Every so often a headline claims the dark web is shrinking: fewer onion sites exist than before, the network is dying out. These claims usually come from counting exercises that sound precise but are built on shaky measurements. If you rely on directory links to find anything on Tor, understanding why these counts are unreliable matters more than the headline number itself.
What these surveys actually count
Most "how big is the dark web" studies count onion addresses that show up in some kind of index, crawl, or request log. That sounds straightforward. It is not.
The dark web is a small part of the broader deep web, the portion of the internet not indexed by ordinary search engines. Onion services do not register themselves in any central, public list. Anyone trying to count them has to infer activity from indirect signals, and those signals behave in strange ways.
Why visit counts do not equal site counts
The Tor Project itself has explained one of the clearest traps in this kind of measurement. A Tor client requests a hidden service directory record the first time it visits an onion address it has not seen in a while. Spend hours on one site and you generate roughly one such request. Spend one second each on a hundred different sites and you generate roughly a hundred.
A small number of people who hop between many sites in a single session can dominate the data, making the network look far more or far less populated than it really is. A survey that treats directory requests as a proxy for "number of active sites" can be badly skewed by a handful of unusual users.
Link rot does the rest of the damage
Even when a count is taken honestly, it goes stale fast. Studies of ordinary web links found that about one in every 200 links breaks each week, giving a half-life of roughly 138 weeks. A widely used web directory that stopped being updated saw its links reach a half-life of about two years.
Onion addresses are more fragile. Many sites are run by individuals or small groups with no obligation to keep an address stable, no uptime guarantee, and no forwarding system when they move or shut down. A directory snapshot from six months ago can already be mostly dead links by the time you read it.
Duplication inflates the numbers too
The flip side of link rot is duplication. When a popular site disappears or gets seized, clones and copycats tend to spring up using similar names, similar layouts, or scraped content. Some are run by the original operators trying to recover. Many are not.
A crawler counting unique onion addresses has no reliable way to tell a genuine new service from a clone, a scam mirror, or an abandoned fork of an old site. Raw counts get inflated in one direction while real churn deflates them in another. The "size" of the dark web, measured this way, is really measuring noise.
This is exactly why directory pages are risky
This matters most for readers, not researchers. If researchers with crawling tools and statistical methods still struggle to tell a live, legitimate onion site from a dead or cloned one, an ordinary visitor clicking through a list of links is in a much weaker position.
Any general-purpose "Hidden Wiki" style directory lacks a single trustworthy version. What circulates under that name today is a sea of clones and copycats, and many are stuffed with outdated, broken, scam, or outright dangerous links. A page looking established or well-organized is not evidence that its links are safe or even still active.
What to do instead
The safest way to find a legitimate onion address is to get it directly from an organization you already trust on the ordinary web, then verify it against that organization's official site. A newsroom, a privacy nonprofit, or a service you already use is a far better source than an anonymous directory page.
- Search for the organization's official surface-web page first, using a search engine you already trust.
- Check whether that organization publishes its own onion address on its official site, verified channel, or documentation, such as how the Tor Project documents its own services.
- Treat any .onion link found only on a third-party list as unverified until you can confirm it another way.
- Use Tor Browser as intended: it protects how your traffic is routed, not what you choose to click.
Keep anonymity expectations realistic
Tor routes your traffic through multiple relays, but it does not make any action you take automatically anonymous. Logging into a personal account, reusing a username, or downloading a file through Tor can still identify you, regardless of how the network itself is measured or sized.
No single tool, directory, or survey number is a safety guarantee by itself. Scepticism about "how big is the dark web" headlines is an extension of the same habit worth applying to any link you are about to click: verify the source, do not trust the list.