gbinal
commented
1 week ago Data is here.
My analysis is as such:
1) On the first question:
- 5,443 out of 15,451 live sites return this header (35%).
- 2,653 out of 9,157 omb/idea canonical sites return this header (29%)
For comparison...
- 260 out of 15,031 live sites use the open graph article_published tag (2%)
- 683 out of 15,031 live sites use the open graph article_modified tag (5%)
- 4 out of 15,133 live sites use the dc.date (0.03%)
- 21 out of 15,133 live sites use the dc.data.created tag (0.14%)
- 16 out of 15,133 live sites use the date tag (0.11%)
- 11 out of 15,133 live sites use the revised tag (0.07%)
- 32 out of 15,133 live sites use the last modified tag (0.21%)
2) On the utility of this data for inferring content freshness
- 4,105 out of 5,443 have a year of 2024
- Another 580 have a year of 2023
- 268 have a year of 2022
- The remaining 629 have a year of 2021 or earlier.
- There are a small number going back to even 2004, 2005, etc.
Another way of looking at this though is to compare sites when they have both the last modified header and open graph article_modified tag in place. The overlap is small 227 sites, and it's not terribly easy to compare/contrast the dates, but when you sort by the open graph tag chronologically, you find that the strong majority, indeed virtually all of the ones that have older open graph article_modified dates (2022 and earlier) have super recent (2024) last modified header dates. Though this sample size is small, it does lead me to question how reliably the header date will be able to indicate that the content on a site is recently updated.
That said, a super old last modifed header would seem to very reliably indicate if content had not been recently updated. As in, if the header has a date of 2015, I don't know that there's any way that the content on that site could be more recent than that.
Following up on #961, parse the data to try and learn how useful it would be.