Google testing "before:" and "after:" commands to help find old pages

[Robert Charlton]

Danny Sullivan, Google's Search Liaison, recently announced via a Twitter thread that Google is beta testing two new search operators, before: and after:. The operators are to be used from within the Google search box, and they currently appear to be alternatives to the time settings in the Google search "Tools" menu.

Google SearchLiaison @ searchliaison 1:03 PM - 9 Apr 2019

For many searches, surfacing fresh information often ensures the best relevancy. However, sometimes people want to find older yet also relevant information. Our tools have long made this possible. Now we’re testing new before: and after: commands to make this even easier....

Danny reviews the evolution of time-search on Google, noting that by reducing the time interval we can surface more pages, thus making results for a chosen interval more granular.
[twitter.com...]

IMO, this very much relates to our current thread...
Google No Longer Indexes all The Web
https://www.webmasterworld.com/google/4941842.htm [webmasterworld.com]

...as it suggests another approach to unearthing old pages, which is to allow the user to choose the time segments, rather than attempting a full historical view in a set of ten links. The segmented approach is particularly useful in finding gold in old pages, which are othewise buried under ever-growing layers of new results. How this will eventually relate to the apparent size of the web that Google is indexing remains to be seen. A lot here, I suspect, will relate to searcher education.

----
I tested this segmented approach with old subjects that are still around. It seemed particularly helpful to me in exploring forum searches, which are necessarily temporal in their nature, surfacing details on topics that were prevalent for a time and then faded from discussion. This, of course, for forums that still exist, with old threads still indexable by Google.

Apart from forums, which preserve an historical time line... I should add that, from what I can see, results of, say, ten years ago do bring up sites that were ranking at the time, albeit some of those sites have now changed considerably. For those sites, of course, you can't possibly get an accurate representation of what content actually was ranking at the time, though the rankings in the serps do seem to be more or less what I remembered.

Where sites and companies behind those sites have disappeared... ie, they've been acquired or gone out of business... Google does not bring the old sites back. The rankings shown are the rankings of the surviving domains of the time in context, at least as I remember them, but the missing sites and pages are gone. So, it's not really a time machine, because too much else has changed.

Danny also explores the flexibility of the new "before:" & "after:" commands, which... for search choices involving intervals of years... I find to be more convenient and flexible than either the preset time ranges or the calendar interface that Google Time search currently offers.

[Robert Charlton]

PS to the above... Google has tackled the question of dates, not just with search operators, but also with guidelines for specifying dates on different types of content. Danny suggest reading this in a recent official Google blog...

Help Google Search know the best date for your web page
Monday, March 11, 2019
[webmasters.googleblog.com...]

News sites are looked at in a more precisely sequential than other types of material.

Worth noting here is this WebmasterWorld thread from 2015, about dates and articles and rankings....

Removing date from articles - Ranking impact?
Nov 2015
https://www.webmasterworld.com/google/4779415.htm [webmasterworld.com]

I've been a proponent of keeping dates as real and as prominent as possible, particularly on technical or search-related topics, where the order of developments is important. It's worth comparing Google's newly stated date deadlines with what our 2015 thread came up with.

I'm very curious where Google is going with this... how many results are simply going to be Supplemental, and how many others are likely to become Vintage, like old wines, brought out for special occasions.

[aristotle]

Where sites and companies behind those sites have disappeared... ie, they've been acquired or gone out of business... Google does not bring the old sites back. The rankings shown are the rankings of the surviving domains of the time in context

In my opinion that's a big drawback. Lots of valuable old articles would be excluded simply because the domain no longer exists.

The Wayback Machine (archive .org) keeps those valuable old pages available, but google doesn't. This is a major deficiency in google's approach.

So Google has gotten it wrong in this case

[tangor]

The bucket is sloshing overfull ... this is an attempt to pour some of it out into smaller containers. Sadly, once started, g will fracture its own index/reputation, or end up with TWO indexes "before" and "after" ... and might be as historical as

BC to AD ... :)

[tangor]

NOTE: BC to AD is

"Before Crisis" and "After Decision".

[Robert Charlton]

Sadly, once started, g will fracture its own index/reputation, or end up with TWO indexes "before" and "after" ... and might be as historical as / BC to AD

tangor, the idea of TWO indexes might make sense if we (or Google) were measuring from one point only, and that was a constant in time. That, however, is not the case.

The "before:" and "after:" operators are on a continuum, most likely changing for each new query, so the idea of only TWO indexes doesn't really make sense. There are many befores, and many afters. These are independent ranges, parameters to be refined as Google refines the interface.

In a sense, these operators are simply an extension of the calendar interface they've had for a long time, only now there are two of them, they're easier to use, and you can use the calendar instead if you're so inclined. I find the calendar mostly clumsy except for specifiying recent intervals, as measured back from "now". Like, "the last year" is pretty easy to check off as a choice. I keep wishing they had a "in the last two years" choice as well, which I think would be a much more useful range for many kinds of research.

Google in my memory once did have a time-line search, when it was indexing the Twitter firehose, and I found it fascinating, as you could really zoom in on a given point an increase the granularity as you narrowed the window you were looking at. But those Tweets were points stuck to a time-line that would remain, much simpler than evanescent websites that might come and go and morph on the web.

I've always thought the time-line search which incorporated a picture of history, was an area that was too much neglected, and have hoped the Google might institute it into channels of culture, commerce, and geo-related topics.

How Google might incorporate the evolution of the web itself, without a WayBack machine type database, would very tricky... particularly as many sites use NOARCHIVE... and there are also gaps and discontinuities (not in Google, but in the entitities themselves); and I think the best that might be done now is to follow a chain of 301 redirects where they exist, albeit (at least for now) you'd have to know what they were before you could follow them.

A history of a company's web acquisitions, eg, can be one interesting area to explore. It's not something, at least for now, that these operators can give you. I can imagine ways of building this into a time range search, but it can only be as continuous as the entities themselves.

[tangor]

the idea of TWO indexes might make sense if we (or Google) were measuring from one point only, and that was a constant in time.

Chuckles. The before and after is cosmetic, with goal posts changing to suit g, not the user. I suspect this will go over as well as G+ did, and with as much falderal. :)

Don't get me wrong, having the option is fine, just don't know how much value might be applied at this time. g has let users down too often in recent years.

[Robert Charlton]

tangor, you seem to be hellbent on taking this thread off topic. You wrote...

The before and after is cosmetic, with goal posts changing to suit g, not the user.

Either it will be a fracture of historical proportions, as you previously said, or it's a cosmetic change of no import. I assume to a degree that if Google ignores the user too often, it will hurt the company. I wasn't happy that they shut down Google Reader, and am dismayed about how much effort got wasted with Google+. I trust they've heard the many murmurs out there. I'm sure that this feature hasn't yet become that significant

The feature is in Beta, which for Google is really writ in water. It is what it is, and we've been advised by Danny how much it might get changed. How often it's used, and what it might bring to general search, probably has a lot to do with whether it will survive.

I gave it a fairly thorough shakedown in exploring past rankings in niches I'm familiar with, and in taking a look at various topics that are less popular now than they were, say, ten years ago... and in both cases I found the operators fairly useful in unearthing good results. I use Time Search a lot, and have used it since the original calendar interface became available. You may well not have found the feature useful for the kind of searching you do.

(Let me add, mod's hat on, that I know that while it must be tempting to complain at Google's failings with every new feature, constantly complaining no matter what the context, and then pontificating about the company... that this is generally off-topic in this forum.)

[tangor]

Merely looking at this from this viewpoint: [webmasterworld.com...]

If there is a before and after, those sites which still exist, but are "before" might not be able to be found in the future. Seems a valid concern.

[martinibuster]

This is not related to the old pages gone missing issue. The function has always existed, in the tools section. This just turns the tool into a command, that's all. The command is new. The functionality is not.

[tangor]

Will the users know about it? Any announcements THEY will know?

Else kind of a limited thing for webmasters...

Why I ask, users have no clue on boolean type searches, much less "before" "after".

[Robert Charlton]

tangor, I think you're making these two operators, before: and after: ... much more complicated that they really are. The syntax, if you want to combine them both, is *after* such and such a start date... and then, if you want to combine both, you also add a date with the before: operatator that's *before* your specified end date. You needn't use both operators, but if you do, they give a a date range. The date formats are as flexible as any I've seen. Here's an example from Danny's Twitter thread I linked to up top...

The before: & after: commands return documents before & after a date. You must provide year-month-day dates or only a year. You can combine both. For example:
[avengers endgame before:2019]
[avengers endgame after:2019-04-01]
[avengers endgame after:2019-03-01 before:2019-03-05]

Power user note! If you provide only a year, before: & after: translates those into full dates that work, such as follows:
[before:2018] = [before:2018-01-01]
[after:2018] = [after:2018-12-31]

More power user notes! You can use either dashes or slashes in dates. Both of these are valid:

2018-12-31
2018/12/31

You can also use a single digit for month or day, so all of these are valid:

2019-3-1
2019-3-12
2019-03-01

Etc etc etc....
As far as I understand it, no queries get engraved in stone. There is no iconic BC / AD divider we're stuck with for thousands of years. But yes, there will be gaps, and there's no way (yet) of filling them without doing a lot of outside research. That may be because many webmasters chose the "noarchive" robots option... or it may be because a brand name got acquired by a large company and they buried it. I'm sure there's a set of rules you can discover if you know the history of a niche well enough and put in a lot of time searching out all the players of the time.

{But a time range search is definitely not a perfect recreation of a past era, and (for sci-fi fans) accidentally misspelling the name of a butterfly while checking out an old result won't change the course of history.}

To see how one of the better known experts on search hacks of the past gives this new interface a trial run, check out Tara Calishain's recent ResearchBuzz article on the new interface....

Google Adds Two New Syntax for Date Search
By researchbuzz2 on April 10, 2019
[researchbuzz.me...]

I have been obsessed with searching Google by date for a loooong time. In 2003, shortly after Google released its API documentation, I figured out how to create a date search form and called it GooFresh. It’s long gone, of course,...//
Google has retained date search all along, making it easier to use with the addition of date ranges (and custom date ranges) under its Tools option. Now, however, Google has launched (via a Twitter thread) two syntax that will allow you to do date searches without filling out a form: before: and after:.

By narrowing down your search results to the new range you've defined (for one search, or many... it doesn't matter)... you've got a tool that can effectively make your queries more focused within the range you've defined. This helps to see results that you might have otherwise missed. It's not really making the web larger... It's just stretching the capabilities of those ten lines by digging deeper into results that might otherwise have gotten buried.

That's why, in my first post in this thread, I posited this as a possibility of effectively stretching our view of the web, which was being discussed here...

Google No Longer Indexes all The Web
https://www.webmasterworld.com/google/4941842.htm [webmasterworld.com]

Yes, Google will have to get better than they have been in publicizing their operators and software to a newer audience, and perhaps providing some contextual help to make their minimalist interfaces more intuitive. But ultimately, this probably is a power users feature, and I'm encouraged that Google is willing to give it a shot and see if people use it.

[aristotle]

In my view, for most people the after: operator would be more useful than the before: operator.

As I mentioned previously, the before: operator is seriously defective because it doesn't include pages that are no longer live on the web. As a simple example, suppose for the period 2004-2010, a certain article about "antique widgets" was consistently number 1 in google's results for searches for that term. But then, for whatever reason, that article disappeared from the web. Thus it wouldn't show up in the before: results for someone using that operator to explore that period. In fact, in my opinion those results would be of little value because of this omission.

I can actually remember some excellent old articles in certain niches that are no longer live on the web. It would be nice if google chose to make them available again.

[Robert Charlton]

In my view, for most people the after: operator would be more useful than the before: operator.

aristotle, I'm going to disagree, as while what you describe might be true for you, and sometimes even for me, it's certainly not going to be true for everybody all the time.

I recently was looking for the date some years back when a feature was introduced, and it was much easier to pick an arbitrary before: date that I knew was too early, along with an after: date closely following.... and then to shift the after date later and later incrementally, until it started including the terms I was after. You can pinpoint something pretty closely that way.

You also wrote:

The before: operator is seriously defective because it doesn't include pages that are no longer live on the web

Keep in mind that this is a Google-only operation. I'm sure there would be complaints and perhaps legal complications if they went to the Wayback machine to get the missing data, and disregarding the "noarchive" tag would be equally problematic... so, until we have a time machine that recreates everthing in a way that causes no one to complain, I'm not sure how we can achieve the perfection you desire.