Linked exploration · exact shares · observed capture rates
See when tracked technologies entered Hacker News’ attention, compare their shapes across time, then move into the forward-captured front page to inspect how individual stories climbed and accelerated.
Start here · verified briefing
Read the finding first, inspect its exact evidence and scope, then follow the supporting view.
Next question: Inspect the AI title-share arc
Show evidenceNext question: Compare the startup and AI arcs
Show evidenceEditorial contract: a local editor may reorder verified candidates, but it cannot author or alter their facts. Invalid guide data leaves this checked HTML visible.
Evidence before interpretation
Loading scope boundaries…
Each analytical surface has a different denominator and clock. This ledger keeps the mature historical census, partial live year, forward capture, and directional samples visibly separate.
Central canvas
Annual share of HN story titles
Every cell is an exact annual title share. Start with within-topic intensity to reveal timing; switch to absolute share to compare magnitude. Select a cell to link that topic to the comparison and captured-story views below.
Keyboard: focus the map, use arrow keys to move cell by cell, then Enter to select. On touch, tap a cell.
Loading the attention map…
Experimental discovery queue
Loading embedded-title coverage…
Curated topics provide stable long-term measures. This separate open-vocabulary view groups embedded titles by meaning to surface candidate themes the fixed tracker may miss; candidates remain provisional until their labels and coherence are reviewed.
Linked view
Indexed to each topic’s own peak
Select up to three topics from the heatmap. The comparison follows the same scale mode, so “within-topic” compares shape while “absolute share” compares actual title share.
Aggregate behavior · privacy-safe
Loading retained-corpus cohorts…
Understand participation without profiling individuals. These views publish cohort totals only—never handles, individual rankings, inferred identities, or comment text.
Story-level outcomes · retained counters
Average comments, discussion per 100 points, silent-story share, and 50+ comment thread share.
How story outcomes vary with retained submission activity.
Submitters with at least five retained stories, grouped when one post format represents 60% or more of their submissions.
Share of each qualifying first-seen cohort observed submitting another retained story in the immediately following calendar year.
Retained comments by depth and the share that received a reply.
Loading collection boundaries…
Source landscape
Registrable domains · retained stories
Compare frequently submitted source domains using the last five complete years by default, including the median annual share to normalize differently sized years. All-time engagement remains labeled context. Selecting a domain links it to Story Lifecycle.
Domain rows are descriptive retained-corpus aggregates. Year bars use share of that year’s retained stories so differently sized collection eras remain comparable. They do not measure publisher quality, audience size, or causal influence.
Score-prioritized linked-page corpus · aggregate framing cues
Loading linked-page evidence…
See which linked pages could be read and summarized, then compare the vocabulary in HN titles with the corresponding local article summaries. Lexical distance can reveal reframing worth investigating; it cannot determine truthfulness, intent, or whether the publisher’s original headline changed.
Direct publisher-title comparison
Publisher-title capture is accruing.
0% of publication threshold
A coverage audit—not a trend line. Only UTC years with at least 50 usable pairs appear.
Share of attempted pages in this score-prioritized acquisition corpus; every attempt receives an explicit terminal status.
Share of meaningful HN-title terms also present in the local summary, among summarized links in the selected corpus.
Score-selected summarized stories only. Outcomes are descriptive within this corpus—not effects or estimates for all submissions.
Selected-corpus coverage and median lexical alignment for domains with at least 250 summaries.
Census structure + directional semantic sample
Loading comment coverage…
Separate the shape of retained discussion trees from the themes in a fixed monthly comment sample. Structural census measures support long-term comparisons; semantic theme shares are directional and never represent all comments.
Share at depth 1 versus depth 4+, plus the share of comments receiving a reply.
Select a reviewed display label to inspect its within-sample share over time.
Aggregate language signals · separate corpora
Loading career evidence…
Explore explicit skill, work-mode, compensation, visa, and seniority language in official HN Jobs and top-level monthly “Who is hiring?” entries. The views remain separate because their publishing processes and coverage differ.
One record can mention multiple skills; shares do not sum to 100%.
Reviewed place names mentioned in the retained record; one record can mention multiple places.
Dollar-denominated annual amounts near explicit salary/pay language; denominator is records with a qualifying amount.
Ask HN · Show HN · Jobs
Loading maturity gates…
Compare complete-year final outcomes now, while the forward trajectory layer accrues. Live rank and dwell comparisons remain visibly gated until the capture reaches meaningful 7-, 30-, and 90-day windows.
Forward capture only
No historical rank paths are inferred
Rank captured stories by their typical positive score, comment, or rank-climb interval, then compare up to three trajectories. Single-interval peaks remain secondary context. Each percentile describes this capture set and is not a claim about all Hacker News submissions.
Loading pattern definitions…
Linked comparison · up to three
Select stories from the table or detail drawer.
Loading coverage and method…