Keegan Skeate was working the night shift when he first heard about the scam. The 26-year-old was only a few months into his new job at Praxis Laboratory, a Washington state lab that conducts consumer safety tests and THC potency analysis for legal cannabis products. During one quiet shift in 2018, a fellow lab technician walked over to Skeate and told him that she thought someone was manipulating her test results. She said that the numbers in her spreadsheets looked irregular and the recorded THC potency values were higher than she remembered measuring.
Over multiple interviews with FiveThirtyEight about his time at Praxis, Skeate explained what happened next. When he reexamined his own test results, he saw a similar pattern.
“I honestly believed her and did think the numbers were changed,” Skeate recently said. “It was a moral dilemma where I really didn’t know what to do. Do you tell someone? Do you quit, or do you just keep working? I kept working and I don’t really think that was the right thing to do.”
That was the spring of 2018. By the end of 2020, Skeate would lose his job, the local police would recommend criminal charges against him for blocking his boss from a company database, and the young software developer would become the latest whistleblower tangled in one of the legal cannabis industry’s thorniest problems: how to stop corrupt pot labs from profiting off manipulated test results.
Confidence Interval: Will marijuana be legal across US by 2024? | FiveThirtyEight
America’s legal weed industry sold over $17 billion of pot last year and the industry’s obsession with THC, pot’s most famous intoxicant, has created financial rewards for every marginal increase in THC potency. Weed shoppers use THC percentages like nutritional labels, purchasing products based on their THC content, yet the lab system entrusted with measuring the compound is vulnerable to corruption. Laboratories across the country have been suspended or fined for manipulating potency results, having deficient procedures for detecting contaminants like mold or faking those contaminant tests altogether.
“THC inflation is pernicious, it’s easy to accomplish and there are strong financial incentives to do it,” said Don Land, a professor of chemistry and forensics at the University of California, Davis, and an adviser to the multistate cannabis lab company Steep Hill. “I believe that it is likely to happen at least infrequently in every market out there, and there’s very little chance that labs would get caught.”
related: Americans From Both Parties Want Weed To Be Legal. Why Doesn’t The Federal Government Agree? Read more. »
Corrupt labs are cheating customers out of the pot they think they’re buying, and the industry’s potential problems with lab fraud are unlikely to go away as more states legalize weed and regulations grow increasingly dependent on measuring THC. New York plans to tax cannabis by THC content when commercial sales begin in the next few years. And there’s a growing nationwide movement to limit the THC content of legal products. But how can you collect taxes or enforce laws based on a product’s THC potency if you can’t trust how much THC is actually there?
OBSESSED WITH THC
Pot advocates have long argued that legalizing weed would improve public health by putting consumer safety regulations on America’s favorite illicit substance. Underground markets are notoriously bad at protecting consumers, whether it’s tainted moonshine killing people in the 1920s or illegal THC vape cartridges killing people in 2019, so establishing safety standards should, the argument goes, make society safer.
In America’s booming legal weed industry, state governments have delegated their regulatory eyes to private, for-profit cannabis labs. Pot farmers and processors must pay private, state-certified labs to conduct batch-specific tests before any product can be sold to consumers. The samples must pass a quality assurance check — ensuring they aren’t contaminated with things like E. coli or mold — and be tested for their cannabinoid potency, usually by measuring the presence of pot’s two most common active compounds: cannabidiol (CBD) and tetrahydrocannabinol (THC).
In theory, this testing system should be more robust than other consumer products, like produce or alcohol, which are usually only tested when there is a suspected safety issue. But in practice, some labs across the country have been accused of fraud. Labs have been caught inflating THC potency levels, passing moldy cannabis as safe, and even making up results entirely. A blistering 2019 audit of Oregon’s testing system found that the state’s testing program “cannot ensure that test results are reliable and products are safe” and said the state regulatory program had “limited authority, inadequate staffing and inefficient processes.”
Experts say that lab corruption is widespread because the incentives to cheat are too high and enforcement is mostly ineffective. “The people paying for the tests are happier to get inaccurate tests that say [their products] are more potent than they really are,” Land said.
These lab tests have serious financial impacts on America’s multi-billion-dollar legal cannabis market. If a sample fails its quality assurance test, a farmer could have to destroy an entire crop of cannabis. And THC potencies have become a benchmark for both wholesale and retail cannabis prices for cured flower products, the common form of cannabis used for smoking. A 2018 study of 31 million cannabis transactions in Washington state between 2014 and 2016 found that listed THC potency had a statistically significant relationship with price, with a 1 percentage point increase in THC potency associated with a 1 to 2 percentage point increase in price per gram. Interestingly, the study found that the state’s market was dominated by pot that had tested over 15 percent potency, with only 7.5 percent of all pot sold during the study period testing below that figure. California’s market has a very similar dynamic, with over 95 percent of all cured flower products sold in 2020 testing at above 14 percent THC, according to Flowhub, a cannabis software company. California’s retail prices were also correlated with THC potency, with low-potency pot (between 7 and 14 percent THC) selling for an average of $5.31 per gram while pot that tested higher than 21 percent THC sold for more than $11 per gram, on average.
More potent pot tends to be priced higher in California
Market share and price of legal marijuana by listed level of THC, 2020
|THC Level||share of transactions||retail price/gram|
For Julia Jacobson, the CEO of Northern California’s Aster Farms, there’s no question that THC defines the terms of pot commerce in California.
“The pressure is real. Full stop. We have some retailers who love us, who sell out of our products, and they will only put our product on their shelves when it tests over 20 percent,” Jacobson said. “The buyers are always caveating, saying, ‘We know there’s so much more to cannabis and its effects [than just THC], but our consumers are still THC hunting.’”
Jacobson said her company doesn’t even try to sell batches that test under 15 percent THC, and instead uses those batches as promotional samples or turns them into oil for edibles or vape cartridges.
The market’s obsession with marginally higher THC potencies does not appear to be based on any kind of science or expert advice. A University of Colorado Boulder study published last year did not find a correlation between higher THC products and stronger subjective “high” effects. Cannabis aficionados have long claimed that THC potency is a poor measure of cannabis quality, equating shopping for pot based on THC potency with shopping for wine or beer based only on alcohol content.
“There’s a sungrown brand and they had a batch that was 12 percent THC but … that weed just did it for me,” said Andrew DeAngelo, a widely respected cannabis advocate and co-founder of Oakland’s Harborside Dispensary. “I don’t want to name the brand because I don’t want people to think it’s a low-potency brand, it’s not.”
A VULNERABLE SYSTEM
Allegations of lab fraud have dogged the legal cannabis industry from the beginning of commercial pot sales, especially in Washington state, where recreational pot has been legally sold since 2014. Observers have known about the problem in part because lab data is public in Washington, allowing data scientists to analyze the test results of individual labs.
One of those scientists is Jim MacRae, a private consultant to pot companies based in the Seattle suburbs. Verbose, profane and tinged with a Canadian accent, MacRae has been a thorn in the side of Washington’s lab industry since 2015, when the state first started releasing its traceability database, which tracks products as they are grown, tested and sold. MacRae has a Ph.D. in experimental psychology and spent over a decade as a data scientist for the pharmaceutical company Merck. He uses the state’s pot data to identify problematic behavior by lab operators, similar to how the Department of Justice uses health care prescription data to spot doctors or pharmacists who are overprescribing opioids. MacRae publishes his analysis in meandering blog posts that are punctuated by damning analysis of how the testing industry’s perverse incentives affect the entire market.
In late 2015, MacRae published data showing that, in a three-month period, four of Washington state’s 14 certified labs had tested tens of thousands of samples without ever failing a sample for microbial contamination, while other labs failed as many as 45 percent of samples. Four months later, the state suspended one of those four labs for six months, finding the lab had given the highest THC averages in the state and “put the public health and safety at risk by exposing the public to … marijuana products that have not been properly or accurately tested for microbial contamination and other risks.”
MacRae’s analysis has also shown how the pot market rewards friendly labs (the labs which give high THC potency results and fail fewer samples) with millions of dollars in lab tests. In March of 2017, he published data showing that one lab had increased its market share of cannabis flower testing from 0 percent to around 20 percent within just 12 months and was giving out the state’s highest average THC potencies. Five months later, Washington temporarily suspended the lab after it found irregular testing results and bad lab practices. The state’s auditor did not find evidence of widespread THC inflation at the lab, but did uncover a sample of Blue Dream cannabis that was tested at 37.2 percent THC potency despite the lab having no paper trail for how it measured such an incredible potency figure. (Marijuana’s maximum potency is usually theorized at around 35 percent THC.) The auditor’s report criticized the lab for not investigating further.
Other states have similar issues. In 2019, MacRae analyzed the cannabis market in Nevada, which has publicly struggled to regulate their industry. (In one five-month period, the state suspended the licenses of nearly half of all of its certified labs.) MacRae’s analysis found that the average THC potency in the state had steadily increased from 19 percent to almost 22 percent between 2018 and 2019, and multiple labs were appearing to release fraudulently inflated THC potencies and rarely failing samples for safety standards. Within days of MacRae presenting his findings to the state, Nevada’s government warned the lab industry that they were actively investigating THC inflation, and within weeks the state had fined a lab for “unsound testing practices.”
Ensuring accurate tests is made more difficult by the fact that pot is essentially a vegetable, which means there’s natural variation between different samples taken from the same plant. Michigan, which sold over $500 million of legal pot last year, has attempted to reduce the potential sampling error by mandating that testing labs collect samples themselves with a standardized sampling process, rather than allowing farms to send in their own cherry-picked samples. Lev Spivak-Birndorf, co-owner of state-certified PSI Labs in Michigan, said even with a required sampling process, he has still seen evidence of THC inflation in Michigan’s market.
“I think it’s challenging for labs in this industry because with low regulatory oversight, there’s almost an incentive to have inaccurate data,” Spivak-Birndorf said. “I have seen labels posted with 106 percent THCA. I don’t think you really need to be a scientist to know that some level of inflation is happening on a label like that.”
Keegan Skeate couldn’t forget about his coworker’s allegations of fraud at Praxis Lab. It was nauseating for him to think that his lab work was defrauding customers in the legal pot market, deceiving them into buying pot that had less THC than its label stated. But he also needed the job. He had moved from out of state to work at the lab and he was dealing with $50,000 in student loan debt after recently dropping out of a Ph.D. program at the University of California, Irvine. Other than his suspicions, things at the lab were going well. He had quickly been promoted from a laboratory analyst to a software developer, and he built the company’s website and a fully functioning information management system.
His software included a version control system, which tracked every granular change to the company’s data workbooks and would, by the summer of 2020, confirm to him what he and his coworker had already suspected: The lab’s scientific director and owner, Dustin Newman, was falsifying THC potency tests on a large scale.
“I saw that the first person signed off on one set of results and then Dustin signed off on a different set of results, and that was just crystal clear proof,” Skeate said. “I tallied it all up and it was hundreds of samples.”
Skeate’s version control showed that over a three-month period in the summer of 2020, Newman had altered test results to boost the THC potencies for over 1,200 samples. The Washington State Liquor and Cannabis Board (WSLCB) officer reviewing Skeate’s whistleblower complaint wrote in his investigation file that the data showed “solid proof of tampering” because the weights showed inconsistent decimal points, varying between two, three, four, or five decimal points, even though the lab’s scales measured to four decimal points.
On September 3, 2020, two days after Skeate sent his evidence of widespread fraud, the WSLCB investigators showed up at Praxis Laboratory in Centralia, a small town about 80 miles south of Seattle. They confirmed that the lab’s internal documents matched what Skeate had secretly sent the agency and then confronted Newman about the modified numbers. According to the state investigation file, Newman denied inflating THC potencies but didn’t have an explanation for the irregularities. In an April interview with FiveThirtyEight, Newman blamed the irregularities on Skeate’s software and claimed the WSLCB failed to actually conduct an investigation.
Five days after the WSLCB’s visit to his lab, Newman emailed the state and said he would “no longer have any role in the operation, or management of the company.” Meanwhile, Skeate continued working, apparently without the lab’s management realizing he was the whistleblower.
“I really should have just quit at that point, but I thought it was my responsibility to keep coming to work,” Skeate said. “And then over those weeks Dustin realized — I can’t believe he didn’t on day one — but he gradually … connected the dots and realized I must be the whistleblower. And then I essentially was fired and it turned into this really dramatic thing.”
Politics Podcast: Summer Mailbag Edition | FiveThirtyEight
Two weeks after Newman claimed to the state that he wasn’t operating the company anymore, he called Skeate into his office and demanded that Skeate return his company computer and provide the login credentials for the company’s cloud storage, where all of Skeate’s version control files were stored. Skeate refused and the meeting turned heated, according to a police report documenting the incident.
Skeate said in hindsight he wished he had just handed everything over to Newman, but at the time he didn’t know if the WSLCB had gathered enough evidence to hold Newman accountable.
“I should have just quit early on and walked away,” Skeate said. “I wasn’t certain if the [WSLCB] would do anything so I didn’t think it was ok for [Praxis] to get into the database and delete what may have been the only copy of the fraud.”
After Skeate left the meeting, Newman and two colleagues reported Skeate to the Centralia Police Department for allegedly stealing company property by keeping his work computer and not giving Newman the password to the company’s cloud storage. The police department then recommended Skeate be arrested and charged with first-degree theft for “refusing to turn over access of the database to Praxis,” according to a police report.
When I told Skeate, seven months after the incident, that the Centralia Police Department had referred a charge of first-degree theft against him to the county prosecutor’s office, Skeate sounded even more melancholy than he had previously in our interviews.
“That’s information to me,” Skeate slowly said, his voice trailing off. “I’ve just not gone to Centralia. I haven’t done anything wrong and I don’t think I should have to change my life but I’m not taking that threat for granted. Arrest … that’s a pretty serious threat.”
AN OPEN DATA SOLUTION
There’s a simple way to keep pot labs honest, according to Jim MacRae, the data scientist in Washington state: Just look at their data. MacRae said state regulators should pore over lab data to spot fraudsters while also releasing it publicly so private firms can help report suspicious activity.
“Looking at the fucking data and just looking for problems is a really good thing,” MacRae said. “Within three months of [Washington state] making the data available back in 2015 I had … informed the regulators that they had a problem. Those data should be available to journalists, to interested data scientists like myself, and social sciences the world over would be fascinated by this shit.”
The public can still request Washington’s pot lab data but data firms across the state say it’s no longer usable because a new state vendor has muddled the system since 2018. That’s left MacRae unable to investigate labs for the past three years.
Federal prohibition of pot also means that cannabis testing labs cannot have any federal business, which makes them even more indebted to the cannabis industry, according to Rosalie Pacula, a professor of health policy, economics and law at the University of Southern California.
“That’s not an effective design of incentives for ensuring labs place priority on consumer protection (rather than industry protection),” Pacula said in an email. “Effective regulation means random compliance checks and secondary testing to ensure that the testing labs are in fact doing the job they are expected to do (and protect consumer’s safety, not industry profits).”
The strictest lab regulations in the country are likely in Oklahoma, where a booming pot economy worth $800 million a year has sprung up on the edge of the American South. Pot labs face regular proficiency tests and the state requires labs to collect two samples for every test and then hold a reserve sample, which is used to investigate complaints. The second sample is also used as a calibration tool, with the state randomly retesting reserve samples. The lab must answer for any deviations between the first and second tests.
“If by some chance they are outside the [testing accuracy] window, the lab director must do some evaluation to figure out how that happened,” said Lee Rhoades, the laboratory oversight manager for the Oklahoma Medical Marijuana Authority (OMMA). “If it becomes a chronic situation where people simply cannot achieve the target value, there are provisions in our rules that allow us to take more action.”
Terri Watkins, a spokesperson for OMMA, said it was obvious to the state that it needed to design a rigorous lab testing program.
“I think it was a question of why take the chance? Two sets of eyes are usually better than one,” Watkins said.
On December 10, 2020, seven months after Skeate first contacted the WSLCB and more than two years after he first heard about the scheme, the state issued a 180-day suspension of Praxis Laboratories and filed a complaint seeking to revoke its license for “falsifying lab records” and reporting “false THC potency analysis” between April 2018 and September 2020. The agency said that the lab’s continued operations were “direct and immediate threats to the safety of the public.”
Newman denies all of the WSLCB’s allegations and initially appealed the state’s decision. In an April interview with FiveThirtyEight he said that the state failed to conduct a full investigation of Praxis but he later did not return follow-up calls after withdrawing his appeal in late May.
Newman told FiveThirtyEight in April that Skeate was only acting because he was about to lose his job and that Skeate had manipulated the files he had sent to the WSLCB. He also said the discrepancies in the decimal points were caused by Skeate’s software.
“Whatever he turned over the LCB did not match anything in any of our backups, we had hard copy backups, and we had a separate server that backed everything up and those all matched,” Newman said.
FiveThirtyEight was not able to independently evaluate Newman’s claims, but state investigators verified Skeate’s whistleblower data with data found at the lab.
These weren’t the first allegations of impropriety leveled at Newman. In 2016, Newman had been working as the laboratory director at a different Washington lab and was fired when questions were raised about the accuracy of lab results. The lab’s license was later temporarily suspended after a state audit found poor laboratory practices. That audit also uncovered a handwritten note sent from a pot farm to Newman that asked him to “not post any flower results under 20 percent” and instead have the farm send a new sample for a retest. The lab’s owner accused Newman of reporting THC potency values that were “pulled out of thin air,” according to the Seattle Times.
Newman told FiveThirtyEight that he’s never retested samples for potency at a farm’s request and said the 2016 note was saved “so anybody could see it and that was the last time we worked with [that farm].” Newman also defended himself by sending FiveThirtyEight an email thread from 2016, when he was the lab director, showing him refusing to retest samples after a client demanded new tests. “I thought we spoke about any flower lots below 25 percent or so would need another sample to be tested,” wrote the client. Newman also blamed his dispute with the lab owner in 2016 on the owner’s rejection of Newman’s demand for more equity in the lab.
As for the first-degree theft charge against Skeate, the Lewis County Prosecuting Attorney’s office said they are currently not pressing charges against Skeate, although Deputy Prosecuting Attorney Karin Phomma said, “If more evidence is provided in the future, our office would certainly review it.”
Skeate is now working on his own startup selling software to cannabis companies and labs. He said he doesn’t regret reporting Praxis to the state, though he wishes he did it sooner given the obvious need for tighter lab regulations. “It still seems like the laboratories are just self regulated,” Skeate said.
That was the case, at least, for Praxis. After all, the only reason the lab’s license was taken away was because of a young software developer’s willingness to risk his livelihood and criminal prosecution to do the right thing. And who’s to say there won’t be another Praxis? These labs are the clearinghouses for a pot economy that grows by billions of dollars every year. With the federal government absent, it’s beleaguered state lawmakers versus a billion-dollar incentive to chase THC. And the most potent test results, however dubious they might be, are winning.
Art direction by Emily Scherer. Charts by Elena Mejía and Jasmine Mithani. Copy editing by Maya Sweedler. Story editing by Chadwick Matlin.