Bugs in WMT.... Ooops

[g1smd]

Suddenly I see:

Crawl errors
Ni bilo mogoče najti 23
Ni uspelo 4
Dostop zavrnjen 3
Napaka v strežniku 22
Programska napaka 404 1
Drugo 0

Eh?

[Whitey]

A bunch of random Vietnamese domains not assoicated with the WMT account are loaded into the drop down select panel. Click on one , i have to re login again, and get taken back to the panel in Vietnamese.

Yesterday it was Thai

Fortunately there is no access to the information on those sites.

[g1smd]

You're in Australia, and it is now showing you languages from nearby countries. Did you ever get the ES-DE-FR-(unidentified eastern European language) rotation?

I'm in the UK and I have mostly been getting languages from nearby countries (as above) but also Arabic on the second day.

[Whitey]

Did you ever get the ES-DE-FR-(unidentified eastern European language) rotation?

Nothing so far.

[g1smd]

Interesting. Today in WMT I have Dutch I think.


Returning to Google search results, if I enter [Toegang geweigerd] I get foreign language results with those words in, and a "tip" to "search for pages only in English" at the top of the SERPs.

Just discovered something "new". If I enter [translate:Toegang geweigerd] I get English language results for "Access denied". That's rather nice, but it doesn't tell me what language my original query was in.

[edited by: g1smd at 9:50 am (utc) on Mar 15, 2012]

[lucy24]

That's rather nice, but it doesn't tell me what language my original query was in.

It's their "detect language" mode.* If you specify a language by name you can see if it comes out the same. Sure looks like Dutch to me. Could theoretically be Afrikaans, but I kinda doubt it ;)

OK, so there's an APNIC random-language group and an RIPE random-language group. (The latter includes the Near East.) What are people getting in the Americas?


* At least here you have the option of specifying a language. The free-translator app won't let you say. This leads to interesting results.

[rlange]

lucy24 wrote:
What are people getting in the Americas?

I'm in the U.S. here and I'm getting Spanish, French, and Portugese.

There seems to be a cycle to it, too, if you refresh the page repeatedly...

* English
* Spanish
* French
* Portugese
* English
* English
* Spanish
* French
* Portugese
* English

etc...

--
Ryan

[g1smd]

Russian!

[lucy24]

Russian!

Or possibly Ukrainian ;)

Isn't anyone here in Africa? Should be an interesting mix of languages there. But they don't seem to distinguish between North and South America.

Now back to sulking and feeling left out. What is the ### variable?

[Andem]

>> Isn't anyone here in Africa? Should be an interesting mix of languages there. But they don't seem to distinguish between North and South America.

In my experience, I can tell you that Google's comprehension of Afrikaans is dismal. I don't know any of the other indigenous languages in Africa, but if Afrikaans is any signal, the whole entire continent has been ignored completely.

[lucy24]

Setting aside the question of whether Afrikaans can really be considered an indigenous language...

Have you looked at the "translate google into your language" area? As far as I can tell, they use whatever text someone chooses to type in. What they consider "worthy of confidence" is anyone's guess. Maybe it means that six consecutive people looked at the first translation and threw up their hands in despair.

In the non-Indo-European, non-Roman-script-using language I've been looking at, there isn't even a consistent file encoding. Some strings come through as html numeric entities. Some seem to have been entered in a legacy font-- which does a fat lot of good if you don't know which one it was, so you can't transcode. Some are purely and simply garbage. I mean literally. No language at all.

E-mail asking specific questions or raising specific points is answered with a form letter referring you to the translation forums-- which have no post newer than 2009. Maybe that's when users figured out how pointless it all is and stopped asking questions; if there's a newer continuation it's very well concealed.

Oops. What was this thread about again?

[g1smd]

Now what?

Hungarian?

Crawl errors
Nem található 22
Nem követett 3
Hozzáférés megtagadva 2
Szerverhiba 19
Lágy 404-es hiba 0
Egyéb 0

Swedish?

Hittades inte 22
Kan inte följa 3
Ĺtkomst nekad 2
Serverfel 19
Soft 404 0
Övrigt 0

Russian?

Ненамерени 22
Непроследен 3
Достъпът е отказан 2
Грешка в сървъра 19
Soft 404 0
Други 0

My 'guess the language' skills are certainly getting a workout.

[lucy24]

I had to detour to Omniglot to get the full list of languages that use Cyrillic. It's Bulgarian.

Still-unanswered question: Is anyone not getting the Multilingual Showoff Bug? What do we have in common?

[g1smd]

The magical mystery tour continues.

Looks like Polish.

Crawl errors
Nie znaleziono 22
Nieodwiedzony 3
Odmowa dostępu 2
Błąd serwera 19
Pozorny błąd 404 0
Inne 0

This thread is going to rank for a heck of a lot of different languages...

[lucy24]

Nieodwiedzony

Did you leave out the space or was that g###'s doing? It gave me much grief ;)

I like "the apparent error 404". Most languages seem to have thrown up their hands and stuck with "soft 404".

Still waiting for someone in the Americas to get

Pa jwenn 22
Pa te vizite twa
2 Aksč Rejte
Erč sčvč 19
Erč a aparan 404 0
lňt 0

but I won't hold my breath on ᐊᓯᖏᑦ 0

[g1smd]

I don't remember seeing a space.

Just looked again and got English ten times in a row. Just as it looked to be fixed, I clicked once more and it was broken with a different language each time from then on.

I also looked in the site-list drop-down (top right) and it listed a whole load of .ru sites that are not mine. I clicked one and WMT lists all my sites but with the rest of the on-screen text everywhere in WMT is now in Cyrillic.

I can't see data from other sites.

/later/ Oh krap. All of the menus everywhere in every Google service are now in Cyrillic for me, even the black bar at the top (with the crawl errors text in WMT in Polish again).

[lucy24]

Possible variables: IP range. OS type. Content of accept-language header. (If and only if you ascribe active malice to a bug ;)) Browser type. Relationship to google (i.e. accounts other than gwt).

:: detour to g### message boards ::

I count seventeen threads asking about the same thing. This suggests that people are not taking the obvious first step of searching (hello? this is the world's #1 search engine we're talking about!) to see whether someone else has already asked the identical question. Heh.

Answers point to
[groups.google.com...]
which says, helpfully, the check is in the mail. Or something like that.

Favorite post, two days into The Problem:

At least until today the system displayed Spanish or English.

Today I'm seeing Spanish, French, German and English.
So if someone is implementing a fix, it appears they're going backwards.

Someone else reported seeing their results in Klingon, but didn't provide a specimen. (Not totally impossible. It's one of the possible Search languages, at least.) One person reported mixed languages (Dutch and Spanish, Dutch and Turkish).

Moral, I guess: Count your blessings.

I thought I'd found out why I'm immune. I have the Dashboard bookmarked-- and my bookmark includes the magic phrase hl=en. But deleting it from the address bar makes no difference. Wouldn't really expect it to, since the rest of the page stays in English.

In fact-- this is entirely fascinating-- if I deliberately set the page to another language, the whole page changes... except the Crawl Errors list, which stubbornly remains English.

[anallawalla]

I'm in Australia and I am seeing Hiragana and Simplified Chinese. No English so far.

[g1smd]

@lucy The Google Groups link is broken for me, probably because I'm still on the old Google Groups settings.

In any case, a Google search for the problem then finds a recent locked thread about the topic which itself contains a link to another thread that also doesn't work for me, and neither does the onsite navigation.

In Larry Page's great quest to synchonise the UA across all Google products it seems like thorough testing and QA has gone out of the window.

[lucy24]

The Google Groups link is broken for me, probably because I'm still on the old Google Groups settings.

Heh. It didn't work for me from here either. Maybe it only works if you are already reading a post in the wmt area, reached via clicking Help and, uhm, clicking a bunch more stuff. Which reminds me that I have yet to figure out how you would find gwt if you didn't know it was there.

If I use my browser's address bar instead of pasting their own link, it comes through as

[groups.google.com...]
instead of
[groups.google.com...]

(That was meant to be an Apache Forums joke, but they botched it by not auto-linking the https version-- which appears to work.)

The difference, besides https:
{snip}#!category-topic/webmasters/webmaster-tools/BVW7aoOmziU
vs.
{snip}#!category-topic/webmasters/BVW7aoOmziU

I'm in Australia and I am seeing Hiragana and Simplified Chinese. No English so far.

Well, you're in the APNIC Bug Group ;)

I discovered that if I set my gwt language to Norwegian (hl=no), one specific link is instead given in Romanian. I really hope this is just a database error and not a foreshadowing of something new. Still looking for the Google In Your Language discussion, though.

[g1smd]

Oh, and the new Google groups layout is insane. Thread list on a netbook has room for three thread titles [twitpic.com...] and only part of the screen scrolls. Horrible!

[g1smd]

Turkish!

Crawl errors
Bulunamadı 20
İzlenemiyor 2
Erişim reddedildi 2
Sunucu hatası 12
Soft 404 0
Diğer 0

[lucy24]

RIPE Bug Group. Check ;) And yup, if I set my page to hl=tr I get the Crawl Errors in English.

[lucy24]

Follow-up: They've changed the blasted thing AGAIN! Now the Dashboard only lists those errors that actually exist. (That is, non-zero value. And yes, Not Found is broken down into 404 and 410 if you go to that subsequent page.)

Still getting English error(s) if I set the page to non-English. And I swear I posted the Romanian/Norwegian thing to the appropriate g### group, but the post evaporated into thin air and I haven't the energy to dredge it up again :(

[g1smd]

The '

crawl stats

' graphs are now in three different colours, blue (pages), red (kb) and green (time). That is new isn't it? I'm not simply imagining that it has changed?

The '

crawler access

' report now includes the "number of blocked URLs". This figure appears to be quite meaningless as it actually just records the number of disallow lines in the robots.txt file which it quotes immediately below.

Today there's a new entry in the WMT '

Pages with missing title tags

' report for a particular site. Nothing strange about that you say; especially when I tell you that the page never had a title tag while it existed.

The reason for querying the accuracy of this report comes from the fact that the page was deleted more than two weeks ago and already appeared in the WMT '

URL errors

' report more than a week ago as having status 404.

[Whitey]

Just raising another WMT bug ..

When submitting a link on "fetch as Googlebot", if you accidentally click a 2nd link during the "pending" process , the submit link process goes into a perpetual refresh attempt without resolving as "submitted to index".

It seems to go on for ages, even if you log out and log back in.

A while back i noticed it took hours to resolve.

[lucy24]

The 'crawler access' report now includes the "number of blocked URLs". This figure appears to be quite meaningless as it actually just records the number of disallow lines in the robots.txt file which it quotes immediately below.

I get a wee question mark leading to a popup that says "Google tried to crawl these URLs in the last 90 days, but was blocked by robots.txt." The number is much higher than the number of lines in my robots.txt. In fact it's higher than was ever listed in the old "blocked by robots.txt".

Maybe they're counting some random fraction of the files that I massively rearranged last year and couldn't be bothered to redirect (they are thin, thin pages). Someone around here brilliantly suggested I robot-them out and let the search engines start from scratch.

Didn't the crawl stats graphs use to be a completely different shape?

Oi! When did Keywords morph into Content Keywords? What other kind is there?

Double oi! A link from your own ### profile is not a ### "link to your ### site"-- especially when they pick two at random and list them in completely different ways.

Wish they'd add a tab for "disavow this link categorically and forever" ;)

:: further investigation ::

Nope. I gotta stop before the things they call "links" send me into screaming hysterics. Just leave the one from my son's site. At least he once looked at the page he's linking to :)

[rlange]

g1smd wrote:
Turkish!

Heh. I've begun seeing this, too. At least now it's only switching between English and Turkish on refreshes.

g1smd wrote:
The 'crawl stats' graphs are now in three different colours, blue (pages), red (kb) and green (time). That is new isn't it? I'm not simply imagining that it has changed?

Same colors as before, just brighter shades. Edit: They also swapped the colors for the kilobytes and time spent graphs.

You can see the old colors by doing a Google Image search for [webmaster tools crawl stats].

lucy24 wrote:
Wish they'd add a tab for "disavow this link categorically and forever" ;)

I'd love an "Ignore all links from this domain" option.

--
Ryan

[g1smd]

I thought something had happened with the colours. Thanks for the clarification.

[g1smd]

Back to Russian yesterday, German and others today.

They seem in no hurry to fix it.

[lucy24]

I'm still getting the Crawl Errors in English if I change the page to another language. Does that happen from your side too?

They tried to tell me the Romanian-in-Norwegian glitch is a manifestation of the same thing, but I still think it's vastly more likely that they entered the wrong string under the wrong language code.

Does anything happen with the word "Labs" in the list on the right? I thought it was just an international-usage thing, but if I go to a language in non-Roman script it still says "Labs" in English. (Russian, Thai, Hindi, Japanese, Hebrew etc.) They managed to translate it into Greek and Arabic, though.