The Network Structure of Opioid Distribution on a Darknet Cryptomarket



The current study is the first to examine the network structure of an encrypted online drug distribution network. It examines (1) the global network structure, (2) the local network structure, and (3) identifies those vendor characteristics that best explain variation in the network structure. In doing so, it evaluates the role of trust in online drug markets.


The study draws on a unique dataset of transaction level data from an encrypted online drug market. Structural measures and community detection analysis are used to characterize and investigate the network structure. Exponential random graph modeling is used to evaluate which vendor characteristics explain variation in purchasing patterns.


Vendors’ trustworthiness explains more variation in the overall network structure than the affordability of vendor products or the diversity of vendor product listings. This results in a highly localized network structure with a few key vendors accounting for most transactions.


The results indicate that vendors’ trustworthiness is a better predictor of vendor selection than product diversity or affordability. These results illuminate the internal market dynamics that sustain digital drug markets and highlight the importance of examining how new anonymizing technologies shape global drug distribution networks.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2


  1. 1.

    Colloquially, anonymous activity on Tor and other encrypted networks are often referred to as activities on the ‘darknet,’ although this is not technically correct.

  2. 2.

    Most literature in this field refers to Tor markets as ‘cryptomarkets.’ We use the term Tor marketplace to avoid confusion with the specific marketplace we examine, Cryptomarket.

  3. 3.

    Decary-Hetu et al. (2016) note that risk is greater in international distribution because sanctions are harsher than domestic distribution and because contraband crossing national boarders is more likely to be seized than contraband traveling within national borders.

  4. 4.

    Cryptomarket was also selected for data collection because it is the only large Tor market that provides the entire username for both vendors and buyers, allowing network ties to be identified. Most other Tor drug markets encrypt usernames to preserve anonymity. The ability to match identifiers is necessary to reconstruct the network.

  5. 5.

    Schedule 1 is the highest degree of control in the US. It is reserved for substances with high abuse potential and no approved medical usage.

  6. 6.

    Many vendors miscategorize their listings purposely to advertise to users who typically use different drugs. We classified each drug according to their chemical category, rather than how they were listed on Cryptomarket.

  7. 7.

    Reasons for poor transactions include long shipment times, products not being delivered as described, poor communication between vendors and sellers, and non-delivery of items.

  8. 8.

    Unfortunately, the country of origin for buyers is not listed on Cryptomarket.

  9. 9.

    It is important to clarify here that all vendors were still active, even though the listing may not be. For example, when the webpages were downloaded, the vendor may no longer sell a particular drug, even though the vendor had sold that drug at some point in the prior 6 months.

  10. 10.

    This did not result in any missing ties in the network data. Missing ties are the source of most missing-data based estimation problems in network analysis (Robins et al. 2004; Wang et al. 2016). Missing tie values were present for 37% of ties, but did not impact ERGM results when we reran the models without the affordability variable—the only variable based on tie values.

  11. 11.

    An actor’s degree score and degree centrality are synonymous.

  12. 12.

    To make interpretation easier, we added ‘6’ to every vendor’s reputation score so that 0 becomes the lowest score possible, indicating vendors who have not made a sale in the past 6 months. We compared this decision to z-score and logarithmic transformations of the reputation score to verify the robustness of our results.

  13. 13.

    We also ran the ERGM with an average evaluation per sale measure for vendors, finding the same pattern as that presented below. Because the ERGM controls for the number of sales made to each vendor, we used the cumulative reputation score in our models.

  14. 14.

    We reran the model measuring vendors’ affordability as the average price per gram of drug for a vendor (e.g. average of $40 for a 1 g drug transaction), encountering the same results: vendors’ reputation was positively correlated (p < 0.001) with making a sale and affordability was non-significant. We elected to use the more parsimonious measure to restrict missing data in the model.

  15. 15.

    It is important to note here that only vendors can be isolates, as buyers could only enter the network after a connection (transaction) has been made. Thus, there are 57 vendors total, but only 44 vendors who have made a sale in the last 6 months.

  16. 16.

    This may be due to vendors using different accounts to purchase. However, we have found no source, scholarly or otherwise, to corroborate this.

  17. 17.

    Technically, bipartite networks are not directed. However, we continue to use the language of indegree/outdegree for the sake of consistency.

  18. 18.

    All network statistics provided are calculated by treating the network as bipartite, with the exception of transitivity, which cannot exist in a bipartite network.

  19. 19.

    We reran this model without the mean transaction variable—the only variable influenced by missing data—with little difference in statistical significance, standard errors, and coefficient size.

  20. 20.

    Prior research has highlighted that many vendors list their geographic location as ‘worldwide’ to avoid identification. Eight of our 57 vendors were listed as worldwide. This information was not included in our vendors’ location variable. As one reviewer pointed out, there may be self-selection among vendors who do not list their country of origin.

  21. 21.

    While unintuitive given our descriptive findings, this merely reflects that the only buyer level variable in our model is their degree score, which tends to be significant if no other controls are included in the model (Lusher and Ackland 2011), and that all buyers in our network have a degree score >1.


  1. Aldridge J, Décary-Hétu D (2014) Not an ‘Ebay for Drugs’: The cryptomarket ‘Silk Road’ as a paradigm shifting criminal innovation. doi:10.2139/ssrn.2436643

  2. Aldridge J, Decary-Hetu D (2016) Hidden wholesale: the drug diffusing capacity of online drug cryptomarkets. Int J Drug Policy. doi:10.1016/j.drugpo.2016.04.020

    Article  Google Scholar 

  3. Andresen M, Felson M (2010) The impact of co-offending. Br J Criminol 50(1):66–81

    Article  Google Scholar 

  4. Baker WE, Faulkner RR (1993) The social organization of conspiracy: illegal networks in the heavy electrical equipment industry. Am Sociol Rev 58:837–860

    Article  Google Scholar 

  5. Barratt MJ (2012) Silk road: EBay for drugs. Addiction 107(3):683

    Article  Google Scholar 

  6. Barratt MJ, Aldridge Judith (2016) Everything you always wanted to know about drug cryptomarkets* (*but were afraid to ask). Int J Drug Policy 35:1–6

    Article  Google Scholar 

  7. Barratt MJ, Lenton S, Allen M (2013) Internet content regulation, public drug websites and the growth in hidden Internet services. Drugs Educ Prevent Policy 20:195–202

    Article  Google Scholar 

  8. Barratt Monica J, Ferris Jason A, Winstock Adam R (2014) Use of silk road, the online drug marketplace, in the United Kingdom, Australia and the United States. Addiction 109:774–783

    Article  Google Scholar 

  9. Barratt MJ, Lenton S, Maddox A, Allen M (2016a) ‘What if you live on top of a bakery and you like cakes?’ Drug use and harm trajectories before, during, and after the emergence of the Silk Road. Int J Drug Policy. doi:10.1016/j.drugpo.2016.04.006

    Article  Google Scholar 

  10. Barratt MJ, Ferris JA, Winstock Adam R (2016b) Safer scoring? Cryptomarkets, social supply and drug market violence. Int J Drug Policy. doi:10.1016/j.drugpo.2016.04.019

    Article  Google Scholar 

  11. Caulkins J, Reuter P (2010) How drug enforcement affects drug prices. Crime Justice 39(1):213–271

    Article  Google Scholar 

  12. Coomber R (2004) Drug use and drug market intersections. Addict Res Theory 12(6):1–5

    Article  Google Scholar 

  13. de Bie JL, de Poot CJ, Freilich JD, Chermak SM (2017) Changing organizational structures of jihadist networks in the Netherlands. Social Networks. 48:270–283

    Article  Google Scholar 

  14. Decary-Hetu D, Laferriere D (2015) Discrediting vendors in online criminal markets. In: Maim Ali, Bichler Gisela (eds) Disrupting criminal networks: network analysis in crime prevention. Lynne Rienner, Boulder

    Google Scholar 

  15. Decary-Hetu D, Paquet-Clouston M, Aldridge J (2016) Going international? Risk taking by cryptomarket drug vendors. Int J Drug Policy 35:69–76

    Article  Google Scholar 

  16. Diekmann A, Jann B, Przepiorka W, Wherli S (2014) Reputation formation and the evolution of cooperation in anonymous online markets. Am Soc Rev 79:65–85

    Article  Google Scholar 

  17. Dolliver DS (2015) Evaluating drug trafficking on the Tor Network: Silk Road 2, the sequel. Int J Drug Policy 26:1113–1123

    Article  Google Scholar 

  18. Dolliver DS, Kenney JL (2016) Characteristics of drug vendors on the Tor network: a cryptomarket comparison. Vict Offenders 11(4):600–620

    Article  Google Scholar 

  19. Dupont B, Cote A, Savine C, Decary-Hetu D (2016) The ecology of trust among hackers. Glob Crime 17(2):129–151

    Article  Google Scholar 

  20. Eurobarometer (2014) Young people and drugs: Results per country. Retrieved from

  21. Hunter DR, Goodreau SM, Handcock MS (2008) Goodness of fit of social network models. J Am Stat Assoc 103(481):248–258

    Article  Google Scholar 

  22. Kennedy D (2008) Deterrence and crime prevention: reconsidering the prospect of sanction. Routledge, London

    Google Scholar 

  23. Knoke D, Yang S (2008) Social network analysis, 2nd edn. Sage, Thousand Oaks

    Google Scholar 

  24. Kreager DA, Schaefer DR, Bouchard M, Haynie DL, Wakefield S, Young J, Zajac G (2016) Toward a criminology of inmate networks”. Justice Q 33:1000–1028

    Article  Google Scholar 

  25. Krebs V (2001) Mapping networks of terrorist cells. Connections 24(3):43–52

    Google Scholar 

  26. Lusher D, Ackland R (2011). A relational hyperlink analysis of an online social movement. J Soc Struct 12(5):1–35

    Google Scholar 

  27. Lusher D, Koskinen J, Robins G (2013) Exponential random graph models for social networks: theory, methods, and applications. Cambridge Press, Cambridge

    Google Scholar 

  28. Morselli C, Giguere C, Petit K (2007) The efficiency/security trade-off in criminal networks. Soc Netw 29(1):143–153

    Article  Google Scholar 

  29. Natarajan M (2006) Understanding the structure of a large heroin distribution network: a quantitative analysis of qualitative data. J Quant Criminol 22:171–192

    Article  Google Scholar 

  30. Newman MEJ (2006) Modularity and community structure in networks. Proc Natl Acad Sci 103:8577–8582

    Article  Google Scholar 

  31. Newman MEJ (2010) Networks: an introduction. Oxford University Press, Oxford

    Google Scholar 

  32. Newman MEJ, Girvan M (2004) Finding and evaluating community structure in networks. Phys Rev E 69:026113

    Article  Google Scholar 

  33. Papachristos A (2009) Murder by structure: dominance relations and the social structure of gang homicide. Am J Sociol 115:74–128

    Article  Google Scholar 

  34. Papachristos A (2014) The network structure of crime. Sociol Compass 8(4):347–357

    Article  Google Scholar 

  35. Papachristos A, Hureau D, Braga A (2013) The corner and the crew: the influence of geography and social networks and gang violence. Am Soc Rev 78(3):417–447

    Article  Google Scholar 

  36. Pons P, Latapy M (2005) Computing communities in large networks using random walks. In: Physics and society. arXiv:physics/0512106v1

    Google Scholar 

  37. Raab J, Milward BH (2003) Dark networks as problems. J Public Adm Res Theor 13(4):413–439

    Article  Google Scholar 

  38. Robins G, Pattison P, Woolcock J (2004) Missing data in networks: exponential random graph (p*) models for networks with non-respondents. Soc Netw 26(3):257–283

    Article  Google Scholar 

  39. Robins G, Pattison P, Kalish Y, Lusher D (2007) An introduction to exponential random graph (p*) models for social networks. Soc Netw 29:173–191

    Article  Google Scholar 

  40. Schaefer DR, Rodriguez N, Decker S (2014) The role of neighborhood context in youth co-offending. Criminology 52(1):117–139

    Article  Google Scholar 

  41. Schaefer DR, Bouchard M, Young JTN, Kreager DA (2017) Friends in locked places: an investigation of prison inmate network structure. Soc Netw. doi:10.1016/j.socnet.2016.12.006

    Article  Google Scholar 

  42. Smith CM, Papachristos AV (2016) Trust thy crooked neighbor: multiplexity in Chicago organized crime networks. Am Sociol Rev 81(4):644–688

    Article  Google Scholar 

  43. Snijders TA (2002) Markov chain Monte Carlo estimation of exponential random graph models. J Soc Struct 3(2):1–40

    Google Scholar 

  44. Soska K, Christin N (2015). Measuring the longitudinal evolution of the online anonymous marketplace ecosystem. In: Proceedings of the 24th Usenix security symposium

  45. Stafford MC, Warr M (1993) A reconceptualization of general and specific deterrence. J Res Crime Delinq 30:123–135

    Article  Google Scholar 

  46. Stephen A, Toubia O (2009) Explaining the power-law degree distribution in a social commerce network. Soc Netw 31(4):262–270

    Article  Google Scholar 

  47. Tenti V, Morselli C (2014) Group co-offending networks in Italy’s illegal drug trade. Crime Law Soc Change 62:21–44

    Article  Google Scholar 

  48. Tremblay P (1993) Searching for suitable co-offenders. In: Clarke RV, Felson M (eds) Routine activity and rational choice. Transaction Books, New Brunswick, pp 17–36

    Google Scholar 

  49. Tzanetakis M, Kamphausen G, Werse B, von Laufenberg R (2016) The transparency paradox. Building trust, resolving disputes, and optimizing logistics on conventional and online drug markets. Int J Drug Policy 35:58–68

    Article  Google Scholar 

  50. United Nations Office on Drugs and Crime (2016) World drug report.

  51. Van Buskirk J, Roxburgh A, Bruno R, Sundresan N, Lenton S, Sutherland R, Whittaker E, Sindicich N, Matthews A, Butler K, Burns L (2016) Characterising dark net marketplace purchasers in a sample of regular psychostimulant users. Int J Drug Policy. doi:10.1016/j.drugpo.2016.01.010

    Article  Google Scholar 

  52. Van Hout MC, Bingham T (2013) ‘Silk Road’, the virtual drug marketplace: a single case study of user experiences. Int J Drug Policy 24:385–391

    Article  Google Scholar 

  53. Van Hout MC, Bingham T (2014) Responsible vendors, intelligent consumers: Silk Road, the online revolution in drug trading. Int J Drug Policy 25:183–189

    Article  Google Scholar 

  54. Von Lampe K, Johansen PO (2004) Organised crime and trust: on the conceptualization of trust in the context of criminal networks. Glob Crime 6:159–184

    Article  Google Scholar 

  55. Walsh C (2011) Drugs, the internet, and change. J Psychoact Drugs 43(1):55–63

    Article  Google Scholar 

  56. Wang C, Butts CT, Hipp JR, Jose R, Lakon C (2016) Multiple imputation for missing edge data: a predictive evaluation method with application to Add Health. Soc Netw 45:89–98

    Article  Google Scholar 

  57. Wasserman S, Faust K (1994) Social network analysis: methods and applications. Cambridge University Press, Cambridge

    Google Scholar 

  58. Weerman F (2003) Co-offending as social exchange. Br J Criminol 43(2):398–418

    Article  Google Scholar 

  59. Wood G (2016) The structure and vulnerability of a drug trafficking collaboration network. Soc Netw 48:1–9. doi:10.1016/j.socnet.2016.07.001

    Article  Google Scholar 

Download references


We would like to thank Benjamin Gilbert for assistance in data coding, and David Melamed for helpful methodological suggestions.

Author information



Corresponding author

Correspondence to Scott W. Duxbury.



In line with Hunter et al. (2008), we evaluate the goodness of fit of ERGM by comparing a distribution of degree statistics from networks simulated from ERGM parameters to the degree statistics of the empirical network. Figure 3a indicates how well the simulated networks match the degree score of the empirical network. Figure 3b indicates how well ERGM coefficients predict the degree scores observed in the empirical network.

Fig. 3

a Goodness of fit for ERGM, degree. b Goodness of fit for ERGM, log-odds of degree

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Duxbury, S.W., Haynie, D.L. The Network Structure of Opioid Distribution on a Darknet Cryptomarket. J Quant Criminol 34, 921–941 (2018).

Download citation


  • Drug distribution
  • Online drug markets
  • Social networks
  • Trust
  • Tor network