“Kaspa: Mining the Internet” is the title of Yonatan Sompolinsky’s December 2025 talk at Tokenize London. It names a proposed role for Kaspa miners: helping the network agree on information from outside the blockchain and act on it quickly. [1][2] In the talk, he describes an internet money system that can receive outside information and carry out a user’s instructions in response. He presents this as a proposed flow, not an already operating service. [1][2]
Why reach beyond the blockchain?
Yonatan draws a distinction between checking a calculation and checking its input. Cryptography can establish that a transaction has a valid signature or that a calculation follows its rules. But that does not, by itself, establish whether an event outside the blockchain actually happened. Connecting a blockchain to the wider world introduces another kind of trust. [3]
In his explanation, that connection needs an honest-majority assumption: confidence that most of the participants responsible for reporting or agreeing on the information are honest. His aim is to use proof-of-work mining, rather than a single service, as the basis for that agreement. This does not turn an outside claim into a mathematical proof. [3][4]
The flow he describes has three basic steps:
- A person defines an instruction that depends on an outside event.
- The proposed internet money system receives information about that event.
- It carries out the instruction in response. [5][1]
The important idea is not the particular event in his example. It is connecting information, agreement and action in one coherent process, rather than leaving the person dependent on separate services to complete each step. [1][2]
Why Kaspa mining matters to this idea
Yonatan describes the central challenge as how quickly the system can sample an honest majority. Mining produces repeated opportunities to gather evidence of agreement. More frequent samples could let the network reach a sufficiently confident decision about an outside report sooner. [4]
In the talk, he illustrates this with a hypothetical rate of 100 blocks per second: half a second could provide 50 votes. That is an illustration of his proposed direction, not a statement that this rate or the internet-information service was operating. His point is that more frequent mining could help reach a verdict within the time of one internet request and response. [4]
A lesser-known detail is that he does not frame this mainly as making a payment feel faster to a person. In that passage, he says the more important gain would be quickly settling an outside-information report. The intended benefit is faster decentralized agreement on outside information, not a payment that merely feels faster. [4]
What miners would do, and what they would not do
Yonatan explicitly says miners would not be forced to participate in the proposed service. They would opt in because they had an incentive to do so. In the December 2025 discussion, he also suggests that competition over which user requests get handled first could provide additional revenue for participating miners. These are elements of the proposal he described then. [2][6]
He proposes integrating the flow into Kaspa’s base network. His argument is that information-dependent actions involving assets and logic on the same network could happen together, without a separate handoff. He calls this benefit “atomicity”: treating linked actions as one coordinated operation. [2]
This should not be confused with miners automatically knowing what is true everywhere online. Yonatan stresses that outside information requires an honest-majority trust model. The proposal concerns agreement on reports and action based on them, not cryptographic proof of every real-world fact. [3][4]
The wider purpose
In his Oxford Union address of March 2026, published on hashd.ag, Yonatan argues that the internet lets people communicate but gives them too few ways to act together: “The internet currently lacks methods to act on agreement.” [7] His later “Project Staghunt” essay proposes an association layer for communities to act on shared interests, and says it must run on infrastructure that stays censorship-resistant, permissionless and fair in real time, which he calls real-time decentralization: the same property his December 2025 talk ties to Kaspa mining. [8][4]
For his explanation of mining’s role, watch “Kaspa: Mining the Internet” at 12:35, where he explains rapid majority sampling, then 17:06, where he describes the proposed integration and miners’ voluntary participation. [4][2]