A crowd you can hide in, from day one.
The network is not live: no miners, no gateways, no contracts, no nodes. v0 runs locally in your browser in the Noise Lab. There is no token.
agreed; changes are announced
on the road, not built yet
part of the design, can change
open question or not checked yet
zkSI is a design for private SI requests based on noise mining. CONCEPT
A fixed volume of traffic hides the real requests. When real traffic is low, miners fill the gap with decoy requests. When real traffic grows, the decoys shrink. CONCEPT
The problem it targets: a private network is only as private as its crowd. With 50 users, there is no crowd. zkSI pays for the crowd until real users replace it. CONCEPT
A fixed target. Miners fill the gap.
The network aims for a fixed target volume, for example 1,000 requests per minute. The real value is not decided. CONCEPT
Miners deliver the difference between the target and real traffic, and are paid for that volume. CONCEPT
| real requests / min | decoys from miners | noise share |
|---|---|---|
| 50 | 950 | 95% |
| 500 | 500 | 50% |
| 900 | 100 | 10% |
Noise matters most at the start and falls on its own as the network grows. What happens when real traffic goes above the target is not decided. [VERIFY]
Cheap at the network. Expensive at the model.
| NZ.NET · network noise | NZ.END · endpoint noise | |
|---|---|---|
| what it is | Fake packets from client into the mixnet | Decoy requests that reach the model |
| where it ends | Dropped at the last node | Answered by the SI provider |
| cost | Cheap, no inference | Expensive, real inference |
| who it confuses | Network observers | The SI provider |
| status | CONCEPT | CONCEPT |
The Noise Pool pays mainly for endpoint noise, because that is where the cost is. CONCEPT
In v0, each user runs a local decoy generator and pays for their own noise. Requests use fixed size classes so size does not reveal which ones are real. PLANNED
Decoys only work if they look like real requests. A generator that does not copy real request patterns can be detected. See Risks. CONCEPT
Pick a seat. Watch the traffic.
| party | sees | should not see (design goal) |
|---|---|---|
| Your client | Your real requests | - |
| First mixnet node | Your IP, that you send traffic, when | Content, which packets are real |
| Last mixnet node | Traffic leaving the mixnet | Your IP |
| Gateway | Request content, audit nullifiers | Your IP. Whether it can tell real from decoy: [VERIFY] |
| SI provider | Content of every request and response, real and decoy | Your IP, which requests are real |
| Miner | Only its own decoys | Your requests |
| Chain observer | Stakes, commits (hashes), audit results, payouts | Any request content |
All rows: CONCEPT. The exact view of each node depends on the mixnet used, which is not decided. [VERIFY]
What zkSI does not hide
The SI provider still reads the content of your real request. If your prompt contains your name, your email or text that only you would write, noise does not protect you. zkSI hides which request is yours, not what it says. CONCEPT
> next: threat model, proof of noise, noise pool, cost, cryptography, verification, risks, roadmap, disclosure, faq. in progress.