University College London report: Bitcoin system is inefficient, but the 1MB block limit is not the main reason (full text download)

A research report from the University College London (UCL) Blockchain Technology Center (download the full text of the Coin Library) shows that "43% of transactions take more than an hour to be written to the blockchain after being sent, and 20% of transactions cannot be confirmed even after 30 days, which also proves the inefficiency of the Bitcoin system."

The UCL study, which examined the bitcoin network over a three-month period, found that 12,000 nodes had improved efficiency when processing large transactions.

The report mentioned:

We found that the processing speed of this process (transaction confirmation) is still slow, but most transactions (93%) can be completed within 3 hours, and only 0.1% of transactions have not been confirmed after 30 days.

The study came to some interesting conclusions. For example, in one week, the number of new blocks reached nearly 200,000, but only 2,000 of them were "real" or meaningful. UCL assistant researcher Giuseppe Pappalardo, one of the authors of the report, explained this phenomenon to us:

When a node receives a block, it must verify all transactions in the block and the block itself. If both the transactions and the block are valid, the node will announce this result to other nodes using the INV message. Therefore, when a node sends a message about a block or transaction using the INV message, it means that a block has passed verification.

The reason why nodes send block information is that not all nodes hold a complete blockchain record, so some nodes cannot verify blocks, or it is possible that the function of these nodes is not to verify blocks at all.

Without proper verification, old blocks will just be broadcasted among nodes in a cycle.

This is clearly evidence of the inefficiency of the Bitcoin system, as many resources are wasted. Regarding transactions, it is unclear whether UCL has studied the size of fees and the role they play in transaction processing, the authors simply note the difference between small and large transactions.

The Bitcoin system cannot guarantee accurate transaction processing times, as some transactions take months to confirm. We found that this is not due to the 1MB block size limit or the fact that each block can only contain a few thousand transactions. The network does not seem to be saturated at the moment, with an average block size of 0.8MB, only 3% of blocks exceeding 0.99MB, and some blocks not containing any transactions at all.

The study began in May last year, and soon after, Blockchain.info said the average daily block size had reached 0.96MB. Since last year, transaction delays have become common, with more than 2 million bitcoins ($2.5 billion) currently waiting to be confirmed.

Inefficient transaction processing

However, the report states that regardless of whether there is a limit on block size, “there is no guarantee that transactions will be processed in a timely manner” because miners are free to choose the transactions that go into a block.

In order to ensure that transactions are processed in a timely manner, Satoshi Nakamoto did propose a mechanism, but it is not mandatory and does not belong to the protocol layer. This mechanism is called first seen. Nodes and miners should process the first transaction they see, and use this method to arrange the order of transactions, but as the block capacity continues to increase, this method has become ineffective. When the block capacity remains at a normal level, this mechanism is not mandatory, but most miners will still use this method because of its rationality.

First seen also makes double spends more difficult (although some miners may still not use first seen), even when the transaction is not confirmed. This method allows Bitcoin transactions to be processed instantly in most cases.

It’s unclear whether the authors of this study were aware of the existence of this mechanism (which has long been replaced by a fee-based prioritization approach). Pappalardo believes that further investigation is needed to determine why some transactions are delayed in confirmation:

We believe that miners do not receive additional economic incentives for including all transactions, so some transactions are ignored. After a period of time, miners are less likely to be willing to process old transactions.

Pappalardo further stated, “Despite the persistent discrediting of the efficiency of the Bitcoin network, the UCL Blockchain Centre is a firm believer in the potential of blockchain and its consensus mechanism. The purpose of this report is to call for the right economic incentive system to create an efficient P2P blockchain system that records transactions accurately and in a timely manner.”

So what kind of economic incentive system is efficient? Pappalardo answered, "A system that does not provide economic incentives for reasonable transaction processing and timely record keeping cannot be efficient."

Block capacity dispute

The block capacity dispute has been going on for two years. Some people believe that the 1MB block cap should be maintained, and the efficiency of the Bitcoin basic protocol should be low, so they choose to support the Lightning Network (LN). However, others believe that the first seen mechanism should be restored by directly increasing the block capacity cap.

Those who insist on inefficiency continue to block any proposed changes to the protocol, including the recent extension blocks proposal. This behavior has ultimately led to nearly half of Bitcoin transactions waiting for more than an hour to be confirmed, and some research has begun to reveal the inefficiency of the Bitcoin payment network.

Whether the inefficiencies of the Bitcoin network can ultimately be resolved remains to be seen, but we welcome any independent research based on scientific facts and a fair attitude that can help end the block size debate.