Tor: 80 percent of ??? percent of 1-2 percent abusive.
Hi, Nick here.
Roger's at 31c3, so I'll post his statement about that article you might have seen:
Tor hidden service traffic, which Dr. Gareth Owen discussed in his talk this afternooon, is only 1.5% of all Tor traffic. Tor gets about 2 million users per day total.
The researcher ran a set of Tor relays for a six month period, and recorded how many times somebody attempted to look up a hidden service (this lookup is one of the steps in visiting a hidden service). Then at the end of that period, he scanned the hidden services he'd learned about, to find out what sort of content was on them.
Dr. Owen's data shows that there's a lot of churn in hidden services, so nearly all of the sites were gone by the time he did these scans. His graphs only show data about the sites that were still up many months later: so his data could either show a lot of people visiting abuse-related hidden services, or it could simply show that abuse-related hidden services are more long-lived than others. We can't tell from the data.
Without knowing how many sites disappeared before he got around to looking at them, it's impossible to know what percentage of fetches went to abuse sites.
There are important uses for hidden services, such as when human rights activists use them to access Facebook or to blog anonymously. These uses for hidden services are new and have great potential.
PS: Law enforcement agencies use Tor to stay anonymous while they catch bad guys. Law enforcement agencies use and run hidden services, too.
More info to follow.
FWIW, the cited percentages refer to the number of look-ups, not to the number of sites. These are two different things: It's possible for a small number of sites to be looked up frequently, or for a large number of sites to be looked up infrequently. Also, there isn't necessarily a one to one correspondence between look-ups and requests.
The total counts in Dr. Owen's presentation were about all hidden services, not just web sites. His two slides with categorizations of hidden web sites (in existence, of which abuse sites were a small percentage, and lookups for them, of which they were the majority) only listed percentages and not the actual total. It would be good to know the total number of web sites that his percentages are about.
Other researchers have estimated that the number of long-lived child abuse sites is very low, certainly a two digit number.
I don't believe long-lived hidden services are currently safe against a powerful nation state adversary, so it seems quite likely to me and others I've talked to that child protection agencies could have deanonymized these services if they wanted to but instead they've found that their existence makes it easier to find and prosecute the producers and consumers of the content.
It seems unknowable what portion of these lookups are child abusers vs how many are crawlers, but it is certainly known that there are many law enforcement agencies that regularly crawl these sites.
>It seems unknowable what portion of these lookups are child abusers vs how many are crawlers
Are there really that many child abusers there? From what I know, the majority only share pictures that they found elsewhere and the child abusers (producers) tend to not share on Tor and only share with friends.