technologybriefs
10:40in productionCh. 1 · What It Measures/ 10:40 · ceiling 15 min
Internet culture

PageRank

1998

PageRank didn’t rank pages — it ranked links, and in doing so, rewired how machines read trust.

PageRank is a foundational, patent-expired algorithm that ranks web pages by modelling hyperlink structure as a probability flow. It works — but only on graphs where links reflect genuine endorsement, not manipulation or noise.

Chapters & takeaways4
  1. 1:03
    What It Measures

    PageRank measures importance through linkage — not content — and applies beyond the web to any system of mutual reference.

  2. 2:54
    How It Thinks

    It treats the web as a graph, defines importance recursively, and assumes importance flows from important sources.

  3. 4:28
    What It Produces

    It outputs a probability distribution, and converges in predictable iterations — even across hundreds of millions of links.

  4. 6:17
    What It Released

    All PageRank patents expired in 2019 — the algorithm is now fully open for implementation and adaptation.

Worth your time?

Yes. Study the whole thing.

4.5/ 5
What works
  • converges reliably on large graphs
  • outputs a stable probability distribution
  • applies to any reciprocal reference system
  • requires no proprietary licensing after 2019
What does not
  • account for content relevance
  • handle adversarial link structures
  • incorporate user behaviour
Study it if
  • researchers building citation tools
  • engineers designing graph-based recommenders
  • teachers explaining recursive ranking
Skip it if
  • developers building real-time search APIs
  • product teams shipping personalisation features
  • policy makers regulating platform ranking
The written brief1 min read

What it is and the problem it solves

PageRank is a link analysis algorithm that solves the problem of measuring relative importance in a hyperlinked set. Before it, search engines ranked pages by on-page text matches — making them easy to manipulate and blind to consensus.

How it works

PageRank is a recursive algorithm that treats the web as a graph: pages are nodes, hyperlinks are edges. It calculates a page’s importance by summing the weighted PageRanks of all pages linking to it — each link’s contribution scaled by the number of outbound links on the source page.

What works

It converges reliably: on a 322 million-link network, it reaches tolerance in 52 iterations. It outputs a stable probability distribution — the chance a random surfer lands on a given page. Its recursive definition and Webgraph grounding make it mathematically tractable and empirically robust at web scale.

What does not

PageRank does not account for content relevance, user intent, freshness, or semantic meaning. It ignores anchor text, page structure, or user engagement signals. It assumes uniform random clicking — a simplification that breaks down under adversarial link farms or attention-driven navigation.

What it changes

PageRank shifted search from keyword frequency to topology. It established that authority could be inferred from network position alone — not just on the web, but in any system of reciprocal references, from academic citations to legal precedents.

Is it worth your time

Yes — if you work with citation networks, recommendation systems, or graph-based ranking. Its core logic remains teachable, implementable, and patent-free since 2019. But it does not scale to modern real-time, multimodal, or behaviourally enriched ranking tasks without heavy augmentation.

Same field · Internet culture4 of 20
9:22
Domain Name SystemPaul MockapetrisDNS is the distributed name-resolution system invented by Paul Mockapetris in 1983. It replaced the ARPAnet’s central hosts file. It uses a hierarchical zone model and originally ran over UDP. Later concerns about reliability, security, and privacy drove TCP use and protocol extensions. RFCs 882, 883, 973, 1034, and 1035 define its evolution. Mockapetris implemented the first version, 'Jeeves', for TOPS-20 the same year.
9:34
PeeringPeering is the voluntary, settlement-free exchange of downstream traffic between autonomous systems — instantiated by physical links and BGP, not contracts. It cuts transit costs and improves performance where traffic is reciprocal. But it offers no guarantees, scales poorly without coordination, and collapses instantly when mutual benefit ends. Its power lies in its simplicity — and its fragility.
8:58
DreamcastSega · 1998The Dreamcast was Sega’s final home video game console. It launched in Japan on November 27, 1998, and globally in late 1999. It used off-the-shelf components, shared hardware with the NAOMI arcade board, avoided DVD licensing with GD-ROM, and shipped with a built-in dial-up modem — the first console to do so. It achieved strong initial sales: 150,000 units sold out in Japan on day one; over 225,132 sold in North America in 24 hours; more than 500,000 in two weeks; and 31% of US video game revenue by Christmas 1999. But momentum collapsed in 2000. Competition from PlayStation 2 and Xbox, weak third-party support, and financial strain forced Sega to discontinue the console on March 31, 2001 — ending its hardware business. Total sales: 9.13 million units. Games released: over 600.
8:50
File Transfer ProtocolAbhay Bhushan · 1971FTP is a foundational but fatally flawed protocol. It delivered interoperable file transfer in 1971, but its reliance on unencrypted channels and rigid two-connection architecture made it unsustainable. It changed how machines shared data — then became the textbook example of inherited insecurity.
Up next in Technology

WebAuthn

· 10:21

WebAuthn doesn’t kill passwords—it just makes them optional for those who can afford the stack.

10:21