Per-field breakdown

The headline "% with any IPTC" hides a lot of detail. Below is the population rate of each IPTC field across the 6,845 images analysed in the latest crawl. Field names link to their definitions in the IPTC Photo Metadata Standard. The "scored" badge marks the four fields that contribute to a site's overall score (the "Four Cs" of news photo provenance); the others are tracked for context but don't affect the score — see the methodology for why.

Field Type Present Total Population Score weight
Creator scored 479 6,766 7.1% 25%
DateCreated tracked 462 6,766 6.8%
CreditLine scored 379 6,766 5.6% 25%
Copyright scored 351 6,766 5.2% 25%
ObjectName tracked 315 6,766 4.7%
CaptionDescription scored 296 6,766 4.4% 25%
Source tracked 259 6,766 3.8%
LocationCreated tracked 221 6,766 3.3%
Keywords tracked 172 6,766 2.5%
WebStatement tracked 108 6,766 1.6%
LicensorURL tracked 80 6,766 1.2%
DataMining tracked 46 6,766 0.7%
UsageTerms tracked 44 6,766 0.7%
Genre tracked 18 6,766 0.3%
LicensorName tracked 2 6,766 0%
DigitalSourceType tracked 1 6,766 0%
AIPromptInformation tracked 0 6,766 0%
AIPromptWriterName tracked 0 6,766 0%
AISystemUsed tracked 0 6,766 0%
AISystemVersionUsed tracked 0 6,766 0%
AltTextAccessibility tracked 0 6,766 0%
ExtendedDescriptionAccessibility tracked 0 6,766 0%
LocationShown tracked 0 6,766 0%
Total 100%

Bars are scaled to the most populated field, not to 100%, so differences within the long tail stay visible. The score weight column reflects the per-field weights in config/scoring.yaml: the four scored fields are weighted 25% each. Tracked fields show "—" because their presence does not contribute to the score.

What this tells us

Even the best-populated field — Creator at 7.1% — appears on fewer than one in ten images we fetched. The fields most relevant to licensing (WebStatement, LicensorURL) are below 3%. The dataset as a whole tells a consistent story: publishers and CDNs are discarding metadata that the photographers and agencies almost certainly embedded upstream.