Approach and methodology

It is difficult to differentiate all transactions on a blockchain based on their end-use. Therefore, we have focused on transactions that we know were for exchange, teller, and payment processing services to analyse a more accurate data sample.

We analysed a sample of 1.2 million transactions across our client base, which includes payment processors and exchanges, which produced the following statistics:

0-conf throughput

+/- 0.1% of transactions GAP600 processed failed to reach the blockchain. This figure has generally remained constant throughout our operational experience in the industry. These represent transactions which in essence were fraud attempts and were initially published as desired transactions but failed essentially at realizing the double-spent attempt. 

While this statistic may appear to be minimal and insignificant, all businesses operating in the crypto space have a low tolerance to fraud. In addition, when a vulnerability is discovered, it is heavily exploited. As such, proactive prevention is vital. Actual double-spend numbers would be much higher if there were no mitigation measures in place.

To contextualise this figure, it is important to understand the transaction process. Every new Bitcoin tx comprises previous transactions, which creates a chain link. The GAP600 system performs a backward analysis of these chain links to identify potentially suspicious transaction ancestors or inputs that could indicate an attempted fraudulent double-spend attempt.

Another interesting statistic based on historical analyses performed by the GAP600 system indicates that only 1.5% of transactions had transaction ancestors or inputs that were too suspicious to confirm instantly. 

While suspicion does not necessarily imply a nefarious purpose, the presence of these factors places numerous transactions outside GAP600’s parameters. They are, therefore, considered too high risk to guarantee.

Tx speed

The time we save on the platform varies according to stress periods on the network. For instance, significant backlogs still occur periodically on the BTC network. On average, it took 15 minutes for transactions from the sample set to be included in a block and be considered as almost final.

Payment processors generally wait for 1 block confirmation, while most crypto exchanges wait for between 3-6 block confirmations before processing a transaction. This means transactions can take between 15 minutes to 1h05 to complete, but certain transactions have taken a day or more depending on network stress conditions. 

A simple solution is just to wait the equivalent of 3-6 confirmations on the BTC network, but this obviously degrades the experience and fails to meet the demand for instant transactions.

Double spend

As previously stated, GAP600’s experience indicates 1.5% of transactions include ancestors or inputs that are too suspicious to confirm, but not every suspicious input indicates a double-spend attempt.

When GAP600 does not confirm these transactions for ancestral or other reasons, there is no incentive for the attacker to follow through with the second transaction, which is required to confirm if the first transaction is a double-spend attempt. As such, in many cases, there are no attempted or completed counter double-spend transactions after the initial transaction is not confirmed by the platform.

In addition, the sensitivity around money, especially digital money, necessitates that responsible services providers mitigate risk wherever possible. This ensures that they generally have clear control of their revenue streams, especially when offering this as a commercial service. 

GAP600 has in the past guaranteed 0-conf transactions which failed to reach the blockchain. In each case, we perform a rigorous analysis to determine the causes and implement changes on a granular level to ensure we don’t experience them again.