard that Congress surely did not intend.9 She found, for example, that Novelty “violated its case limit policy [i.e., its own policy that it would only sell one 24-count package to a store per visit] some 85 times” during a six-month period, 73 Fed. Reg. at 52699. That sounds pretty bad, but the DA omitted the denominator; according to the record, the number of transactions during that year was “approximately 100,000 to 120,000.” The error rate thus may have been less than .2% (no one knows for sure the exact rate because the DA did not even attempt to discern it using any methodology whatsoever, Judge Tatel’s or otherwise, see Tatel Op. at 1183). Despite the burden resting with the government, the DA did not explain how much risk is too much, nor did she give any context to those “85” violations (a number, by the way, that Novelty says is at least forty too high). All we know is sales staff may have mistakenly sold extra packages of product to convenience stores' — not to meth makers, mind you — less than two times per thousand transactions. Yet somehow Novelty’s “failure to enforce its own policies provides reason alone to conclude that it cannot be trusted to adhere to compliance conditions”? 73 Fed.Reg. at 52704.

IV.

In revoking the registration, the DA also stressed Novelty’s attempt to find a new way of doing business once its registration was suspended. Though conceding Novelty did nothing illegal, she nonetheless found Novelty did more than “merely think[ ] about a legal option or seek[ ] legal advice about the scheme,” but also “affirmatively sought out one of its suppliers and attempted to induce it to enter the scheme only to be rebuffed by the supplier.” 73 Fed.Reg. at 52703. The DA concluded “[i]t would fundamentally undermine the [Controlled Substance Act’s] purpose of protecting against diversion to allow an entity whose registration has been revoked to subsequently engage in *1178the same or related activities as an agent.” Id.

Here are the real facts. In January 2008, the DEA suspended Novelty’s registration. After receiving the suspension order, Novelty sent a letter, which I quote, explaining its “plan to get back in the ephedrine business” to one of its suppliers. Letter from Novelty, Inc. (under seal). Under the “plan” — which was “still [being] finaliz[ed] ... with the DEA to insure everything can work smoothly and we do not have any issues” and was awaiting “approval” before any sales would occur— Novelty’s sales force would “tak[e] orders” which would be sent “to a third party logistics company for shipping and handling” to “fill the order ... via UPS or similar carrier.” Id. The “costs [to the owner of the store would] remain the same,” and “[a]ll invoices [would be] from Novelty.” Id. The owner of the store would “have the choice of stocking the [drug] cabinet or holding it for [Novelty’s] sales person.” Id. The effect would be to “basically keep[ ] the business the same,” except with “a UPS box.” Id. The supplier declined to participate, but a Novelty executive testified during the administrative process that Novelty was “still continuing to explore” this idea.

The DA’s reading of the letter is unfair. Making tentative plans — plans contingent upon DEA authorization no less — suggests nothing nefarious. Novelty, moreover, planned on substantially changing its business, including no longer using its own distribution network (i.e., the very network at issue in this case). Under 21 U.S.C. § 822(c)(1), a sales agent need not be registered if he “is acting in the usual course of his business or employment,” which, under this completely new business model, seemingly would include Novelty. Granted, Novelty said its plan would “basically keep[ ] the business the same,” but, in context, the only plausible reading of the sentence is the business would remain the same for Novelty’s customers, not for Novelty itself. How could the business be unchanged for Novelty which would not longer do the shipping and handling of these drugs?

Neither of my colleagues reaches this issue (though both hint they might agree with my analysis). This is a mistake. The DA employed a totality of the circumstances test, but the only evidence she relied upon in assessing 21 U.S.C. § 828(h)(5) — the statute’s catch-all category — was Novelty’s letter. See 73 Fed.Reg. at 52702-03. Because the DA erred as to this factor, and because we do not know how much weight she gave it in her analysis,10 at a minimum this court should remand for her to reweigh the evidence. See, e.g., PDK Labs., Inc. v. DEA, 362 F.3d 786, 799 (D.C.Cir.2004) (“If the agency’s mistake did not affect the outcome, if it did not prejudice the petitioner, it would be senseless to vacate and remand for reconsideration. But in this case we cannot say that the Deputy Administrator’s error was of that sort ... [as he] stated that it was ‘the totality of circumstances’ that led him to sustain the suspension orders.”).

*1179I close with a few words on how easily the administrative state can slip its leash. A familiar argument for enhanced administrative authority (and hence diminished judicial review) is the need for “flexibility,” as old-fashioned courts are ill-suited to deal with the complexities of the modern world. E.g., Cass R. Sunstein, Law and Administration After Chevron, 90 Colum. L.Rev. 2071, 2079 (1990) (citing, inter alia, J. Landis, The Administrative Process 6-12 (1938)). That may be true. But the flipside of flexibility is certainty, consistency, evenhandedness, and predictability— those Rule of Law values that mark a free society. E.g., Thomas Paine, Common Sense 31-32 (Dover Thrift ed., 1997) (1776) (“[L]et a crown be placed thereon, by which the world may know, that so far as we approve of monarchy, that in America the law is king. For as in absolute governments the king is law, so in free countries the law ought to be king; and there ought to be no other.”). If my splintered-on-seemingly-all-points-but-outcome colleagues are right that despite her many errors, the DA’s decision to revoke Novelty’s registration still must be upheld, then I fear we have traded away far too much law in our bargain for elasticity.

But in the end, because of this “flexibility,” it is probably for the best that the DA’s decision is upheld today. If the registration was not revoked this time, it surely would have happened next time. On this record, I have no confidence that Novelty would ever receive a fair shake from the DEA. Indeed, if we were reviewing a district court that acted like this, I would not remand to that court. But the DEA is the only game in town for drug registrants, and in deciding whether to revoke a registration the DA balances a number of open-ended factors with no requirement “to make findings as to all of the factors,” and with power to “give each factor the weight [s]he deems appropriate.” Morall v. DEA 412 F.3d 165, 173-74 (D.C.Cir. 2005). Add Chevron deference to that mush and we have a very powerful Deputy Administrator. In other words, if in the future the DA were to find any errors in Novelty’s records or operations (and, given that Novelty sells millions of doses a year, it is inevitable that she would), then she could revoke the registration by claiming a pattern of dangerous recordkeeping and distribution errors, knowing full well that Novelty would face an uphill scramble to persuade a court that she has abused her overflowing discretion. Given the law of large numbers, such a pattern could be found in any registrant’s records, meaning all registrants live by the grace of the Deputy Administrator.

No, old-fashioned law will not save Novelty and the jobs of its employees. It does not matter that no Novelty executive has ever been convicted of a crime. It does not matter that notwithstanding Novelty’s millions of sales, the best evidence the DA can point to of diversion is one — one!— instance from over six years ago. It does not matter that the DEA inspected Novelty’s records for years and never peeped about a problem before deciding to bring down the full weight of the Executive Branch on Novelty’s head. It also is irrelevant that Novelty has credibly offered to overhaul its internal processes to comply with the DA’s whims. When an agency has gone rogue, and when judicial review is gutted, the only thing left is the Law of the Jungle, the weak versus the strong. And in this war of all against all, who can withstand the might of the federal government?

. As the Administrative Law Judge (ALJ) perceptively observed, no party "disputéis] that illegal methamphetamine is a major drug problem in the United States,” but the agency "seems to be trying to remedy this problem by restricting or eliminating the availability of such over-the-counter products by removing the distributors of [these] products to convenience stores from the market place” altogether, despite lacking sufficient record evidence. In re Novelty Distrib., No. 08-33, slip op. at 93 (May 21, 2008) (Recommended Ruling of the Administrative Law Judge) (“AU Ruling”).

. Alas, these are not the only things the Drug Enforcement Administration (DEA) did wrong. Agents interfered with Novelty’s understandable efforts to document in real time the search of the company’s own facility by forbidding the use of and disabling video and audio recorders. These agents also “requested 12,000 additional pages of documents,” giving only "three hours to produce them,” and an executive believed he would be arrested because he could not do the impossible. AU Ruling, slip op. at 62-65. Such tactics do not reflect well on the United States.

. Though the DA did not uphold Robbin in this case, he has been relied upon in more than twenty other cases, even though his approach — aside from being internally defective — does not appear even to yield consistent results. Compare Sunny Wholesale, Inc., 73 Fed.Reg. 57655, 57658 (Oct. 3, 2008) ("Mr. Robbin explained ... that during 2005, '[t]he expected average monthly convenience store sales of nonprescription drug products containing pseudoephedrine (hcl) in Georgia were * * * $82.’ ”) with Holloway Distrib., 72 Fed.Reg. 42118, 42119 (Aug. 1, 2007) (“Mr. Robbin explained that a monthly retail sale of $60 of pseudoephedrine (Hcl) * * * would be expected to occur less than one in 1,000 times in random sampling, and a monthly retail sale of ... $100 in pseudoephedrine (Hcl) ... would occur about once in a million times in random sampling.”) with H & R Coip., 71 Fed.Reg. 30168, 30169 (May 25, 2006) ("Mr. Robbin ... determined that the normal expected retail sales of pseudoephedrine tablets in a convenience store would ... average ... about $20.00 and the sales of more than $100.00 in a month would be expected to occur in a random sampling about once in one million to the tenth power, a number he characterized as nearly equivalent to the number of atoms in the universe.”). How many jobs have been lost and reputations ruined because of this dubious analysis? One wonders if the DEA feels remorse for preying on companies that lack the wherewithal or sophistication to fight back against sloppy statistics.

. The relevant definition of ''principal” is: "1: most important, consequential, or influ*1174ential: relegating comparable matters, items, or individuals to secondary rank.” Webster's Third New Int’l Dictionary 1802 (1976) (emphasis added). Though the DA elected to not even quote it, much less consider it, 73 Fed. Reg. at 52701, the word "most” is particularly meaningful.

. As the ALJ found, nearly all Novelty’s sales were to "major retail convenience store chain customers” — " 'billion dollar companies' " with "full-time staff” and "legal counsel, to ensure compliance with state and federal regulations.” AU Ruling, slip op. at 6.

. DEA investigators also “visited a number of [Novelty's] customers,” and, after looking at the “logbooks” used to record drug sales, found five stores with questionable patterns (though not all sales were illegal, they nonetheless were "suspicio[us]”). 73 Fed.Reg. at 52694-95. Novelty, however, “cannot review the logbooks.” Id. at 52699 n.43. The DEA, it seems, did not investigate further to see if these drugs were being illicitly used.

. I am confused, however, about the logic in the Raber Letter. Registrants are allowed to keep drugs in the truck "overnight” for "long delivery route[s],” provided the delivery is for "pre-placed customer specific orders for List I chemical products." Raber Letter. But if the concern is "a thief can steal the vehicle’s cargo,” 73 Fed.Reg. at 52698, then why does it matter if the drugs are for "pre-placed customer specific orders” or for on-site orders?

. Novelty, moreover, showed no reluctance to police its own operations. When Store # BPM55, its largest customer, purchased noticeably large quantities-aZihongk not more than federal law allowed -Novelty prohibited the sale of 100-count ephedrine to that store. Perhaps Novelty could have done more, but if a single debatable judgment call about what extralegal steps must be taken to avoid diversion can cost a company its registration, then all the world’s a cage, with every registrant just waiting to be culled.

. Just as "safe” does not mean "risk-free,” Industrial Union Dep’t v. American Petroleum Inst., 448 U.S. 607, 642, 100 S.Ct. 2844, 65 L.Ed.2d 1010 (1980) (Stevens, J., plurality), "maintenance by the applicant of effective controls against diversion” as used in 21 U.S.C. § 823(h) cannot mean perfect controls, an impossible standard.

. In context, I charitably assume the DA’s otherwise staggering suggestion that Novelty’s "failure to enforce its own policies provides reason alone to conclude that it cannot be trusted to adhere to compliance conditions” was not intended to be taken literally, particularly as it was not the first reason given for revocation, 73 Fed.Reg. at 52703-04, and if Novelty’s failure to self-police were intended to be sufficient on its own, then there was no reason for the rest of the DA's lengthy opinion. But if the statement was not hyperbole, then we obviously should vacate this decision because, as explained, the DA entirely failed to consider the relevant denominator. What can be more arbitrary and capricious than that?