[metrics-team] How large is the Tor-network

[Iain Learmonth]

Hi Leon,

You've contacted a list that is primarily used for discussion about the
development of Tor Metrics software, and you'll probably get more
answers from the tor-project at lists.torproject.org mailing list instead.
I will however answer your questions.

On 26/02/2019 10:13, Leon Hoekstra wrote:
> I’ve read that the surface web is only 5% of the total internet. The deep web should be the largest part of the internet, people say. And a part of the deep web there is a layer thats been called the dark web with has a few different kind of networks like I2P, Freenet, Open bazaar and off course the largest network of these, the Tor-network. 
> 
> My question is, when the surface web should be 5%, how much of the 95% would be covered by the Tor-network?
My guess is that you've seen a picture of an iceberg alongside these
statistics? There is some truth to the idea of a surface web and a deep
web, and there's also a lot of sensationalism.

Let's start with some definitions:

The "surface web": Things you can Google, generally available to the
public via a browser. This excludes anything you need to log in to see.
Wikipedia would be in here along with many news outlets. Google search
results would be in here too, which is an interesting question: Do
Google search results pages that are dynamically generated count towards
the size of the surface web or is it only static content?

The "deep web": Anything you need to log in to see. This is all the
posts on Facebook, news articles behind paywalls, scientific journals,
records in your human resources system, medical databases, corporate
intranets, etc. There is a scale here with regard to how public you want
this stuff to be and there are very legitimate reasons for hiding some
content behind a login.

The "dark web": Anything you cannot access with an ordinary browser
would be a popular definition. This would include onion services,
corporate intranets that need you to connect via VPN, IoT networks, I2P,
Freenet, etc. Some of these offer anonymity properties, some strong
authentication and some just offer robust access.

The picture isn't so clear now. Depending on who you ask a corporate VPN
might be in the "deep web" category because you have to login to see it,
but I believe really it's in the "dark web" category because you need
specialized software to access it (the VPN client). You also have
Tor2Web instances which make Onion services available to the public
Internet so maybe they are just "deep web" and not "dark web".

Perhaps even the distinction is the wrong one to be making, an
oversimplification, and really we should be talking about the security
properties that are provided or required for access.

Let's get to some numbers:

Assuming that traffic consumed is a proxy measurement for "size"
~150GBit/s traffic is consumed daily in the Tor network
~1.7 GBit/s traffic is consumed daily in the Tor network for Onion services
~98.9% of total Tor traffic is for "surface web"
~1.1% of total Tor traffic is for Onion services

https://metrics.torproject.org/bandwidth-flags.html
https://metrics.torproject.org/hidserv-rend-relayed-cells.html

I would guess the most likely explanation for this number being so low
is that the 95% number has been made up just to scare people. Users of
Tor Browser are just ordinary people that want the security, anonymity
and privacy properties.

Thanks,
Iain.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.torproject.org/pipermail/metrics-team/attachments/20190226/3ddacc41/attachment.sig>