For a long time I assumed that google indexed pretty much everything it, and it was only a question of providing a specific enough set of search terms to drag up older content.
But what you hint at might be more correct these days. They are running a reverse wayback machine in that anything not changed in the last year gets removed. If you click the advanced search its "updated within" and the max timeframe is a year.
> They are running a reverse wayback machine in that anything not changed in the last year gets removed.
Sometimes I wish that were true! Try Googling for, say, PostgreSQL documentation and the top result will often be for a 10-year-old version of the software.
Nitpick, that's kind of a non example because the official Postgres docs let you swap versions more or less seamlessly. IME I will click the top result then click the correct version for my Postgres SE queries. 2 clicks, 0 scrolling.
You're right. I tend to click back, then revise my search to include the version, and then click the result, so Google gets the message that when I search for Postgres docs, I want the most recent version. I have no idea if this actually works, but I heard Google uses bounces to determine relevancy, so I thought it was worth a shot.
Before the omnibar became popular I used to use google as my homepage and I'd type website keywords then "I'm feeling lucky" to get to my frequent websites. I think bookmarks took up too much space vertically so this was my solution /shrug
Well it might be easy to switch but it is a good example
Why is it that Google is thinking the older page is more relevant? Does PageRank outrank content (and Google is oblivious to similar pages that have different versions?)
> For a long time I assumed that google indexed pretty much everything it,
They did that for a long time, but some years ago the index grew so big, they started restricting it. I thing the general timeframe is 10 years or less till the last update.
> If you click the advanced search its "updated within" and the max timeframe is a year.
Because it makes no sense to go further. For older content you can define individual dateranges. And yes, it works fine for me. Tested a search for 2015 just now, first side had entries all from 2015.
But what you hint at might be more correct these days. They are running a reverse wayback machine in that anything not changed in the last year gets removed. If you click the advanced search its "updated within" and the max timeframe is a year.
In fact it seems the date range example doesn't even work: https://developers.google.com/custom-search/docs/structured_...
If I fiddle with it, it returns a result, but I see an hit from just a few days ago at the top...