Yonatan Sompolinsky said competitive sequencing means making miners compete to decide the order of transactions, rather than letting one participant control that order for a round. His argument for moving from 10 to 100 blocks per second was not mainly about faster confirmations. It was about increasing competition among the participants doing the ordering. [1]
Why transaction order matters
A sequencer is a participant that chooses which transactions go into a block and in what order. In his AusCryptoCon 2024 talk, published on 21 May 2025, Sompolinsky contrasted competition to produce a block with competition over transaction ordering: many miners may compete, but in a leader-based system one winner decides the order for that round. [1]
That temporary control matters for financial transactions. Sompolinsky named front-running, back-running, sandwiching and temporary censorship as ways an ordering participant can exploit its position. These involve placing transactions before or after someone else’s transaction, surrounding it with other transactions, or delaying its inclusion. He argued that people whose outcomes depend strongly on ordering would benefit from competition within the round itself. [2]
Why he wanted 100 blocks per second
Sompolinsky introduced a measure called SPR, to ask how many sequencing participants compete during each consensus round. His focus was competition within a round, not simply how often the network starts another one. [1]
Kaspa’s blockDAG, a structure that accommodates blocks created in parallel, provides the basis for this idea. In the talk, he showed a slice containing three miners’ parallel blocks: they had created their blocks without yet seeing one another’s work. He used that example to explain how multiple miners could participate in sequencing at the same time. [2]
He said 10 blocks per second could already feel sufficiently responsive to users. Raising that to 100 might make little noticeable difference to their waiting time, while making a large difference to sequencing competition. The 100-block figure was an argument for a design direction in that talk, not an announcement of an activated upgrade. [1][2]
The proposed auction mechanism
Sompolinsky said he was working with Aviv Yaish on a mechanism that would make parallel miners compete for users’ transactions. He presented it as research in progress. [3]
His example worked like this:
- A miner identifies value it could gain from ordering a user’s transaction and offers to return part of that value to the user. [4]
- Another miner offers the user a larger return. [4]
- The mechanism uses these competing bids to help decide which miner should sequence the transaction. [4]
He compared the idea with first-price and second-price auctions, but explicitly warned that those simple models would not work directly: a practical mechanism would need more sophisticated methods. He argued that more competing participants could improve the outcome for users. [4]
He also suggested that, in his hypothetical 100-block-per-second setting, reaching even seven or eight percent of sequencing participants could provide useful competition. That was an illustration of the proposed mechanism, not an instruction for using a deployed service. [4]
A further application: outside data
Sompolinsky extended the idea to oracles, services that report outside information to a blockchain. He argued that parallel participation could support competition among data providers, helping reject outliers and combine reports. He described this as ongoing research, not an operating guarantee. [5]
Watch his AusCryptoCon explanation of competitive sequencing, starting at 12:44, then the auction example at 19:36. [1][4]