stoic garnet
Hi all!
Do we have a way to remove last-edit dates from search engine results? (27 Dec 2021 in this case). I've noticed a lot of other mira wikis also have this by default. We have autodescription and clean autodescription enabled in SEO tools if that relates at all.
(https://rainworld.miraheze.org)
robust palm
Hi all!
Do we have a way to remove last-edit dates from search engine results? (...
That's google being lazy and not crawling
If the domain is super popular, crawl time would be each hours instead of years
radiant scarab
Our wiki doesn't have AutoDescription (just SEO + TextExtracts + PageImages) and the dates also appear. What's interesting is that a few newer pages don't have it
stoic garnet
oh that is interesting, huh
stoic garnet
That's google being lazy and not crawling
If the domain is super popular, crawl ...
Oh so it's due to the search engine crawling? We get it on duckduckgo and bing as well
robust palm
no idea but should be
stoic garnet
Surely the date would remain the same no matter how often the page is crawled, as it is just the date the page was last edited on the wiki?
robust palm
last time I checked my wiki and it says 7 days ago
stoic garnet
robust palm
oh
radiant scarab
yeah it does look to be last edit date
robust palm
that also, but to be precise, Google does sometimes behave weird
if it can't find the "updated" date, it looks into revision history
stoic garnet
Looking at other wiki farms (and large wikis in general) it doesn't come up there, so I do think it's something we'd be able to turn off?
If so it makes sense that most wikis just don't get around to it
radiant scarab
I feel like it's not something we can control from wiki level
stoic garnet
it does seem very consistent for mira wikis hm
radiant scarab
A quick google search had shown articles on how to remove the date. So I assume it's a default behavior
stoic garnet
:o
radiant scarab
I feel like it's not something we can control from wiki level
but ^
radiant scarab
radiant scarab
these are two tags I suspect cause it on my wiki
You can try to inspect one of the pages on yours and check if these appear too!
stoic garnet
Ooh taking a look now 👩🔬
radiant scarab
Sadly removing those needs to be done higher up in the wiki hierarchy 😆
radiant scarab
no idea but isn't Google doesn't read OpenGraph?
not sure, but it's the only date related thing I could find
in the source
robust palm
I bet Google read date somewhere which user can see
stoic garnet
(Just to say obviously, the logic for not wanting a date shown is that for the average user it can imply that the page or wiki is quite old or outdated, when in reality it's just... the very static front page, haha)
robust palm
Like this one
stoic garnet
I suppose we could update it every month or so but that's a little hacky haha
radiant scarab
Like this one
Unlikely, google uses abbreviations for the dates in the results. It would be weird if it managed to automatically obtain a relevant date about it from thousands of differently structured sites
robust palm
Unlikely, google uses abbreviations for the dates in the results. It would be we...
As far as I know, Google use AI to read those
and I even see sometimes Google does override custom snippet text with text from articles
radiant scarab
Like ignoring a custom description specified via wikiSEO?
robust palm
yep
radiant scarab
hmm
anyways I like the idea of disabling the date even if it was for every wiki on Miraheze. Of course doing so globally would require agreement of many users on meta
But the first step would be pinging someone who knows if it's even possible
stoic garnet
aha
radiant scarab
But the first step would be pinging someone who knows if it's even possible
CosmicAlpha probably
stoic garnet
yes, for homepages it does feel a little clunky or superfluous
robust palm
radiant scarab
Could open a phab ticket I suppose :o
phab is more for extension issues, unless you want to suggest it to wikiseo
Maybe we should first find out if it's even a possibility with how Miraheze is structured
robust palm
I would suggest looking into application/ld+json which Google actually reads
gray mountain
hmm, let me see
interesting
you can control that field via WikiSEO
and it seems to support magic words such as {{REVISIONTIME}}
radiant scarab
[[mw:Extension:WikiSEO]]
gray mountain
perhaps setting it blank would make WikiSEO stop returning a value?
robust palm
you can control that field via WikiSEO
oh wait, then which extension controls ld+json?
stoic garnet
Oh very interesting! I'll try it out!
gray mountain
oh wait, then which extension controls `ld+json`?
I think also WikiSEO
though it seems GoogleRichCards does too
radiant scarab
The only instance I see in the documentation is
|modified_time={{REVISIONYEAR}}-{{REVISIONMONTH}}-{{REVISIONDAY2}}
published_time is the new value which also controls modified_time
radiant scarab
odd that it isn't in the list of parameters
ah, makes sense
Something that'd be good to test yeah
I'm actually force crawling my main page to test rn
stoic garnet
hmm this isn't testable via Special:ManageWiki/settings/wikiseo, is it? aha
robust palm
robust palm
hmm this isn't testable via `Special:ManageWiki/settings/wikiseo`, is it? aha
per page setting
{{#seo: |published_time={{REVISIONYEAR}}-{{REVISIONMONTH}}-{{REVISIONDAY2}} }}
stoic garnet
oh right!
gray mountain
{{#seo:|published_time=-}} seems to work
radiant scarab
`{{#seo:
|published_time={{REVISIONYEAR}}-{{REVISIONMONTH}}-{{REVISIONDAY2}}
}}...
this means we can technically make it display the date of last purge? >:D
robust palm
nope, that's only set to last revision
radiant scarab
this is yes. But if other magic words work then we can
radiant scarab
Btw setting published time to empty does nothing, but that makes sense
radiant scarab
is there a magic word to change date when purge?
there is one to display the current date which of course won't change if the page is cached
hmmm a 503
stoic garnet
I'm compelled to seek further clarification; looking at the wikiseo extensions page this is being done in.. localsettings.php?
radiant scarab
Afaik you add a seo parser function to a page to change this.
radiant scarab
`{{#seo:|published_time=-}}` seems to work
^
radiant scarab
Now, it changes the published time but not modified time 
But google maybe looks at published time for this
radiant scarab
published_time is the new value which also controls modified_time
It would mean that this is false then
robust palm
ld+json is slightly below OpenGraph
gray mountain
Google looks at the ld+json though first, no?
radiant scarab
ld+json has the same values
robust palm
Google looks at the ld+json though first, no?
ld+json is snippet
gray mountain
ah
hmm
seems like something to make a WMF Phabricator task for then
to disable that value
robust palm
Fun fact: ld+json also handles search which you can search contents inside wiki directly without pressing search button
like this
radiant scarab
ld+json has the same values
gray mountain
ah
radiant scarab
`|modified_time=`
modified_time=- has no effect actually
maybe it's meant to take the value of published_time but for some reason isn't
hmmm maybe it detects that published_time is an invalid value and because of that sets to an automatic value?
radiant scarab
hm
stoic garnet
right :o
radiant scarab
Now, if google takes the value of modified_time, there's no way currently to affect the displayed date
robust palm
modified_time is depending on published_time so either set published_time properly or disable the extension
radiant scarab
let's see if modified_time copies published_time if it's a valid date then
nope it doesn't
it's uncontrollable
There's also nothing about it depending on published_time in the documentation, just that it's a removed tag that's set automatically
robust palm
we only need to get schema.org to work
OpenGraph is what Discord, Twitter, etc... uses
oh, doesn't work also
radiant scarab
Yes, it really seems to be impossible to customize via the extension
robust palm
oh bruh, didn't see the dot
so we can't do anything with the modified time
some interesting reading
Google even looks at content date, not just ld+json or OpenGraph
radiant scarab
interesting
It seems like if dateModified is specified, it would look at it instead of datePublished
but it isn't explained anywhere so not sure
robust palm
It seems like if `dateModified` *is* specified, it would look at it instead of `...
that's Google issue sometimes
I do see result on Google that says 2022 but the content is 5 yrs ago
radiant scarab
Since what we've been trying to do since is "hacking" the dateModified value to make the date disappear in search result, maybe we should request a change to the extension to not add this data at all when needed
radiant scarab
It wouldn't prevent google from searching through other content for it but would hopefully hide it
heady grove
Looking at other wiki farms (and large wikis in general) it doesn't come up ther...
it appears for wiki.gg, so it's more than likely Fandom that have removed that from core MW.
radiant scarab
it appears for wiki.gg, so it's more than likely Fandom that have removed that f...
Or, maybe wiki.gg uses wikiSEO as well?
heady grove
Or, maybe wiki.gg uses wikiSEO as well?
radiant scarab
Hmm I cannot find any data in the source though. Looking at Coromon wiki btw, and there was no date in the main page search result (but was in another of a different page)
It's possible that Google gathers the date from the page itself there
heady grove
wiki.gg probably has some custom extension to add tags to the <head>
What makes you think that?
azure kettle
because they have custom meta tags that MediaWiki does not support by default
like for open graph
azure kettle
so it looks like it's just this extension
though I don't discard them using a forked version of MediaWiki, just not as drastic as Fandom's
heady grove
As far as I’m aware — and one of the devs and the wiki.gg project lead are both former colleagues of mine and we keep in touch — they’re not using a forked version of MW.
gray mountain
Last I saw, they were running an obsolete version of MW but it seems they upgraded, neat
also heh, they use CosmicAlpha's PortableInfobox extension!
stoic garnet
it appears for wiki.gg, so it's more than likely Fandom that have removed that f...
ahh I see
It's interesting that for a lot of wiki.gg wikis the date only seems to show up on google, not bing or ddg :o
thanks for the investigation! We'll have a root around with this info; it definitely seems infeasible to easily affect at the moment
stoic garnet
Since what we've been trying to do since is "hacking" the `dateModified` value t...
If it was thought a wikiseo update might help there I'd be happy to request it as well!
That would specifically be... requesting the option for "dateModified" ( and "datePublished"?), in application/ld+json, to not be added to specific pages through a parser function, I suppose?
radiant scarab
Yeah, I believe that would fix it. Not adding those parameters at all
stoic garnet
Sounds like a good step, I'll write that up now :)
gray mountain
Just to confirm that I have the right idea? :P
You do but that's the wrong Phabricator
We can't do anything here on Miraheze
You'd need to ask the developer of the extension to do that on Wikimedia Phabricator
stoic garnet
Ah I see!
https://phabricator.wikimedia.org/T317921
Submitted :) thanks for the advice
radiant scarab
Looks great! Thanks for submitting that
quaint geyser
I noticed that, hate this shit 
so, what's the solution w/ dates, or there's no solution?
heady grove
so, what's the solution w/ dates, or there's no solution?
There’s no solution.
radiant scarab
Well removing WikiSEO would do it. But it's not a good idea 😆
heady grove
Well removing WikiSEO would do it. But it's not a good idea 😆
No it wouldn’t? WikiSEO doesn’t cause the dates to appear.
quaint geyser
any solutions atm?
same question was just before yours ...
shy bone
ah
radiant scarab
No it wouldn’t? WikiSEO doesn’t cause the dates to appear.
Oh it doesn't? Sorry I got an impression it does since documentation mentions setting the data... my bad
robust palm
?
