Publishing a hidden service and promoting it are two separate decisions. A Tor onion service or I2P site can be reachable to anyone who has its address while remaining absent from public directories, search tools, proxies, and clearnet announcements. That distinction gives a publisher useful control: the site can begin with a small, invited audience and add discovery channels only when the benefits justify the added exposure.

This is not secrecy by default. Addresses can be copied, visitors can share them, and a public service should never rely on obscurity as its only protection. It is better to think of discovery as a set of doors. Each door helps a different audience enter, creates a different maintenance burden, and reveals a different amount of information about the project. Good publishing starts by deciding which doors belong in the plan.

Reachability and discoverability are different

A service is reachable when a visitor who already has the correct address can connect to it. A service is discoverable when a visitor can find that address through another channel. Conventional websites tend to bundle those ideas together: domain registration, search indexing, social links, and browser navigation all point toward the same public name. Hidden networks separate them more sharply.

Tor onion addresses are derived from cryptographic identity material and are used only through the Tor network. I2P destinations are also cryptographically addressed, while human-readable .i2p names are resolved through naming systems and address books. I2P documentation notes an important consequence: names are locally meaningful rather than guaranteed to be globally unique. Two people can have the same readable name mapped to different destinations in their local address books. That makes the source of a name or address part of the trust decision.

A publisher therefore needs an authoritative way to announce the address. It may be a signed statement, an established account, a trusted newsletter, a printed card, a known clearnet page, or a direct message through an already verified channel. A directory entry can help people discover a service, but it should not be the only proof that an address is genuine. A copied listing can be outdated, mistaken, or malicious even when the underlying network is working correctly.

Tor’s Onion-Location mechanism illustrates the difference well. A conventional website can announce that it has an onion counterpart by returning a header or page tag that Tor Browser recognizes. This gives visitors a visible bridge from a known clearnet origin to its onion service. It is useful when the publisher wants that relationship to be public. It is not mandatory, and it would be the wrong choice for a project whose clearnet and hidden-network identities are intentionally separate.

Four discovery layers, four different tradeoffs

The first layer is direct distribution. The publisher gives the address to selected people or posts it in a limited community. This is simple and keeps promotion narrow, but it makes onboarding and address changes harder. Visitors need a reliable way to distinguish the current address from an old or impersonated one. Direct distribution works best when the publisher maintains at least one durable announcement channel and explains how changes will be authenticated.

The second layer is a directory or index. A directory can add a title, description, category, screenshot, reputation signals, and search. That context lowers the effort required to find a useful service. It also places the service in a browsable public catalog, where its name, topic, and address can be copied or monitored. A publisher should check how a directory verifies updates, handles complaints, labels unavailable sites, and removes listings. Submission should be intentional; discovery should not be treated as an automatic consequence of hosting.

The third layer is proxy or gateway access. A gateway may make hidden-network content reachable to people who are not running the original network software. That can substantially widen the audience, especially for education, previews, or public-interest material. It also changes the route and the trust relationship. The visitor is no longer making the same end-to-end journey as a native Tor or I2P visitor. The gateway may process requests, headers, caching, abuse controls, and logs according to its own design and policy. Publishers should describe gateway access accurately and avoid presenting it as identical to native access.

The fourth layer is a conventional clearnet counterpart. A project may publish the same material on both networks, use a clearnet landing page to announce a hidden service, or maintain two different editions. This offers familiar access and can improve ordinary search visibility, but it creates another hosting, DNS, certificate, analytics, and complaint surface. It can also make two identities easy to correlate. That may be acceptable for a public organization and unacceptable for a deliberately separated pseudonymous project.

These layers can be combined. A public documentation project might use direct announcements, Onion-Location, a directory listing, and a clearnet edition. A small community site might distribute its address privately and decline all other promotion. Neither pattern is universally more private or more responsible. The right pattern follows from the audience, threat model, content, support capacity, and desired relationship between identities.

Build an opt-in discovery ladder

A useful way to plan is to create a discovery ladder before launch. Begin with the narrowest channel that reaches the intended readers, then define what evidence would justify the next rung. For example:

  1. Verified direct address: establish one authoritative announcement channel and a method for communicating address changes.
  2. Community sharing: allow trusted groups or partners to share the address with clear, current descriptive copy.
  3. Directory listing: opt into a catalog after reviewing its verification, update, moderation, and removal practices.
  4. Gateway or proxy: add easier access when the larger audience is worth the different trust and logging boundary.
  5. Clearnet promotion: publicly connect identities only when that connection supports the project’s goals.

For every rung, record an owner, a purpose, and a removal plan. Who updates a listing when the title changes? Who confirms that a screenshot does not reveal information the site itself no longer shows? What happens if an address rotates? How will visitors learn that an old address has been retired? Promotion creates operational work, and abandoned discovery records can be more confusing than having no record at all.

Also separate audience growth from security. A hard-to-find service still needs authentication, software updates, access controls, backups, and an abuse process appropriate to what it publishes. Conversely, a widely promoted service is not automatically unsafe if it was designed for public traffic. Discovery settings manage visibility; they do not repair an insecure application or guarantee the identity of every visitor.

The Invisible-Internet services guide documents the intended division: Darknet Now is the privacy-first creation and publishing layer, while Darknet Proxy is the discovery, access, and optional promotion layer. A publisher should be able to bring a site online without being enrolled automatically in broader clearnet exposure. Promotion belongs to a separate, informed choice, with clear information about what route is being added and how it changes the audience. The related article Hidden-Service-Only or Clearnet Too? covers the earlier visibility decision.

Practical Takeaway

Write a one-page discovery policy for the project. Name the intended audience, the authoritative source for the address, the channels currently approved, and the conditions for adding or removing each channel. Treat directory listings and proxy access as maintained publications, not one-time submissions.

Before adding a new route, ask four questions: Does it reach people the project actually wants to serve? Does it publicly connect identities or infrastructure that were meant to remain separate? Who operates the intermediary, and what changes about visitor trust? Can the publisher keep the record accurate and respond to problems there?

The goal is not to make a public site impossible to find. It is to prevent accidental exposure from becoming the default. A deliberate ladder lets a project widen access gradually while preserving the ability to say where, how, and for whom it is being promoted.

Sources and Further Reading