The Anatomy of a Large-Scale Hypertextual Web Search Engine
Computer Networks and ISDN Systems, 1998 · paper · Computer science
Describes Google, a prototype large-scale web search engine that uses hyperlink structure to improve results. It details how the system crawls and indexes the web, with a prototype database of at least 24 million pages, and how it copes with uncontrolled content.
Checked: PDF from snap.stanford.edu, fetched 4 Oct 2026. Page numbers are those of that PDF.
2 findings
Substance Sec. 2.1.1 prints PR(A) = (1 − d) + d(PR(T1)/C(T1) + ... + PR(Tn)/C(Tn)) and then says "the PageRanks form a probability distribution over Web pages, so the sum of all Web pages' PageRanks will be one"; the formula as printed does not give a sum of one.
- p. 4 Section 2.1.1, PageRank definition
so the sum of all Web pages’ PageRanks will be one
- p. 4 Sec 2.1.1
so the sum of all Web
- p. 4 Section 2.1.1
the sum of all Web pages’ PageRanks will be one
Only the authors can say what is right here.
Judge's note: I checked the formula and sentence on the crop and on page 4. My own working: sum the equation over all N pages and assume every page has outlinks. Each page's PR is split over its C(T) outlinks, so the d term adds up to d·ΣPR. That gives ΣPR = N(1−d) + d·ΣPR, so ΣPR = N, not 1. Dangling pages would make the sum differ further, but it still would not be 1 in general. The sum comes to one only if the constant is (1−d)/N. The formula and the claim cannot both hold as printed. Only the authors can say which was intended (ref [7] may give the normalised form), so this goes to author_query.
Second check: another model agreed · checked by hand
Report a problem with this finding
Substance "the d damping factor is the probability at each page the 'random surfer' will get bored and request another random page". In the printed formula, d weights link-following and (1-d) is the random-jump term, so the probability of getting bored is 1-d.
- p. 4 Section 2.1.2
the d damping factor is the probability at each page the “random surfer” will get bored and request another random page
- p. 4 Sec 2.1.2
the d damping factor is the probability at each
- p. 4 Sec 2.1.1 formula
PR(A) = (1 -d)
Judge's note: Page 4 shows d multiplying the link sum and (1 - d) as the random-jump term. With the stated d = 0.85, the wording implies the surfer jumps at random 85% of the time, which is the opposite of the formula.
Second check: another model agreed
Report a problem with this finding
4 minor (typos and the like; most checked by the judge only)
Typo Sec. 4.2 reads "far less bit manipulation than Huffman coding.ding. Our compact coding uses two bytes"; the stray fragment "ding." should be deleted.
- p. 6 Section 4.2, hit list paragraph
coding.ding.
- p. 6 Section 4.2, hit list paragraph
coding.ding. Our compact coding uses two bytes
- p. 6 Sec 4.2, hit lists
coding.ding
- p. 6
far less bit manipulation than Huffman coding.ding.
Judge's note: The page 6 crop shows "coding.ding." in the typeset text, so this is not an extraction artefact. It is a leftover duplicated fragment.
Second check: none (judge only)
Report a problem with this finding
Typo "very small for a reasonable number of buckets" uses 'buckets' where the rest of the paper says 'barrels'.
- p. 6 Section 4.2, forward index
small for a reasonable number of buckets
- p. 6 Section 4.2, forward index paragraph
for a reasonable number of buckets
Judge's note: Crops C007-a and C007-b show 'barrels (we used 64)' earlier in the same paragraph and 'buckets' only here. The meaning is clear, so this is a cosmetic terminology inconsistency.
Second check: none (judge only)
Report a problem with this finding
Typo "These results demonstrates some of Google's features" has a subject-verb agreement error; it should read 'demonstrate'.
- p. 7 Section 5
These results demonstrates
Judge's note: The wording is visible in crop C009-a.
Second check: none (judge only)
Report a problem with this finding
Typo The Conclusions say "including page rank, anchor text", writing the term as two words where the rest of the paper uses 'PageRank'.
- p. 9 Section 6 Conclusions
including page rank, anchor text
Judge's note: Page 9 (crop C011-e) shows 'page rank, an-chor text'. This is a cosmetic inconsistency.
Second check: none (judge only)