Opinion
21-14071 21-14074
08-01-2024
COMPULIFE SOFTWARE, INC, Plaintiff-Appellee, Cross-Appellant, v. MOSES NEWMAN, AARON LEVY, BINYOMIN RUTSTEIN, DAVID RUTSTEIN, Defendants-Appellants, Cross-Appellees. COMPULIFE SOFTWARE, INC, Plaintiff-Appellee, Cross-Appellant, v. BINYOMIN RUTSTEIN, DAVID RUTSTEIN, Defendants-Appellants, Cross-Appellees.
Appeal from the United States District Court for the Southern District of Florida D.C. Docket No. 9:16-cv-81942-BER
Appeal from the United States District Court for the Southern District of Florida D.C. Docket No. 9:16-cv-80808-BER
Before Jordan, Brasher, and Abudu, Circuit Judges.
BRASHER, CIRCUIT JUDGE
We must once again consider an intellectual property dispute between Compulife and its competitors-defendants Moses Newman, Aaron Levy, Binyomin Rutstein, and Binyomin's father, David Rutstein.
Compulife created software to generate life insurance quotes. To create these quotes, the software relied on Compulife's secret database of insurance rates. Compulife accuses the defendants of infringing on its copyright by copying the software's code and using it for their own website. And it says that they stole its trade secret by acquiring portions of the database through improper means. In a previous appeal, we remanded for a trial on Compulife's claims for copyright infringement and misappropriation of trade secrets. See Compulife Software Inc. v. Newman, 959 F.3d 1288 (11th Cir. 2020) (Compulife I).
After a bench trial, the district court ruled against Compulife on the former, but in favor of Compulife on the latter. All parties have appealed again. Their appeals raise three questions.
First, did the district court err in concluding that Compulife's competitors did not infringe on its copyright? We think that by failing to consider the copyrightability of the code's arrangement, the district court erred. And, because of that error, we must reverse and remand for the district court to make new fact findings on the copyright claim.
Second, did the district court err in concluding that the defendants acquired Compulife's trade secret through improper means? We believe that under Compulife I and E. I. duPont deNemours & Co. v. Christopher, 431 F.2d 1012 (5th Cir. 1970), the district court did not err.
Third, did the district court err in holding the defendants jointly and severally liable for misappropriating Compulife's trade secret despite their varying levels of culpability? We believe the district court did not err. Joint and several liability is the standard for trade secret claims, and that sort of liability ignores different degrees of wrongdoing. Accordingly, we affirm the judgment on the trade secret claim.
So we affirm in part, reverse in part, and remand for further proceedings.
I.
Compulife makes life insurance comparison and quotation software. It both licenses the software to customers and has an online version that public users can access to generate quotes.
Compulife's software relies on its proprietary database-a factual compilation of insurance rates used as the raw materials to develop quotes for customers. Some rates are independently available, but the whole compilation is not, and it includes some rates not publicly available. Compulife has developed working relationships with various insurance companies. It obtains rates monthly and often ahead of their public release, making Compulife's database and software especially valuable. As a result, Compulife encrypts its database.
The software works by looking up information in the database and compiling a quote. It has different blocks of code to correspond to different areas of the database. The major components of Compulife's code were arranged as follows: state, birth month, birthday, birth year, sex, smoking status, health classification, insurance type, payment option, sorting output, face amount, and minimum life insurance company rating. For the software to work with the database, it must be arranged in exactly that manner, or the user will get an error.
The code uses some elements that Compulife claims are creative, including: (1) the names it came up with for the various variables throughout the code, (2) radio buttons when making certain input selections, and (3) camel case when writing out the names of variables in the code. The latter two require a brief explanation. Radio buttons are (typically) circular buttons that allow a user to identify single inputs from a defined field of mutually exclusive options. For example, in the picture below, radio buttons constitute the input method on the left, while a dropdown menu is the input method on the right.
(Image Omitted)
Camel case, meanwhile, is a typographical practice when phrases are written without spaces but with capitalization. For example, a variable in Compulife's code collecting birth year information was written in camel case as "BirthYear," instead of in uppercase as "Birth Year," or in lowercase as "birth year."
A group of Compulife's competitors allegedly infringed its copyright and stole its trade secret. David Rutstein was an insurance agent at one time, but he was permanently barred from the profession. He created several websites that used Compulife's software without a license. The sites were registered in his son Binyomin Rutstein's name, and one of them was later owned by Aaron Levy. David got the software by misleading Compulife into thinking that he worked with someone who was licensed to use it. Binyomin allowed his father to conduct these insurance activities with his insurance agent license number. On Levy's and David's demands, an employee, Moses Newman, supervised a scraping attack of Compulife's website to get many millions of quotes generated by its website-far more than a human could ever physically obtain. The men then used those quotes for their own websites. Compulife's sales declined as a result.
Compulife sued these competitors for copyright infringement and misappropriation of trade secrets, among other claims. The parties consented to a magistrate judge and waived their right to a jury trial. The magistrate judge held a bench trial. On appeal from the bench trial, we clarified some standards that apply and directed the district court to make more specific findings. See generally Compulife I, 959 F.3d 1288.
On remand, a new judge conducted a second bench trial but incorporated all our findings as the law of the case. The district court found that most of Compulife's code was not protectable, and the protectable parts were not substantially significant to the defendants' code. The district court thus concluded that Compulife's copyright claim failed. The district court did not consider the arrangement of the code as a whole, though it did examine the arrangement of some of the variables. Separately, the district court concluded that Compulife prevailed on its misappropriation claim because the scraping constituted improper means. Based on those rulings, the district court granted Compulife injunctive relief and entered judgment for $184,225.87 in compensatory damages, relevant prejudgment interest, and $368,451.74 in punitive damages against all defendants jointly and severally.
Both parties appealed. Compulife appeals the judgment against its copyright claim. Meanwhile, the defendants say Compulife should not have won its misappropriation claim, and in the alternative, that they should not have been held jointly and severally liable on that claim.
II.
"On appeal from a bench trial, the district court's conclusions of law are reviewed de novo, but its findings of fact shall not be set aside unless clearly erroneous." Compulife I, 959 F.3d at 1301 (cleaned up). In other words, reviewing a bench trial raises a mixed question of law and fact, and the relevant standard depends "on whether answering it entails primarily legal or factual work." Id. "Separately, when an appellate court discerns that a district court has failed to make a finding because of an erroneous view of the law, the usual rule is that there should be a remand for further proceedings to permit the trial court to make the missing findings." Id. (cleaned up).
III.
We address three main issues in this intellectual property dispute. First, we must determine whether the district court erred in finding that the defendants did not infringe on Compulife's copyright by copying their code. Second, we must determine whether the district court erred in finding that the defendants misappropriated Compulife's trade secret by improper means when they obtained a portion of Compulife's database through scraping. Third, we must determine whether the district court erred in holding the defendants jointly and severally liable. We address each issue in turn.
A.
We will start with Compulife's appeal of the district court's judgment against its copyright infringement claim. To establish copyright infringement, a plaintiff must prove "(1) ownership of a valid copyright[] and (2) copying of constituent elements of the work that are original." Bateman v. Mnemonics, Inc., 79 F.3d 1532, 1541 (11th Cir. 1996) (internal quotation marks omitted) (quoting Feist Publ'ns, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, 361 (1991)).
The first element is easy. We earlier recognized that the existence and validity of Compulife's copyright is undisputed. Compulife I, 959 F.3d at 1301. "Under the law-of-the-case doctrine, [the resolution of] an issue decided at one stage of a case is binding at later stages of the same case." Schiavo ex rel. Schindler v. Schiavo, 403 F.3d 1289, 1291 (11th Cir. 2005) (alteration in original) (quoting Toole v. Baxter Healthcare Corp., 235 F.3d 1307, 1313 (11th Cir. 2000)). So we can proceed to the second copyright infringement element-copying.
The copying element "comprises two subparts, 'factual and legal copying,' both of which Compulife, as the plaintiff, has the burden to prove." Compulife I, 959 F.3d at 1301. The factual copying inquiry is "whether the defendant, as a factual matter, copied portions of the plaintiff's program." MiTek Holdings, Inc. v. Arce Eng'g Co., 89 F.3d 1548, 1554 (11th Cir. 1996) (quoting Gates Rubber Co. v. Bando Chem. Indus., Ltd., 9 F.3d 823, 832 (10th Cir. 1993)). Again, we already determined that Compulife established factual copying. Compulife I, 959 F.3d at 1302.
Compulife's appeal turns on the element of legal copying. As relevant here, "legal-or actionable-copying occurs when those elements of the copyrighted work that have been copied are protected expression and of such importance to the copied work that the appropriation is actionable." Id. (cleaned up). The district court found no legal copying, but it came to that conclusion after slightly erring in applying our test.
We have adopted a three-step test for legal copying: the abstraction-filtration-comparison test. First, a court should "abstract"-"break down the allegedly infringed program into its constituent structural parts." Id. at 1303 (cleaned up). Second, a court should filter-"sift out all non-protectable material." Id. (cleaned up). Finally, the court should compare the protected material with the copycat. When considering the literal elements of a program, like source code and object code, the plaintiff need establish only a "substantial similarity" between the two works. Id. at 1302 n.6. As we held in Compulife I, because the dispute here concerns source code, the substantial similarity standard applies. Id.
Compulife argues that the district court erred at each step of this three-part test; the defendants argue that it did not. We agree in part with Compulife and in part with the defendants.
1.
Turning first to "abstraction," we must "break[] down the allegedly infringed program into its constituent structural parts." Id. at 1303. Compulife argues that the arrangement of its various components of source code-state, birth month, birthday, birth year, sex, smoking status, health classification, insurance type, payment option, sorting-is creative and therefore protectable. That is, Compulife argues that one "constituent" part of its program is the arrangement of its source code.
Just as we have held that the arrangement of yacht listings within a used boat guide could be protectable, see BUC Int'l Corp. v. Int'l Yacht Council Ltd., 489 F.3d 1129, 1145 (11th Cir. 2007), Compulife argues that the arrangement of the variables in its code could be protectable. And, generally, we have held that the arrangement of elements in a program may be protectable. See MiTek Holdings, Inc., 89 F.3d at 1558 (recognizing that the arrangement of elements of a user interface may be protectable as a factual compilation); see also Oracle Am., Inc. v. Google Inc., 750 F.3d 1339, 1356 (Fed. Cir. 2014) (holding that the sequence, structure, and organization of various packages in a software are protectable).
A careful reader might argue that this caselaw is distinguishable. Our decision in MiTek addressed the potential protectability of the arrangement of nonliteral elements of a program, like a user interface. Here, however, we consider the arrangement of a literal element-source code. So one might conclude that, unlike the arrangement of a nonliteral element, the arrangement of a literal element is not protectable.
But we think there is no principled distinction to draw between those two contexts. We explained in Compulife I that the virtual-identicality standard governs the arrangement of nonliteral elements of a program, even though the substantial-similarity standard is the default standard for other cases. See 959 F.3d at 1302 n.6. The substantial-similarity standard asks, unsurprisingly, whether there is a "'substantial similarity' between the allegedly offending program and the protectable, original elements of the copyrighted works." Id. at 1302. The virtual-identicality standard, on the other hand, asks whether the two works are virtually identical, which is "a level of similarity greater than the 'substantial similarity' standard." MiTek, 89 F.3d at 1558 n.24. This difference, though, concerns only the analysis that a court engages in when comparing the copied and copying works.
But we're not comparing, we're abstracting. There is no other meaningful difference between the virtual-identicality standard and the substantial-similarity standard to make us think that, when abstracting, courts ought not consider the arrangement of a program. In fact, we have applied the latter standard to the arrangement of information in other contexts multiple times. See e.g., BUC Int'l Corp., 489 F.3d at 1149 n.42 (explaining that BellSouth Advert. & Pub. Corp. v. Donnelley Info. Pub., Inc., 999 F.2d 1436 (11th Cir. 1993) (en banc), "established the 'substantial similarity' standard as the default mode of analysis for compilation copyright claims"). Our conclusion in MiTek compels courts, when abstracting, to assess the arrangement of literal elements as a potentially protected constituent part of the program.
Although the district court considered the selection and arrangement of Compulife's code to some degree, the district court never identified the entire arrangement of these variables in the code as a constituent component of the code. For example, the district court expressly evaluated the arrangement of the birth month, birthday, and birth year variables before filtering. But it didn't look at the arrangement of all the variables together. And, relying on Feist Publications, Inc. v. Rural Telephone Service Co., 499 U.S. 340, 347-48 (1991), the district court recognized that factual compilations, "like the ones performed by Compulife's software in compiling facts . . . to generate a quote" can be protectable. But the arrangement of the code itself can be protectable, not just the results produced by the software.
Given that the arrangement of the code may be protectable, we agree with Compulife that the district court should have abstracted the "arrangement" as something to be analyzed at the subsequent filtration step. The defendants offer two arguments to the contrary, but neither works.
To start, the defendants argue that we should not consider whether the arrangement of code is protectable because Compulife forfeited this argument below. After all, typically, we do not consider issues for the first time on appeal. See Access Now, Inc. v. Sw. Airlines Co., 385 F.3d 1324, 1331 (11th Cir. 2004). But Compulife raised this issue below. In its proposed findings of fact and conclusions of law before the district court, Compulife said that its "code contained numerous creative elements including a creative way to identify and organize variables related to requests for insurance quotation information."
On the merits, the defendants argue that the abstraction-filtration-comparison test forecloses analyzing the arrangement of the entire code because abstraction requires the court to break down the code into its constituent structural components. But by recognizing that the arrangement of elements of a program can be protectable, precedents such as MiTek and Oracle suggest that the arrangement is a constituent structural component suitable to be analyzed in the abstraction and filtration steps.
We draw no conclusion about whether the arrangement of this code is protectable. We simply hold that the district court erred by failing to consider whether it was.
2.
Although we must remand for the district court to assess whether the arrangement of the code is protectable, we will continue with Compulife's arguments about the district court's application of other elements of the test. See United States v. White, 846 F.2d 678, 688 (11th Cir. 1988) (explaining that we can address additional alleged errors to provide guidance to the district court in further proceedings).
We will start with filtration. "Before comparing two works to determine if they display the required substantial similarity, a court must 'eliminate from comparison the unprotectable elements of' the copyrighted work." Compulife I, 959 F.3d at 1303 (quoting Bateman, 79 F.3d at 1545). Beyond the arrangement of the variables, which we leave for the district court to consider on remand, Compulife argues that the district court erroneously filtered out other protectable content: (1) the chosen name of the variable that collects birth years (i.e., using "BirthYear" instead of another option like "year"); (2) its use of camel case when naming variables; and (3) the use of radio buttons for making certain selections such as smoker status or gender, rather than other options like a dropdown menu, check box, or manual entry in a text field.
We disagree with Compulife. The district court was right to conclude that those elements are not protectable.
We'll consider first the name of the BirthYear variable. Compulife argues that its competitors could have chosen any other name for the variable. But "obvious label[s] . . . lack the requisite originality for copyright protection." BellSouth, 999 F.2d at 1444 (holding that a heading of "Attorneys" for a list of attorneys or "Banks" for a list of banks is not original expression and therefore not protectable). Common sense instructs that the name "BirthYear" for a variable that invites users to input their birth year is too obvious to be original.
Now we'll turn to Compulife's choice to use camel case there and elsewhere. Under Compulife I, expression dictated by external factors like industry standards is not protectable. See 959 F.3d at 1304-05. There was testimony at trial that camel case is an industry standard. It is perfectly reasonable for the district court to have found that this industry standard dictated Compulife's choice of camel case for "BirthYear" and other variables. So the district court did not err in filtering these elements out.
Finally, we'll address Compulife's use of radio buttons for certain user selections. "Under the merger doctrine, 'expression is not protected in those instances where there is only one or so few ways of expressing an idea that protection of the expression would effectively accord protection to the idea itself.'" BellSouth, 999 F.2d at 1442 (quoting Kregos v. Associated Press, 937 F.2d 700, 705 (2d Cir. 1991)). For example, because there are so "few ways of visually presenting the idea that an activity is not permitted," an expression of that idea in the form of "a circle with a diagonal line crossed through it" is not protectable under the merger doctrine. See BUC Int'l Corp., 489 F.3d at 1143.
Likewise, the merger doctrine makes Compulife's use of radio buttons unprotectable. There are very few practicable ways to invite users to make single selections among a set of defined options. A few of Compulife's proposed alternatives don't fit the bill. Checkboxes generally allow you to make multiple selections simultaneously. Open text fields don't limit you to a defined set of options. Only radio buttons and dropdown menus remain. With so few options, allowing Compulife to copyright its use of radio buttons would merge the expression (the use of radio buttons) with the idea (making single selections from a set of defined options). That is contrary to our law and, at times, Compulife's own opinion. After all, Compulife inconsistently concedes that its use of radio buttons was not protectable, saying "[n]or did Compulife claim that its use of a radio button was copyrightable," but later arguing the opposite.
For all these reasons, the district court did not err in filtering out all the other elements besides the arrangement of the code.
3.
Finally, let us turn to comparison. "The last step is to compare any remaining kernels of creative expression with the allegedly infringing program to determine if there is in fact a substantial similarity-comparison." Compulife I, 959 F.3d at 1303 (cleaned up). Under this step, "both the quantitative and the qualitative significance of the amount copied" must be assessed. Peter Letterese & Assocs., Inc. v. World Inst. of Scientology Enters., 533 F.3d 1287, 1307 (11th Cir. 2008).
Compulife argues its competitors' copying was both quantitatively and qualitatively significant. It complains that the defendants copied 84 percent of its code. And it says the copying was qualitatively similar, arguing that the defendants copied the way it named and formatted its variables. Again, we leave Compulife's arguments about the arrangement of the code for the district court to consider on remand. But we disagree with Compulife that there was significant copying of the remaining elements of the code.
Start with quantitative significance. Yes, Compulife's competitors copied 84 percent of its code. But "even substantial similarity between a copyrighted work's unprotectable elements and a purportedly infringing work isn't actionable, regardless of how many unprotectable elements are copied or how important they may be." Compulife I, 959 F.3d at 1303. Of the elements the district court did find protectable, Compulife's competitors copied only 27 of the 347 lines. That's not the 84 percent that Compulife alleges, but a mere 7.78 percent.
Qualitative significance protects against even quantitatively insignificant copying of a computer program, when that small, copied portion gives the program "distinctive features" or makes "the program especially creative or desirable." Id. at 1302 (quoting 4 Nimmer on Copyright § 13.03 (2019)). Qualitative significance is used to protect quantitatively insubstantial but otherwise incredibly significant copying-like copying "the most powerful passages" from President Ford's autobiography, concerning his pardon of President Nixon (one of the most significant moments of his presidency). Harper &Row Publishers, Inc. v. Nation Enters., 471 U.S. 539, 565 (1985).
Compulife does not offer any reason why the remaining elements are qualitatively significant. It argues that these elements were necessary to perform "key functions" of the program. But their importance is not the key to their protectability. Nothing in the record suggests the district court erred in finding the remaining elements not distinctive or especially creative.
B.
Having addressed copyright infringement, we will move on to Compulife's trade secret claim. Compulife says that the defendants stole its database of insurance rates, which it argues is a trade secret. A trade secret is a piece of information that has independent economic value from being kept secret and is in fact subject to reasonable efforts to maintain its secrecy. See Compulife I, 959 F.3d at 1311. A plaintiff in Florida can sue a defendant who steals his trade secret under both federal and Florida law.
Compulife brings claims under the federal Defend Trade Secrets Act and the Florida Uniform Trade Secrets Act. We already held, and the parties seem to agree, that the federal law is so much like the Florida law that an analysis of a Florida claim amounts to an analysis of a federal one. Compulife I, 959 F.3d at 1311 n.13. So like we did in Compulife I, we will proceed under the Florida law.
To prove that the defendants stole its trade secret, Compulife "must demonstrate that (1) it possessed a trade secret and (2) the secret was misappropriated." Yellowfin Yachts, Inc. v. Barker Boatworks, LLC, 898 F.3d 1279, 1297 (11th Cir. 2018) (quotation and internal quotation marks omitted). First, we will consider whether Compulife's database constituted a trade secret, and second, we will consider whether the defendants misappropriated it.
1.
We concluded previously that it was not clearly erroneous for the district court to find that Compulife's database was a trade secret. Compulife I, 959 F.3d at 1311. On remand, the district court reasoned that it must respect that finding as the law of the case. The law-of-the-case doctrine has three exceptions: when "(1) a subsequent trial produces substantially different evidence, (2) controlling authority has since made a contrary decision of law applicable to that issue, or (3) the prior decision was clearly erroneous and would work manifest injustice." Wheeler v. City of Pleasant Grove, 746 F.2d 1437, 1440 (11th Cir. 1984) (quoting United States v. Robinson, 690 F.2d 869, 872 (11th Cir. 1982)).
The defendants argue that the first exception to the law-of-the-case doctrine applies here. After we remanded the case in Compulife I, the district court held a new trial. The defendants say that agree, that the federal law is so much like the Florida law that an analysis of a Florida claim amounts to an analysis of a federal one. Compulife I, 959 F.3d at 1311 n.13. So like we did in Compulife I, we will proceed under the Florida law. substantially different evidence emerged that suggests that the database was not subject to reasonable efforts to maintain its secrecy. Specifically, the defendants point to new evidence about how Compulife protected its software. None of this evidence undermines our assessment in the initial appeal.
First, contrary to what was originally known, the licensee through which David Rutstein accessed the database was not required by Compulife to sign a license until 2015, long after the misappropriation happened. But there was also testimony from the licensee that he had always understood that his company was bound by a license when using the Compulife software, even before he formally signed one; he said that he treated the software the same way he treated his company's own trade secrets. None of this new information undermines our conclusion on the first appeal that Compulife goes to great lengths to control the use of its software through licenses.
Second, David Rutstein separately accessed the database by sending a misleading email to Compulife, which gave him access without verifying his statements. But David Rutstein lied. He said he had an account with Compulife through a real Compulife licensee it recognized and asked for help in using the service on the licensee's behalf. Perhaps Compulife should have had a second-factor authentication, but it is not unreasonable for it to believe that someone asking for specialized help in using its service on a licensee's behalf works for that licensee. At the very least, it does not negate all the other work Compulife does in encrypting, licensing, and otherwise securing its database. See Compulife I, 959 F.3d at 1312 ("So long as the precautions taken were reasonable, it doesn't matter that the defendant found a way to circumvent them. Indeed, even if the trade-secret owner took no measures to protect its secret from a certain type of reconnaissance, that method may still constitute improper means.").
Third, the public has always been able to pull insurance quotes from the database through the website, with no limitations on how many quotes they could pull. But we previously mentioned that even if individual quotes that are publicly available lack trade secret status, the whole compilation of them (which would be nearly impossible for a human to obtain through the website without scraping) can still be a trade secret. See id. at 1314.
One final point on the law of the case: the defendants also say, with no argument, that we should conclude that our previous decision was clearly erroneous, such that following it would work a manifest injustice. Because the defendants make merely a passing reference to this exception and offer no argument about why it should apply, we consider this issue abandoned. See Sapuppo v. Allstate Floridian Ins. Co., 739 F.3d 678, 681 (11th Cir. 2014).
So the law-of-the-case doctrine applies, and the database is a trade secret.
2.
Now, let us consider whether the defendants misappropriated Compulife's trade secret. "One party can misappropriate another's trade secret by either acquisition, disclosure, or use." Compulife I, 959 F.3d at 1311 (citing Fla. Stat. § 688.002(2)). "Compulife alleges misappropriation both by acquisition and by use- but not by improper disclosure." Id.
Misappropriation by acquisition occurs when a person acquires a trade secret and knows or has reason to know that it was acquired by improper means. Id. Misappropriation by use occurs when a person uses a trade secret without consent and either: (1) used improper means to acquire the trade secret; (2) at the time of use knew or had reason to know that it was (a) derived from a person who used improper means, (b) acquired in a manner giving rise to a duty to maintain secrecy, or (c) derived from a person who owed a duty to maintain secrecy to the owner; or (3) before a material change in his position, knew or had reason to know that it was a trade secret and had been acquired by accident or mistake. Id.
Two key questions are at issue. First, did the defendants use improper means? And second, did the defendants acquire the trade secret? The district court answered "yes" to both. And we cannot say the district court erred.
Let's start with improper means. As we previously explained, "[a]ctions may be 'improper' for trade-secret purposes even if not independently unlawful." Id. at 1312. (citing Christopher, 431 F.2d at 1014). "Under the broad definition adopted in [our precedent] Christopher, misappropriation occurs whenever a defendant acquires the secret from its owner 'without his permission at a time when he is taking reasonable precautions to maintain its secrecy.'" Id.
Here, Compulife alleges that the defendants used "scraping" to acquire its trade secret-i.e., its database of insurance quotes. It is important to note that scraping and related technologies (like crawling) may be perfectly legitimate. Much of the modern internet is built on those technologies. See Han-Wei Lui, Two Decades of Laws and Practice Around Screen Scraping in the Common Law World and Its Open Banking Watershed Moment, 30 WASH. INT'L L.J. 28, 29 (2020) (explaining that scraping "is now used for targeted advertising, price aggregation, budgeting apps, website preservation, academic research, journalism, and more") (citations omitted). But the defendants in this case did not take innocent screenshots of a publicly available site; instead, they copied the order of Compulife's copyrighted code and used that code to commit a scraping attack that acquired millions of variable-dependent insurance quotes. If they had not formatted and ordered their code exactly as Compulife did, they would not have been able to get the millions of quotes that they got. As we explained in the previous appeal, this deceptive behavior resembles the acquisition of a trade secret through surreptitious aerial photography, which we addressed in Christopher.
Now let's consider acquisition. The district court found that the defendants stole anywhere from 3 million to 43.5 million quotes. It is true that we don't know the total percentage of the database that was taken. But the district court found that Compulife's revenue declined after the scraping attack, that it lost business it otherwise expected to receive, that the number of customers looking for free trials declined, that the number of free trials that converted to four-month subscriptions declined, and that the number of four-month subscriptions that converted to annual subscriptions declined. All the while, quotes generated from Com-pulife's software continued to appear on the defendants' website. A fair reading of this record supports the district court's finding that the defendants obtained so much of the database that they posed a competitive threat to Compulife. Faced with this record, the district court did not err in concluding that Compulife's trade secret was acquired by its competitors.
C.
Now let us consider the joint and several liability of the defendants for Compulife's claim of misappropriation of trade secrets. The defendants think it unwarranted because they each argue that they are less at fault than others. Compulife thinks it appropriate.
We start with a few basic principles. First, Florida is a comparative fault regime with an exception for intentional torts. See Barton Protective Servs., Inc. v. Faber, 745 So.2d 968, 976 (Fla. Dist. Ct. App. 1999). Second, in Florida, trade secret misappropriation is an intentional tort. See Vance v. Tire Eng'g & Distrib., LLC, 32 So.3d 774, 776 (Fla. Dist. Ct. App. 2010). Third, joint and several liability is applied "when the defendants acted in concert" and their "separate independent" tortious acts "combined[] to produce a single injury." Basel v. McFarland &Sons, Inc., 815 So.2d 687, 691 (Fla. Dist. Ct. App. 2002) (quoting Smith v. Dep't of Ins., 507 So.2d 1080, 1091 (Fla. 1987)). The upshot is that defendants in trade secret misappropriation cases are jointly and severally liable under Florida law when their conduct produces a single injury.
The doctrine of joint and several liability does not require equal fault among all defendants. "Joint and several liability was established through the common law and later codified by the legislature. It allows a claimant to recover all damages from one of multiple defendants even though that particular defendant may be the least responsible defendant in the cause." Agency for Health Care Admin. V. Associated Indus. Of Fla., Inc., 678 So.2d 1239, 1257 (Fla. 1996). Joint and several liability's whole purpose is to treat the defendants as equally responsible, regardless of their fault. It must hold true when applied to trade secret claims. At least one of our sister circuits agrees that joint and several liability applies to trade secrets cases and copyright infringement cases like this one. Salton, Inc. v. Philips Domestic Appliances & Pers. Care B.V., 391 F.3d 871, 877 (7th Cir. 2004). To hold otherwise, we would need to conclude that comparative fault, rather than joint and several liability, is the default regime in trade secret cases.
Applying those principles here supports joint and several liability. There are enough facts in the record to support the district court's finding that each of the defendants acted in concert to commit the wrongs causing Compulife's injury: David Rutstein lied in an email to get Compulife's web quoter put onto his website; Newman supervised the scraping attack at David's and Levy's direction; Binyomin was the investor and owner of the company that his father David used to design the website (the infringing website was registered in his name), and Binyomin gave his father permission to use his insurance license and his agency to collect fees from insurance sales generated by the stolen trade secret. These acts caused a single injury-lost sales and revenues-and, under Florida law, the defendants are jointly and severally liable for that single injury.
The defendants offer no argument for why we may or should create an exception to the typical rules of joint and several liability in the trade secret context. The defendants say that several courts have suggested that joint and several liability is appropriate when the degree of wrong is the same among the several defendants. See Brocade Commc'ns Sys., Inc. v. A10 Networks, Inc., 873 F.Supp.2d 1192, 1218 (N.D. Cal. 2012); Fishkin v. Susquehanna Partners, G.P., No. CIV.A.03-3766, 2007 WL 853769, at *3 (E.D. Pa. Mar. 19, 2007). But they are not right that multiple defendants must have the same degree of wrong to apply joint and several liability. To draw that conclusion would be "to commit the fallacy of the inverse (otherwise known as denying the antecedent): the incorrect assumption that if P implies Q, then not-P implies not-Q." N.L.R.B. v. Noel Canning, 573 U.S. 513, 589 (2014) (Scalia, J., concurring in the judgment).
In fact, the very authorities the defendants cite suggest that joint and several liability is appropriate in this circumstance. The courts in both Brocade and Fishkin relied on a leading treatise, Milgrim on Trade Secrets. It suggests that the general rule for misappropriation of trade secrets is joint and several liability. See Brocade, 873 F.Supp.2d at 1218 (citing Milgrim for the proposition that "[c]ommentators on trade secret law also suggest that, in general, liability for misappropriation claims is joint and several" before quoting the same degree of wrong language); Fishkin, 2007 WL 853769, at *3 (citing Milgrim for the identical point). The treatise proposes a narrow exception to this general rule for a hypothetical corporate employe who is independent of his corporate employer. 4 Milgrim on Trade Secrets § 15.02 (2023). The treatise hypothesizes that in some cases a court could split a single loss into multiple losses across time, such that if a corporate employee quits his job "at some time during the period of wrongful use, as to him damages might only be computed for wrongful use prior to such termination." Id. The narrowness of that hypothetical exception-which is not present on the facts here-emphasizes the breadth of the presumption in favor of joint and several liability.
Binyomin Rutstein separately suggests that his culpability was so low that he should not have been found liable at all. But we cannot say the district court clearly erred. The district court found that David Rutstein's infringing websites were registered in Binyomin's name, Binyomin owned an insurance agency which benefited from the stolen database, and Binyomin allowed David to use his license in violation of a consent decree to illicitly engage in the insurance business. Nothing in the record suggests that those findings are clearly erroneous. And those findings support the district court's conclusion that Binyomin participated in and benefited from the misappropriation of Compulife's trade secret.
We will mention one final issue about joint and several liability. One could argue that, even though joint and several liability is the rule for compensatory damages in a case like this one, the district court should have considered the defendants' different levels of culpability in the context of punitive damages. The general principle is that "exemplary damages imposed on a defendant should reflect 'the enormity of his offense.'" BMW of N. Am., Inc. v. Gore, 517 U.S. 559, 575 (1996) (quoting Day v. Woodworth, 54 U.S. 363, 371 (1852)). And several of our sister circuits have recognized that, under certain state laws, joint and several liability is inappropriate for punitive damages. See, e.g., Rodick v. City of Schenectady, 1 F.3d 1341, 1349 (2d Cir. 1993); Bader Farms, Inc. v. BASF Corp., 39 F.4th 954, 973 (8th Cir. 2022). For their part, however, the Florida courts at least occasionally approve of joint and several liability for punitive damages. See Parton v. Palomino Lakes Prop. Owners Ass'n, Inc., 928 So.2d 449, 454 (Fla. Dist. Ct. App. 2006); Wrains v. Rose, 175 So.2d 75, 79 (Fla. Dist. Ct. App. 1965). Whether the district court should have considered imposing separate punitive damages is a complicated issue, which would have benefited from the parties raising it. Nobody did. So the issue was forfeited. See United States v. Campbell, 26 F.4th 860, 873 (11th Cir. 2022). We therefore do not address it.
IV.
The district court is AFFIRMED in part and REVERSED in part. The matter is REMANDED for further proceedings consistent with this opinion.