Lessons from the World’s Largest Darknet Marketplace: Part 3

Death Star — November 22, 2023

This is an original translation of a foreign scientific study on the HYDRA darknet marketplace and its operations.
Authors: phØBoŠ & hŮmAnǿ1Đ

3.5 Price Decomposition

Within a single store, pricing on HYDRA tends to be nonlinear: the price per unit decreases as the order volume increases. While this nonlinearity is partly due to wholesale deals—where dealers buy from larger suppliers—there are also fixed operational costs imposed on suppliers due to the illegal nature of the market. One of the main components of these fixed costs is delivery expenses. Given the high risk couriers face when storing drugs, their compensation likely makes up a significant portion of the transaction cost, which remains relatively unchanged regardless of the amount of drugs stashed. The cost of a stash can be seen as an example of costs associated with illegal drug trafficking, typically understood within the risk and price model. According to this model, prices for illegal goods, especially drugs, initially include risks related to law enforcement activities and the internal social organization of the illegal market. Other costs likely borne by sellers include the initial delivery of stashes to the courier and store management expenses. To separate these, we use the following strategy.

In section 3.4, we established a very high level of competition among sellers on HYDRA. In a competitive market, the price set by sellers should equal the total cost of supplying the product. This includes raw material costs, participants’ time, and compensation for legal risk. These costs can be divided between delivery organization and the acquisition or production of the drug. Since HYDRA was a large employment market (see sections 2.3 and 2.4), we assume delivery costs are roughly the same for all sellers. Additionally, if each type of drug has little product heterogeneity and sellers use similar production technologies or buy from the same competitive market, the cost per gram should be the same for each seller. Under these simplified assumptions, the price set by sellers should be determined by the following formula:

where DC is the fixed delivery cost, and PGC is the cost per gram. Based on this simplified economic model, we run the regression model below, allowing the fixed cost of stashes to depend on the type of drug. That is, for each drug type, we estimate the regression using the following formula:

where the sample consists of all listings i for a given drug. We also include fixed effects Ys(i) at the seller level, where s(i) is the seller of listing i. We set the constants so their sum equals zero. In this model, α represents the average fixed costs, which we expect to reflect the costs associated with hiding the stash. Seller-level fixed costs reflect possible differences in costs and prices among sellers. We limit our estimation to small retail listings.

The results for Moscow are shown in Table 5. Each column presents results for a specific drug. For cocaine, the most expensive drug in Russia (column 1), we see that each additional gram increases the listing price by $97.5 on average. For marijuana buds and hashish (columns 2 and 3), the marginal weight effects are much lower: $19.7 and $17.7, respectively. The fixed costs for listing cocaine average $32.4, which we primarily associate with courier expenses for placing the stash. For marijuana buds, hashish, amphetamine, mephedrone, and heroin, fixed costs are similar, ranging from $13.8 to $17.3. Figure B.1 in Appendix B shows our estimates are close to the piece rates found in job ads for couriers on the market that became popular after HYDRA’s closure (see section 4 for details on HYDRA’s closure). Finally, the R2 obtained is very high for all drugs, further supporting the chosen linear specification.

Table 5 reveals two properties of HYDRA’s price components related to fixed costs. First, the fixed effects in the regressions differ by drug type. We believe this reflects differences in delivery costs for various drugs. For example, cocaine stashes are the most expensive, likely due to several factors. First, cocaine is usually hidden in central areas (see Fig. 4), where police activity is more intense, increasing courier risk. Second, cocaine stashes are expected to be of higher quality: the cost of losing a cocaine stash is higher, so they must be better hidden. Finally, the risk of couriers absconding with a batch of drugs was a major concern for stores, and the higher street value of cocaine made them especially vulnerable. As a result, couriers were paid more to discourage theft.

The second property is that the estimated fixed costs are high relative to the per-gram costs. The stash cost is a significant part of the price for small orders. For example, comparing these results to the averages in Table 3, for cocaine, delivery costs are about one-third of the average stash price. For amphetamine, delivery costs are more than half the average price. This suggests that, although the market adapted, Russian laws imposing stricter controls on mailed packages still somewhat restrain online drug trade by raising prices and thus implicitly reducing demand.

3.6 Reviews and Ratings

Table 6 shows the distribution of ratings users left after purchases. As with other darknet marketplaces, HYDRA ratings were mostly at the maximum value. Despite this, our informants consistently say that reputation was important on HYDRA, especially for consumers choosing between sellers. We assume the review text was more important than the numerical rating. For example, the fourth review in Fig. 2 criticizes the product, stating “low quality,” but still gives a 10/10 rating. Table C.1 in Appendix C provides more examples of such reviews.

To confirm this idea, we analyzed the content of the text data. The average review length is 16.3 words. Table 7 lists the most frequently used words in HYDRA reviews. The language used by reviewers reflects buyer preferences on the platform, supporting our hypothesis that reviews play a key role in the market’s reputation system. In particular, many words are related to the delivery process, such as “place,” “touch” (used to mean the stash could be picked up quickly), “find,” “simple,” and “quest.” Other words discuss the product itself, like “quality” and “material.” Finally, many words describe overall satisfaction: “thanks,” “neatly,” “cool,” “super,” “precisely,” etc.

To better understand how reviews inform consumer experience, we also analyzed negative reviews separately, focusing on those with the lowest possible rating of 0. Negative reviews are much longer, averaging 48.0 words, indicating users describe negative experiences in more detail. Table 8 lists the most common words in negative reviews. Two main groups of words stand out: first, words like “dispute,” “favor,” “moderator,” and “close” relate to the dispute system (see section 2.5.2); second, words like “stash,” “location,” and “not found” relate to negative stash-finding experiences.

3.7 Data Limitations

There are three main limitations to the dataset used in this analysis. First, the data only covers April 2020, when Russia first imposed COVID-19 lockdowns. This likely affected both supply and demand for drugs on HYDRA, impacting couriers’ ability to hide stashes and consumers’ ability to retrieve them. Studies on the pandemic’s impact on darknet transactions show it was linked to more failed deliveries. Research on online drug sales via Telegram in the Netherlands shows the pandemic had mixed effects on the popularity of different drugs: “party drugs” like ecstasy, cocaine, and amphetamines declined, while psychedelics became more popular. Although drug popularity may have shifted during this period, analysis by Flashpoint and Chainanalysis (2021) shows HYDRA’s revenue dropped by only about 10% in April 2020 compared to February 2020. Thus, we believe our policy conclusions based on HYDRA’s size remain relevant despite the limited data period.

The second limitation is that the short data period prevents us from analyzing market development over time. We hope future research will use longer panel datasets to study HYDRA’s evolution.

Finally, we used secondary data sources and cannot publicly name the individuals and organizations who provided data due to anonymity concerns. We recognize this may further limit the external validity of our results due to replication challenges.

4. The Shutdown of HYDRA

Note: Added by the Death Star editorial team

Here is a brief overview of HYDRA’s shutdown and how the drug market landscape changed afterward. These changes highlight HYDRA’s impact on the Russian drug market and will have important implications for our policy analysis in section 5.1.

HYDRA was shut down on April 5, 2022, when German police, in a joint operation with U.S. law enforcement, seized the servers hosting the marketplace. On the same day, the U.S. Treasury Department imposed sanctions on HYDRA, and the U.S. Department of Justice charged the alleged HYDRA administrator. There is no evidence that Russian authorities were involved in or informed about the operation. The shutdown by two Western governments can be seen as part of the growing number of Western sanctions against Russia in 2022, though there is no evidence these events are connected. Moreover, Germany’s Federal Criminal Police Office stated in a press release that the investigation began in August 2021. Despite some market participants hoping for HYDRA’s return, the marketplace has remained offline for over a year as of this writing.

In the days following the shutdown, buyers and sellers tried to trade using two darknet forums popular among HYDRA users: LegalRC and RuTor (see also section 1.2). Additionally, sellers began using their own websites or Telegram bots as sales channels. While Telegram bots remain popular, demand for a centralized marketplace quickly led to rapid growth of previously small darknet platforms. The largest platforms to emerge after HYDRA’s closure are currently OMG, Blacksprut, Mega, and Solaris. Another popular marketplace, KRAKEN, launched in December 2022 and presents itself as HYDRA’s successor, run by people connected to the closed marketplace.

According to media reports, there is fierce competition among the new marketplaces. They have repeatedly tried to advertise themselves offline or via social media. Notably, the competitive environment has fueled a series of large-scale hacking and distributed denial-of-service (DDOS) attacks organized by markets against their rivals. These attacks, along with the platforms’ lower technical capabilities, have made it difficult for them to handle large numbers of visitors, making them much slower and less user-friendly than HYDRA. The fact that individual marketplaces regularly go offline forces market participants to operate on multiple platforms at once. This increases search costs for consumers, who must spend time looking for products and sellers across several marketplaces. For sellers, it raises administrative costs due to the extra effort needed to manage multiple platforms.

The consequences of market shutdowns go beyond reduced convenience. As discussed in section 2.5.1, reputation mechanisms are crucial for darknet markets to function in otherwise high-risk environments. However, in the new unstable environment, sellers have less incentive to maintain a good reputation compared to when HYDRA was a de facto monopoly. Reports of fake listings—where drugs are not actually present—are becoming more common, and concerns about product quality have increased. These problems are worsened by the platforms’ low functionality. For example, users note the lack of reliable moderation, which, along with less trustworthy reputations and histories, reduces the chances of satisfactory dispute resolution.

5.1 Policy on Market Shutdowns

The HYDRA example shows that a darknet marketplace allowed to grow unchecked can dominate the online market, reaching enormous size and sophistication. Given limited law enforcement resources, the decision to target darknet marketplaces requires analyzing the overall consequences of such actions. We believe HYDRA’s counterfactual example highlights the main costs of this outcome and illustrates the trade-offs governments face when deciding whether to invest in shutting down individual darknet marketplaces. Below, we outline the main arguments for and against shutting down darknet platforms, supplemented by evidence from the HYDRA analysis. We start with arguments in favor of closing large darknet marketplaces.

1. Large marketplaces can increase drug demand.
   We lack reliable data on changes in drug consumption due to HYDRA’s emergence, as there are no surveys or other data. However, we suggest several mechanisms by which a large marketplace could increase consumption. In particular, a major darknet marketplace reduces many of the main costs of drug use. Having a single well-known platform lowers search costs for consumers entering the market. This applies not only to inexperienced users but also to those wanting to try new drugs or use drugs away from home, given the wide variety of drugs (see section 3.2) and cities (see section 3.3) available. HYDRA was also more user-friendly than smaller platforms, offering a well-designed, fast, and reliable site. Such reliability reduces risks for consumers in the illegal market. The risk of exit scams was low, reducing the chance of users losing funds. The platform’s moderation system and long-standing reputation system also reduced the risk of being scammed by sellers. Lowering these significant costs compared to smaller markets should not only encourage some users to switch from street dealing to the online market but also prompt some potential users to start using drugs on the platform. This is partly supported by correlation data: Russia’s Federal State Statistics Service reported an increase in overdose deaths in 2020 and 2021, with experts naming HYDRA as a key factor (Meduza, 2022).
2. Large marketplaces increase drug supply.
   We found that HYDRA hosted a significant number of wholesale drug listings (see Table 2). We also recorded active trading of precursors. Finally, the platform had job ads for individual producers, such as chemists and marijuana growers. This suggests a large darknet market facilitates connections among supply-side actors. This should have two effects: first, it should bring more participants to the market by lowering fixed costs of organizing drug sales, such as finding staff, precursors, and wholesale batches. Second, increased openness in the precursor and wholesale markets should boost competition and lower production costs. In HYDRA’s highly competitive environment (see section 3.4), lower costs should be passed on to consumers as lower prices. Thus, increased competition among both retail and wholesale suppliers should lead to lower prices and higher overall trade volumes.
3. Lasting impact of large marketplaces.
   After HYDRA’s closure, several marketplaces quickly expanded to fill the gap (see section 4). This trend mirrors the Western experience, where users soon migrate to replacement markets after shutdowns. However, this does not mean shutdowns are ineffective. The Russian case illustrates the long-term consequences of allowing HYDRA to grow, creating a foundation for significant online transactions. Although the current generation of Russian darknet marketplaces is smaller than HYDRA at its peak, TRM Labs estimates that in the five months after HYDRA’s closure, Bitcoin wallets linked to these new platforms received 24% more Bitcoin than HYDRA did in the five months before. Existing darknet users—both buyers and sellers—have already paid the “fixed costs” of learning how to use these platforms, including understanding TOR and Bitcoin for anonymity, finding valid darknet links, and, for sellers, managing logistics. This means many participants can quickly migrate to new markets. Thus, if a large darknet marketplace encourages potential users to learn these skills (due to its reliability and convenience), these factors persist. Allowing the market to grow unchecked may cause law enforcement to lose the chance to curb online drug trade in the future. This is evident in the West, where regular market shutdowns limit the number of users able to migrate to new platforms. TRM Labs estimates that the four largest new Russian darknet markets account for 80% of global darknet crypto transactions, while the largest Western market, ASAP Market, accounts for just 7% (TRM Labs, 2022).
4. Eliminating dominant marketplaces increases competition and destructive attacks.
   As described in section 4, after HYDRA’s closure, new markets began launching DDOS attacks against each other to gain dominance. The resulting competition destabilizes online drug trade without requiring extra government resources. Markets formed after HYDRA’s closure actively undermined each other. For example, Solaris was temporarily shut down by hackers working for its competitor Kraken. In contrast, HYDRA’s dominant position before closure seemed to protect it from such threats.

The above reasons point to serious negative consequences of creating a large dominant market like HYDRA. However, there are also important arguments against aggressively targeting darknet marketplaces and instead allowing a single dominant platform to develop.

1. Reputation systems improve drug quality.
   Sections 2.5.1 and 3.6 describe the importance of reviews for consumers choosing sellers on HYDRA. Previous studies also highlight the importance of reviews and reputation on darknet markets (see Espinosa, 2019; Hardy & Norgaard, 2016; Aldridge et al., 2018; Bhaskar et al., 2019). HYDRA’s longer history and higher transaction volume suggest its reputation system was more informative than others. This allows consumers to better distinguish sellers offering purer drugs from those selling dangerous fakes. It also raises equilibrium quality compared to smaller platforms, as the incentive to maintain a good reputation is stronger. Thus, a large monopoly market can improve drug quality, which may benefit consumer health compared to other sources. However, higher drug purity may also increase overdose cases, especially with opiates.
2. Large marketplaces have incentives to self-regulate.
   Section 1.2 notes that HYDRA banned the sale of weapons, poisons, contract killings, explosives, state secrets, and pornography. Importantly, it also banned fentanyl and other drugs considered too dangerous. HYDRA claimed to check drug quality (see section 2.5.3) and had representatives who allegedly provided medical advice (see section 2.5.6). This suggests large darknet platforms have incentives to regulate participants and protect their reputation. For example, a large marketplace has more incentive to reduce overdose or arrest risks for users, as it expects future profits from them (unlike a small, fragmented market with little chance of user return). Thus, a large marketplace can proactively reduce harm to its users. Less focus on shutting down the marketplace encourages self-regulation in other ways, as shown by HYDRA’s rules. If the government does not try to close all darknet platforms, those earning mainly from drugs benefit from banning “socially dangerous” goods like contract killings, which attract police attention and increase shutdown risk.
3. Large marketplaces can be used for harm reduction efforts.
   A centralized market is a convenient place for information campaigns aimed at drug users. For example, Davitadze et al. (2020) describe an NGO’s attempt to provide HIV and hepatitis C information and promote harm reduction services on HYDRA’s forum. HYDRA cooperated with this NGO and promoted the information on its Telegram channel. Similarly, Bancroft (2017) shows how a large marketplace can facilitate harm reduction information sharing among drug users. Barratt et al. (2016) provide evidence of more controlled drug use on online platforms.
4. Darknet marketplaces may reduce violence.
   Currently, there is no quantitative data on HYDRA’s impact on crime in Russia. However, previous studies show online markets lead to less violence than traditional street dealing due to fewer face-to-face contacts. HYDRA’s scale meant it could facilitate supply-side operations, including recruitment and wholesale drug trade. Thus, the benefits of reduced violence likely extended throughout the supply chain, and were probably greater than if only retail trade had moved online.
5. Darknet marketplaces increase drug market transparency.
   By nature, the illegal drug market is extremely hard for governments to monitor. Online markets can be parsed, and this data can inform future drug policy. For example, it helps track changes in national drug consumption and may allow authorities to spot dangerous trends, such as shifts to high-risk substances, earlier.

Both the positive and negative effects of shutting down large darknet marketplaces can be greatly amplified, as large platforms benefit from economies of scale. In other words, large marketplaces tend to grow as they become more efficient. First, a large marketplace can invest more in developing and maintaining its site. Section 2.5 demonstrated the sophistication of a large platform like HYDRA. Second, the site’s popularity can grow further as it increases the number of available listings. Our informants mentioned that in one city, HYDRA’s popularity surged after stashes became available in all neighborhoods, and drug users learned they could always find a listing nearby.

The HYDRA case highlights new and important aspects of the trade-off governments face when deciding whether to pursue mass shutdowns of popular darknet marketplaces. In particular, we show that a large platform like HYDRA can facilitate communication between drug producers and sellers while reducing non-monetary costs for consumers. We also show potential benefits of large darknet marketplaces that have received little attention. Specifically, HYDRA had an effective reputation system that could improve drug quality and related health outcomes, was able to move even wholesale drug trade online (potentially reducing gang violence), and was self-regulating in several respects.

5.2 Policy Implications

As a marketplace, HYDRA was similar to Western platforms in terms of the sociotechnical practices used to address three main problems in online drug trade: evaluation, competition, and cooperation. At the same time, HYDRA operated in an environment with several distinctive features compared to the Global North: a punitive drug policy, lack of officially approved harm reduction practices, and strict mail delivery controls. Combined with the country’s size, large population, and geopolitical position, this affected online drug trade in three main ways. Below, we discuss these factors and their policy implications.

Stash Delivery

As an unofficial mechanism, stash delivery is similar to mail delivery, as it does not require a face-to-face meeting. At the same time, stashing is an innovative way to bypass the risks of mailing illegal drugs in Russia. Its advantage is also speed. With mail deliveries carrying a high risk of law enforcement interception, the Russian online drug market adopted stashing as the dominant method. In response, Russian police seem to focus on detaining people engaged in suspicious activities, such as hiding or searching for stashes in parks and urban areas. Given the undesirability of search policies in the U.S., this delivery method presents a complex trade-off for law enforcement: choosing between controversial and potentially harmful policies and reducing the ability to curb retail drug supply. Our analysis supports Aldridge and Askew’s (2017) view that Western governments should pay close attention to this possibility.

Financial Management

One element of the Russian market that seems to have greatly contributed to HYDRA’s popularity was the ability to top up accounts via QIWI terminals (see section 2.2). This payment method protected client anonymity and lowered technical barriers to darknet drug trade participation. The literature emphasizes the importance of financial regulation in limiting darknet marketplace activity. As cryptocurrency exchange regulation tightens and crypto transaction anonymity decreases, alternative payment methods may become more important in darknet drug trade. HYDRA’s experience shows that regulating money transfer services like QIWI is a key step in limiting crypto drug markets.

Harm Reduction

Sections 2.5.3 and 2.5.6 describe how HYDRA used harm reduction methods: medical consultations on drugs and selective quality checks of drugs sold. These platform practices indicate high demand for harm reduction services in countries currently using punitive drug policies. This highlights the need for governments to provide broader harm reduction information, giving drug users access to reliable information from medical experts. Given that the long-term deterrent effect of shutting down shadow markets on crime is small (if it exists at all), this suggests shifting from a punitive/prohibitive approach to online drug markets toward harm reduction may be a more effective government response.

This aligns with existing data on drug supply interventions before darknet marketplaces: “Neither prevention, nor treatment, nor law enforcement has demonstrated the ability to affect the scale of drug use and addiction; the best that government measures can do is reduce harmful consequences.”

CONTINUED IN THE NEXT PART