As illuded to in the beginning, the initial aim was to examine the role of network effects in performance of anomaly detection. This was using the Bitcoin blockchain, however the dataset proved unmanageable, or at least unfeasible, to use within the scope of this dissertation. The role of network effects in detecting anomalous banking behavior, have been mentioned in academic literature, but not yet measured in a way in which findings are directly replicable – as they would be if researchers used a completely open dataset like the bitcoin transaction network. Currently the paradigm within fraud detection at banks dictates that each bank is to fend for themselves. If it was proven that there are significant network effects, and that larger networks disproportionately gives bigger banks better chances of spotting illicit activity, arguments could be made for a centralized solution. While it’s outlandish to claim that such a finding would completely upend the way anti-money laundering and fraud detection are carried out today, it would be a significant challenge to the present paradigm within this domain. Banks spend a lot of money on compliance, and especially AML/KYC procedures tare into the private lives of ordinary citizens. If nothing except from the metadata of transactions were needed to label it with a high degree of confidence, this could perhaps significantly limit the necessity of transaction monitoring at a bank level. If not banks, then who? For international transfers the SWIFT organization has been established. SWIFT (Society for Worldwide Interbank Financial Telecommunication) was established by a consortium of banks, as a cooperative, meaning that no one owns SWIFT. They are funded by fees from institutions who choose to be members and use the transaction system (SWIFT, 2023). A centralized organization, taking care of all transaction monitoring and reporting, may draw inspiration. By statute it could include representatives of financial institutions, consumers, and the authorities, to ensure that the transaction data is handled in manner which satisfies all parties and doesn’t violate the privacy of individuals. While these imaginative reflections may be a little high minded, it does cement the reason why this was seen as an important area of research. Lexis Nexis made an estimate in 2017, that the total cost of AML compliance in Germany, France, Italy, the Netherlands, and Switzerland amounted to a combined USD 83.5 billion, when factoring in lost efficiency, customer friction and lost business (Lexis Nexis, 2017). Lexis Nexis obviously want to sell their ‘risk solutions’ to banks etc. so this figure may be exaggerated a bit. Though it is an illustration of the monetary incentive, for banks to find better solutions. On the other hand, there’s the authorities. What’s their incentive to participate in alternative solutions to the existing paradigm? According to EU, the VAT gap in all member countries were around EUR 90 billion in 2021 (European Commission, Directorate-General for Taxation and Customs Union, 2022). Europol estimates that up to EUR 50 billion of this may be directly caused by fraud (Europol, 2023). The exact method of fraud is called missing trader fraud – or missing trader inter-community (MTIC). This is a type of fraud where criminals benefit from the lack of cross-border cooperation, as cases become hard to follow when money move between jurisdictions (Europol, 2023). In a utopian world, where all transaction monitoring in all of Europe were combined, and could see these transactions, perhaps it could be caught before the money disappears. Leaving a potential incentive for the authorities too. But before dreaming up such a world, it would be nice knowing whether transaction monitoring would even benefit from such a paradigm shift. This question is still left for researchers to answer.