Lessons from the World’s Largest Darknet Marketplace: Part 2

Author: Death Star
Date: November 22, 2023

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

2.4 Sellers

Like other darknet marketplaces, HYDRA allowed sellers to register and list a wide variety of drugs for sale. However, HYDRA’s seller functionality had two unique features. First, its features were tailored to the delivery system used by the marketplace: HYDRA automated the assignment of couriers and management of stash drops. Second, the marketplace interface allowed multiple users to operate a single store. There were three types of users: administrators, operators, and couriers. Administrators could assign roles, change store profiles and settings, and conduct financial transactions. Operators handled mid-level tasks, such as resolving disputes and answering customer questions. Couriers were responsible for delivery, i.e., making stash drops.

2.5 Marketplace Services

This section describes the main services and features HYDRA provided to its users.

2.5.1 Reviews and Ratings

Previous research has highlighted the crucial role of reputation on darknet marketplaces. HYDRA’s system included two feedback channels: ratings and reviews. After each purchase, buyers had 24 hours to leave a review.

The simplest information for potential buyers to assess a seller and product was the average rating. On HYDRA, users could rate from 0 to 10. Both the seller’s average rating and the average rating for each product were visible. As with AlphaBay and SilkRoad, most ratings on HYDRA were very high. Notably, the average rating was close to 10, and only 4% of orders had a rating below 10 (see section 3.6 for detailed analysis). This can be partly explained by the fact that if a user did not leave a rating within 24 hours, the platform automatically assigned a rating of 10. There was also a “reputation inflation” effect, where users left higher ratings even if not fully satisfied. This is supported by many cases where users expressed negative sentiment in the review text but still gave a 10/10 rating, consistent with findings that review text is less sensitive to “inflation” than numerical scores. (See Table C.1 in Appendix C for examples of such reviews on HYDRA.)

2.5.2 Dispute Resolution

HYDRA allowed buyers to initiate disputes with sellers, usually related to issues with finding the stash or with the product’s quality or quantity. Disputes began with communication between buyer and seller via the internal messaging system. Typical outcomes included refunds, replacement products, or discounts on future purchases. If the parties could not resolve the dispute, a platform moderator would review the messages and make a decision. Moderators were more likely to side with buyers who had a long purchase history and few disputes, incentivizing users to maintain their accounts and internal ratings rather than creating new ones.

Note from Death Star: The opposite situation also occurred—if you had a bad account history, even a legitimate claim might not go your way!

2.5.3 Quality Control

HYDRA also implemented methods to ensure high product quality. Administrators regularly published information about product checks, which they claimed were conducted on some items listed for sale. Only a small number of products were checked each month. For example, in June 2021, only 47 test results were published (15 in Moscow, 32 in St. Petersburg). The administration claimed that priority was given to sellers/products with more negative reviews or disputes, and that tests were conducted blind (the tester did not know the seller’s identity). Results were posted on the marketplace forum and included details about the stash search process, packaging, quantity versus stated amount, and chemical test results. For powdered drugs, purity was reported (e.g., very high quality, VHQ, or high quality, HQ). This provided buyers with more information and directly incentivized sellers to maintain high standards, given the risk of penalties for poor test results.

2.5.4 Special Statuses

HYDRA allowed sellers to obtain two special statuses to stand out and improve business processes, similar to status ratings on some Western darknet marketplaces.

• Trusted Sellers: This status could be obtained by sellers with over 1,000 total sales and a dispute rate below 7%. It increased buyer trust, improved search ranking, and allowed trusted sellers to make decisions in disputes before HYDRA moderators intervened. Trusted sellers could also have franchisees—smaller partners who produced synthetic drugs or grew marijuana and collaborated with the seller using HYDRA as an intermediary.
• Certified Manufacturers: For synthetic drug sellers, HYDRA offered a certified manufacturer status, which could boost sales by signaling relative safety. Buyers could filter search results to show only certified manufacturers. This required significant involvement from HYDRA staff, including verification of the production cycle, chemicals, equipment, and worker qualifications. Sellers had to provide photos and detailed information about their facilities as part of the application.

2.5.5 Professional Training

HYDRA offered a “Courier School” page, selling online courses for couriers and dispute resolution services between employers and employees. According to reviews, HYDRA’s training reduced risks associated with stash drops. HYDRA also provided consulting services for novice manufacturers, including help acquiring precursors and equipment and training production staff.

2.5.6 Medical Assistance

HYDRA provided free medical consultations for drug users, actively recruiting medical professionals for this service. Consultations were conducted via Telegram chatbot, the site’s messenger, or the platform’s forum. While the popularity or effectiveness of these services is unknown, their provision and promotion represent a new harm reduction innovation in the darknet market.

2.6 Business Model

HYDRA had several main revenue streams. First, it charged sellers a monthly fee and commissions on all marketplace transactions. Second, HYDRA auctioned off positions on the site’s main page. Third, it sold special statuses to major sellers.

Platform Fees

HYDRA charged sellers to operate on the platform. As of March 19, 2022, opening a new store cost about $300, with a monthly rental fee of $100 (unchanged since 2019, according to Proekt).

Transaction Commissions

HYDRA charged a sales commission depending on the type and value of the drug. For sales under 200,000 rubles, the commission was usually 5%. For some drugs, it was higher, up to 10%.

Note from Death Star: Marijuana and hashish had a 5% commission, “salt” 7% or more. At one point, commissions on synthetic drugs were raised.

Main Page Positions

Sellers could pay for one of 20 positions on the site’s main page, distributed via monthly auction. The highest bidder got the first spot, the second-highest the second, and so on. Store owners could watch the bidding. According to Proekt (2019), main page positions cost tens of thousands of US dollars per month at the time of writing.

Note from Death Star: One record-breaking bid was when GangBang outbid Alchemist and paid over 40 million rubles for first place (around 2019–2020).

Fines

HYDRA could fine sellers for violating platform rules. Fines ranged from 25,000 rubles (for placing stashes too close together) to several hundred thousand rubles. The largest fines were for maintaining monopoly positions. As mentioned in section 2.2, stores could be fined for encouraging customers to buy drugs off-platform or register on other darknet marketplaces.

Special Statuses

The “trusted seller” status cost $1,000 per month. The “certified manufacturer” status required sellers to share an additional 3% of all sales with the platform.

Escrow (Added by Editors)

The final revenue stream was escrowed transactions. Escrow deals were only possible for pre-orders. During a pre-order, the buyer invited an escrow agent to the deal. The seller and buyer discussed terms (stash location, delivery time, packaging) in the presence of the escrow agent. If both agreed, the buyer transferred funds to the escrow agent, who then notified the seller to prepare the order. The seller had to provide video/photo evidence of the product and packaging, shot in a way that ruled out substitution. Once the product was ready, the seller provided the video and began delivery (making the stash or sending the package), then gave the address or tracking number to the buyer. The buyer had 24 hours to confirm receipt. If the buyer did not respond within 24 hours, the deal closed in favor of the seller.

In short, escrow ensured fair transactions. Regular buyers rarely used escrow, as they mostly bought instant positions, but for stores, escrow was a necessary (and costly) service. The escrow commission ranged from 3% to 10% of the deal amount:

• 1,000–50,000: 10%
• 50,000–500,000: 7%
• 500,000–1,000,000: 5%
• Over 1,000,000: 3%

3. Marketplace Activity Analysis

3.1 Data

For quantitative analysis, we used two complementary datasets. The first was a set of drug sale listings scraped from HYDRA, containing daily screenshots of listings from April 1, 2020, to May 2, 2020. Each listing included drug type, quantity, price, title, seller name and ID, and delivery method (pre-order or instant, mail or stash drop). For instant listings, stash type and approximate location were noted. The dataset contained 31,035,506 listings, acquired from an independent collector who also provided data for several journalistic investigations (Knife Media, 2020; Proekt, 2019).

Note from Death Star: Shoutout to DrugStat for the screenshots above!

Sample HYDRA listings have also been used in drug use research, showing that opiate listings strongly correlate with fatal overdoses.

The second dataset came from a data provider in Pennsylvania, USA, who continuously collects data from the world’s largest darknet marketplaces. Details can be found in Christin (2015, 2022). This dataset included a large subset of reviews left on the platform, with information on the product, seller, buyer nickname, review time, and rating. Reviews for job ads or non-drug items were excluded. After removing duplicates, there were 325,740 unique reviews from January 2019 to February 2022. The sample covers only part of all reviews, as the provider could not scrape all site pages due to technical issues. Because scraping coverage varied daily, we did not use these data to estimate the number or types of drugs sold on HYDRA, but only to analyze the reputation system. Our conclusions do not depend on observing all market reviews.

3.2 Drug Sales

Table 2 (section 2.5.1) shows the composition of stash drops on HYDRA in April 2020. The original dataset contains 31,035,506 listings. We focused on instant listings, which are more directly linked to actual sales, and excluded pre-orders and mail orders, leaving 2.8 million listings. However, listings for the same stash appeared multiple times, as they were posted daily until sold. Thus, in Table 2, we removed duplicate listings at the “supplier–quantity–location” level, resulting in about 417,000 unique drug sale listings in April 2022. Most listings were for stimulants or euphorics.

The most popular drugs on the platform were mephedrone (31%), cannabis (18%), amphetamine (13%), and alpha-PVP (12%). Our finding that mephedrone had the most listings on HYDRA matches other data on its extreme popularity in Russia.

Note that a single listing could represent several stashes with identical characteristics hidden in one place, so the table likely underestimates the true number of ready stashes across Russia. According to Proekt (2019), the real number could be five times higher. Thus, ≈417,000 is a conservative lower bound for the number of ready stashes in Russia in April 2020. For comparison, at the peak of major Western darknet marketplaces like Silk Road, Silk Road 2.0, Evolution, and Agora, the number of drug listings was under 25,000. However, since HYDRA’s definition of a listing differs from most other marketplaces, a more meaningful comparison is the number of sellers. Using this metric, HYDRA was much larger than major Western platforms, with 4,907 unique sellers in one month versus 2,604 on Agora over a much longer period (December 2013–August 2015).

Table 3 provides summary statistics for individual drug types. As with Western darknet marketplaces, heroin made up a small share of the market. Although heroin sales were allowed on HYDRA, our informants suggested that most heroin trade in Russia occurs through more informal means, such as Telegram bots or offline/street sales. Other drugs were widely available, many sold in about 500 different cities and towns. Unlike most other darknet platforms, a significant share of HYDRA listings appeared to be for redistribution. Demant et al. (2018) estimated that only 2.1% of deals on Agora and Silk Road 2.0 were wholesale (over $1,000), with the rest for personal or social use. Using this definition, we found that for the categories in Table 3, the share of wholesale sales ranged from ≈50% for alpha-PVP to ≈15% for heroin. If we limit the data to instant orders only, the share of wholesale listings drops to 1–5%. Thus, most business-to-business deals occurred via pre-orders, as confirmed by several informants.

Figure 3 shows the distribution of mass/quantity for several drug types. The most common quantities were 0.5, 1, 2, 3, and 5 grams for most drugs. Semi-legal and pharmacy drugs like Xanax were sold as tablets in larger quantities. Methadone was rarely sold in amounts over 1 gram.

3.3 Geography

3.3.1 Within-City Distribution

In interviews, convenience of purchase location was a key factor differentiating listings. The distribution of listings by city district reflects spatial demand. Figure 4 shows the share of three main drug categories in total listings for each Moscow district. As expected, cocaine was usually hidden in the city center, near business districts, while cheaper drugs like synthetics and marijuana were more popular in the outskirts. Similar patterns were seen in other major Russian cities. Figure A.1 in Appendix A shows this for St. Petersburg.

3.3.2 Between-City Distribution

HYDRA operated in 1,129 localities across all Russian regions, covering over 100 million people who had access to at least one stash in April 2020. This means about 69% of Russia’s 144 million residents had access to HYDRA drugs in their city or town at the time. Figure 5 shows the distribution of cities with at least one instant HYDRA listing. Site availability was positively correlated with population size. In Moscow and St. Petersburg, there were 3 and 9 stashes per 100 residents, respectively. Sochi stood out among smaller cities with 16 listings per 100 residents, likely due to its popularity as a resort. Many small satellite towns near Moscow (like Aprelevka or Solnechnogorsk) had especially high per capita listings, likely due to easy access, less law enforcement, and more parks—making them convenient for storage, transport, and serving some Moscow consumers.

3.3.3 Regional Distribution

Figure 6 shows HYDRA activity by region. Some regions had higher per capita listings, especially Moscow and neighboring regions (Moscow, Kaluga, Tula, Yaroslavl), as well as St. Petersburg and regions with major ports or long borders (e.g., Karelia, Kaliningrad). HYDRA’s regional reach reflected population and urbanization: more people live in European Russia and southern Siberia (especially Krasnoyarsk Krai). Kamchatka Krai also had high listing density, but its low population makes conclusions difficult.

3.4 Market Concentration

In April 2022, HYDRA had over 4,500 stores, indicating high market competition. To quantify competition, we analyzed store characteristics: number of listings, cities/regions served, specialization, and overall market competition. We summed listing prices to estimate each store’s volume and market share.

Ideally, market share would be based on transaction counts, but our data only included listings. While Tzanetakis (2018a) estimated that only 69% of AlphaBay listings led to sales, we assumed a strong link between instant listings and sales on HYDRA, as instant listings represented ready stashes that required significant effort and risk to maintain. Thus, 69% is a lower bound for the share of listings that resulted in sales. Our analysis of market shares should not be affected as long as the unsold listing ratio is similar across stores, though this adds uncertainty.

We calculated the Herfindahl-Hirschman Index (HHI) for unique retail listings in April 2020 (see section 3.2), excluding wholesale and artificially overpriced listings (only listings under $1,000 were included, removing 6,000 of 417,000 listings). HHI ranges from ≈0 (perfect competition) to 1 (monopoly), with values below 0.15 indicating a competitive market. HYDRA’s overall HHI was 0.004, indicating a highly competitive market.

Looking at the largest and typical sellers, the top retail seller, “Prazdn1k,” had 4,527 listings worth $471,096 (2% of total retail volume that month), selling nine drug types (mostly mephedrone or cocaine) in 34 regions and 54 cities. “PokemonGo,” the second-largest, had 5,416 listings worth $406,577, active in 45 regions and 102 cities, selling eight drug types (mainly mephedrone and amphetamine). All other sellers had 1% or less market share, with 30% of stores having fewer than 10 unique listings in the month. The average HYDRA seller was small, operating in one or two cities, with 10 listings for one or two drug types.

We also calculated HHI for the most popular drugs and cities. Cocaine (HHI = 0.027) and mephedrone (HHI = 0.014) markets were the most competitive, but still highly competitive overall (see Table 4). There was a negative correlation between city population and market concentration (see Figure 7). Large cities had low HHI and competitive online drug markets. Smaller towns showed more variation, with some appearing to be monopolies on HYDRA.

To be continued…