Timeline for Improper use of rel=canonical is hurting search
Current License: CC BY-SA 3.0
20 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 18, 2021 at 12:13 | history | edited | CommunityBot |
replaced http://data.stackexchange.com/ with https://data.stackexchange.com/
|
|
| Jun 3, 2020 at 13:30 | history | edited | CommunityBot |
Commonmark migration
|
|
| May 23, 2017 at 12:36 | history | edited | CommunityBot |
replaced http://stackoverflow.com/ with https://stackoverflow.com/
|
|
| Mar 20, 2017 at 10:30 | history | edited | CommunityBot |
replaced http://meta.stackexchange.com/ with https://meta.stackexchange.com/
|
|
| Oct 21, 2015 at 11:00 | comment | added | John Dvorak |
Questions with plenty of answers are par for the course on PPCG. They might even be worth searching for (eg: answer-chaining challenges, cops might want to search for their robbers and vice versa). Should rel=canonical be disabled on a per-site basis?
|
|
| Apr 24, 2014 at 13:34 | history | edited | CommunityBot |
Migration of MSO links to MSE links
|
|
| May 2, 2012 at 16:38 | comment | added | Arjan | @Ilmari, indeed. I reported it to Nick in some comment on a post about that very robots.txt. | |
| May 2, 2012 at 12:48 | comment | added | Ilmari Karonen |
@Arjan: Oh, wow. That "page 37273" thing does look like a silly thing to index, particularly since it will be stale just as about as soon as it's indexed. Perhaps we should add Disallow: /questions?page= to robots.txt? (I assume we have proper sitemaps etc. so that Google will find our posts without having to keep recrawling all those pages.) Would probably improve crawl performance too...
|
|
| May 1, 2012 at 21:50 | comment | added | Arjan | Yes, you might be right that those "hottest", "newest", ... most often link to tag pages. Those tags might not always be in my search term though; see the long list at site:stackoverflow.com webkit disable retina sprites, which incidentally also shows Newest Questions - Page 37273 (!) on page 3. Unless I report back later, I guess most are tag results indeed, which I understand is okay to you. (Though not useful for me, as often outdated when I click such link.) | |
| May 1, 2012 at 11:37 | comment | added | balpha StaffMod | @Arjan Hmm yeah, that one actually makes some sense... At least that's filtered by tag; I thought you were talking about the unfiltered active/hot/etc. view | |
| May 1, 2012 at 10:53 | comment | added | Arjan | @balpha, searching for webkit disable retina sprites just gave me Newest 'retinadisplay' Questions - Stack Overflow as the first hit on Google.nl, and as the second hit on Google.com. Still not an example with many hits like that, and also a bad search query I guess. Still, just in case it helps. | |
| Apr 30, 2012 at 14:19 | comment | added | balpha StaffMod | @IlmariKaronen That last one is a great point, thanks for noting. | |
| Apr 30, 2012 at 11:06 | comment | added | Ilmari Karonen | Ps. I just looked at the robots.txt link you gave and realized that we have a much bigger SEO problem there. (All the short permalinks we generate for sharing point to disallowed URLs.) But that's for another post, let's not discuss it here. | |
| Apr 30, 2012 at 11:03 | comment | added | Ilmari Karonen | The online store case seems pretty analogous to me; yes, we have questions and answers instead of categories and products, but to Google they're all just text. Anyway, if you're worried that rel=next/prev won't work properly for our use case, we could a) just ask Google, or b) provide a "view all" page and point the canonical URLs to it, which would solve the problem without any risk of duplicate content or irrelevant search results (and which is what Google recommends as the preferred solution anyway). | |
| Apr 30, 2012 at 10:56 | comment | added | balpha StaffMod |
@IlmariKaronen "which Google has explicitly said is improper and not supported" -- No; as I explained, we're not even the use case that Google talks about, and as I explained as well, we're using rel=canonical to point out duplicate content, which is precisely what it's there for.
|
|
| Apr 30, 2012 at 10:44 | comment | added | Ilmari Karonen | Besides the one problem you wrote this would solve, there's another one which I didn't explicitly mention in my post, since it's not an acute issue yet: we're currently using rel=canonical in a way which Google has explicitly said is improper and not supported, and which can be (and AFAIK has been) used by black hat SEOs to game search results. Yes, there's a (small) risk, as always, that we may lose ranking if we change something; however, there's also a (small) risk that we may lose ranking if we keep doing the same old wrong thing and Google eventually decides to get tougher on it. | |
| Apr 30, 2012 at 10:36 | comment | added | Arjan |
I've been browsing through my history before posting the above, but I guess I mostly end up there at work -- and I'm not there right now. And yes, Google knows I love SE as I use site:stackoverflow.com ;-) Not the best example, but: mvn repo disable ssl site:stackoverflow.com shows a few too. I'll post here, if I find a better one. But so far I did assume that my searches were just bad; mostly cases where I did not really know where to start searching...
|
|
| Apr 30, 2012 at 9:50 | comment | added | balpha StaffMod | Interesting, I hardly ever see that these days. I would think that in such a case, Google would usually give you the particular question page, not a list that just contains the title. Do you use Google while being logged in? May they noticed you love SO and desperately try to give you results from there :) Examples would certainly be interesting. | |
| Apr 30, 2012 at 8:41 | comment | added | Arjan | Very, very insightful! Maybe related: are the "hottest", "newest", "recently active", "unanswered" and "highest voted" results in Google simply caused by Google crawling the site, or do they have some special search merit too? (Whenever my searches are too specific, I often get loooong pages of references to such --outdated-- questions lists in the search results, instead of references to questions. Probably as then my combined search terms are only found in combined, unrelated question summaries that happened to be on the same "hottest '..' questions" list, some day.) | |
| Apr 30, 2012 at 7:19 | history | answered | balphaStaffMod | CC BY-SA 3.0 |
