Tor: The Second-Generation Onion Router
2004 · paper · Computer science
Presents Tor, a circuit-based, low-latency anonymous communication service for TCP applications that improves on the original onion routing design. The authors describe features such as forward secrecy and hidden services, their deployment experience, and open problems.
Checked: PDF from usenix.org, fetched 4 Oct 2026. Page numbers are those of that PDF.
5 findings
Reference or label Section 6.1: "flow control and rate limiting (discussed in Section 4.6)"; rate limiting is the subject of Section 4.5 ("Rate limiting and fairness"), and Section 4.6 ("Congestion control") covers flow control only.
- p. 10 Section 6.1, first paragraph
flow control and rate limiting (discussed in Section 4.6)
- p. 8 Section 4.5 heading
Rate limiting and fairness
- p. 6 Section 4 roadmap paragraph
resource limiting in Section 4.5. Finally, Section 4.6 talks about congestion control and fairness issues.
Judge's note: Read the sentence on page 10 and Sections 4.5 and 4.6. The token bucket is in 4.5; 4.6 opens "Even with bandwidth rate limiting, we still need to worry about congestion", treating rate limiting as already covered, and then describes the windows. The pointer is right for flow control and wrong for rate limiting; it should name Sections 4.5 and 4.6. The roadmap wording raised in one reviewer note is judged under C001.
Second check: another model agreed
Report a problem with this finding
Reference or label Section 6.2: "The hydra model (many entries, few exits) presents a different compromise ... see Section 10." The hydra topology is discussed in Section 9 (Open Questions), not Section 10.
- p. 11 Section 6.2, paragraph beginning 'A mixture of open and restricted exit nodes'
needs to work harder to watch all the clients; see Section 10.
- p. 15 Section 9
Does the hydra topology (many input
- p. 15 Section 9, cascade/hydra paragraph
Does the hydra topology (many input nodes, few output nodes) work better against some adver-
- p. 11 Section 6.2, last paragraph of exit-policy discussion
to work harder to watch all the clients; see Section 10.
Judge's note: Read the sentence on page 11, the Section 9 paragraph on page 15 ("Does the hydra topology (many input nodes, few output nodes) work better against some adversaries?") and the whole of Section 10 in the text. Section 10's items are scalability, bandwidth classes, incentives, cover traffic, caching at exit nodes, directory distribution, specification review, interoperability and wider deployment; a search for 'hydra' finds it only in Sections 6.2 and 9. The pointer should be to Section 9.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference [10]: "Communications of the ACM, 4(2), February 1981." The volume for 1981 is 24, so this is probably 24(2) with the leading digit dropped.
- p. 17 Reference [10]
Communications of the ACM, 4(2),
- p. 17 Reference [10]
Communications of the ACM, 4(2), February 1981.
- p. 17 References [10]
Communications of the ACM, 4(2), February 1981
- p. 17
digital pseudo-nyms. Communications of the ACM, 4(2),
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: Communications of the ACM began in 1958, so the 1981 volume is 24, and Chaum's paper is in volume 24, number 2. The page image prints "4(2), February 1981", so this is not an extraction artefact. Volume 4 and the year 1981 cannot both be right given the journal's start date.
Second check: another model agreed
Report a problem with this finding
Reference or label Reference [52] (Tangler, CCS-8): "pages 86–135" is a 50-page range for a conference paper; the first page is probably misprinted (the paper is believed to occupy pages 126–135).
- p. 18 Reference [52]
pages 86–135. ACM Press, 2001.
- p. 18 References [52]
pages 86–135
- p. 18 Reference [52]
nications Security (CCS-8), pages 86–135. ACM Press, 2001.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that the Tangler paper occupies pages 126–135 of the CCS-8 proceedings. The reviewers do not agree on the correct range (126–135, 86–96 and 86–93 were all offered), so the right pages need checking against the ACM record. What the page itself shows is "pages 86–135", a 50-page span, against 6 to 20 pages for the other conference entries in the list.
Second check: another model agreed
Report a problem with this finding
Wording Section 5.1: "signing the advertisement with his public key"; a signature is made with the private key of the pair.
- p. 9 Section 5.1, second bullet
signing the advertisement with his public key
Judge's note: Read the bullet list on page 9. The preceding bullet says Bob "generates a long-term public key pair", so the intent is clear, but as printed the signing key is misnamed. Loose wording with an evident fix ("private key", or "the key pair").
Second check: another model agreed
Report a problem with this finding
7 minor (typos and the like; most checked by the judge only)
Reference or label Reference [42]: "ACM TISSEC, 1(1):66–92, June 1998." The issue date of TISSEC volume 1, number 1 is believed to be November 1998.
- p. 17 Reference [42]
ACM TISSEC, 1(1):66–92, June 1998.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: that ACM Transactions on Information and System Security volume 1, number 1 is dated November 1998. The page prints "June 1998"; nothing in the files can confirm or refute the month, and a widely copied bibliography entry carries June. Volume, issue and pages are not in question. Worth a check against the ACM record before it is stated as an error.
Second check: another model agreed
Report a problem with this finding
Wording The Section 4 roadmap says "resource limiting in Section 4.5. Finally, Section 4.6 talks about congestion control and fairness issues", but the headings are "4.5 Rate limiting and fairness" and "4.6 Congestion control"; the fairness discussion (preferential service for interactive streams) is in 4.5.
- p. 6 Section 4 roadmap paragraph, before 4.1
resource limiting in Section 4.5. Finally, Section 4.6 talks about congestion control and fairness issues.
- p. 8 Section 4.5 heading
Rate limiting and fairness
- p. 9 Section 4.6 heading
Congestion control
- p. 6 Section 4 roadmap paragraph, before 4.1
Finally, Section 4.6 talks about congestion control and fairness issues.
Judge's note: Read the roadmap on page 6 and the headings and text of 4.5 (page 8) and 4.6 (page 9). 4.5 carries the token bucket and the interactive-versus-bulk preferential service; 4.6 covers circuit- and stream-level windows only and does not use the word fairness. Congestion control can loosely be called a fairness matter, so the roadmap is not strictly false; it is inconsistent with the paper's own headings. The Section 6.1 pointer mentioned in one reviewer note is judged under C007.
Second check: another model agreed
Report a problem with this finding
Typo The naming parenthetical is given twice in one paragraph: "the first node in her chosen path (call him Bob)" and, two sentences later, "encrypted to the onion key of the OR (call him Bob)", after Bob has already been named and used.
- p. 6 Section 4.2, 'Constructing a circuit'
a create cell to the first node in her chosen path (call him Bob)
- p. 6 Section 4.2, 'Constructing a circuit'
encrypted to the onion key of the OR (call him Bob)
- p. 6 Section 4.2, 'Constructing a circuit', first paragraph
a create cell to the first node in her chosen path (call him
Judge's note: Read the 'Constructing a circuit' paragraph on page 6. Both parentheticals are printed, and the sentence between them already uses "Bob". A redundant repeat; the meaning is unaffected.
Second check: none (judge only)
Report a problem with this finding
Typo Figure 1, label between OR 1 and OR 2: "(link is TLS-encryped)" should read "TLS-encrypted", as in the label between Alice and OR 1.
- p. 6 Figure 1, label between OR 1 and OR 2
(link is TLS−encryped)
Judge's note: The enlarged crop of Figure 1 shows "(link is TLS-encrypted)" on the first link and "(link is TLS-encryped)" on the second, missing the 't'. It is in the rendered figure, not only in the text layer; the reviewer who thought both labels read 'encrypted' was looking at too low a resolution.
Second check: none (judge only)
Report a problem with this finding
Typo Reference [10]: "and digital pseudo-nyms" has a stray hyphen; the word is "pseudonyms".
- p. 17 Reference [10]
and digital pseudo-nyms
Judge's note: On the page 17 image "pseudo-nyms." sits in the middle of the second line of entry [10], followed by the journal name, so it is not an end-of-line hyphen or an extraction artefact.
Second check: none (judge only)
Report a problem with this finding
Typo References [7], [19] and [49] give the proceedings title as "Designing Privacy Enhancing Technologies: Workshop on Design Issue in Anonymity and Unobservability"; the workshop title has "Issues" (plural).
- p. 17 Reference [7]
Workshop on Design Issue in Anonymity and Unobserv-
- p. 18 Reference [49]
on Design Issue in Anonymity and Unobservability
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the title of the LNCS 2009 proceedings is 'Workshop on Design Issues in Anonymity and Unobservability'. All three entries on pages 17 and 18 print "Design Issue", evidently from one shared bibliography string. The singular is also ungrammatical in the phrase. Cosmetic.
Second check: none (judge only)
Report a problem with this finding
Typo Reference [50]: "A. Tannenbaum. Computer networks, 1996." The author's name is spelled Tanenbaum.
- p. 18 Reference [50]
A. Tannenbaum. Computer networks, 1996.
Only the authors can say what is right here.
Judge's note: Rests on outside knowledge: the author of 'Computer Networks' (third edition, 1996) is Andrew S. Tanenbaum, with a single 'n' in the first syllable. The page 18 image prints "Tannenbaum".
Second check: none (judge only)