Smaller issues

Okay, you’ve got a fall “search weather” forecast, and it’s mostly quiet. Most of the recent infrastructure changes are out at most/all datacenters. What now?

Well, I thought it might be a good time to collect potential bug reports (not general comments, not questions, and not spam reports). If you know of a query that appears buggy, post it in a comment here. I’ll start the ball rolling. 🙂

– high site: results estimates. I believe that more accurate site: results estimates are live everywhere now.
– [site:mattcutts.com -inurl:blog -inurl:files] returns a result or two with “blog” in the url from Supplemental results. This is a known issue that’s pretty far off the beaten path (NOT terms in the url in Supplemental Results is pretty much the definition of off the beaten path query-wise 🙂 ), but the supplemental folks have promised to look at this.
– I think two people noticed a difference for queries at 1-2 data centers. One person was someone that watches Oxford very closely. The other person was Danny Sullivan, so I heard about it :). I believe this is fixed/working now. The differences for the queries was because we were sending out a newer binary/executable and different data centers had different versions of the binary. Every data center has the newer binary now.
– For a brief while last week, site: only returned three results from a host. Someone mentioned it to me by email, but the first web report I saw was by DaveN on Friday (there’s your link, Dave). Fixed/working by the end that day, I think. It was related to a binary/executable that was going out, but a different binary than the one mentioned above.
– results estimates go up if you add a minus/NOT term. This isn’t a huge deal, but there’s a fix ready to go into a future binary/executable push. That particular push might be a month or two down the road.
– link: queries sometimes showed nofollow links on Google. This is a byproduct of the switchover to the newer infrastructure for PageRank in the Google Toolbar, info: queries, and link: queries. I expect in another month or two that we’ll be back to normal, i.e. nofollow links won’t be shown for link: queries.

Just to be clear, pruning will be ruthless for this post: I only want to see specific queries that seem to show bugs, and the more concisely you can explain something, the better. I’ll probably keep just the first example of what looks like a bug. I’ve got a meeting at noon tomorrow to talk about search bugs, so I’ll probably lock the comments after that.

84 Responses to Smaller issues (Leave a comment)

  1. Ooh, I thought of a couple more guidelines:
    – if you can’t give specifics (e.g. a real query), that doesn’t help me. I want real queries.
    – it needs to be something that’s happening now, not in the past.

    So “I do a search for [site:mattcutts.com] and I get some results back from spamhuntress.com” is a great comment. “Two weeks ago, I did site: on my domain and I got results from a different domain. It’s working now though.” is a comment that I will probably delete, because it’s not actionable. Hope that makes sense, and don’t be offended if I prune general comments–use the weather forecast post instead and use this post only for buggy queries.

  2. And I’m not looking for spam reports either. 🙂 See
    http://www.mattcutts.com/blog/seo-mistakes-spam-in-other-languages/
    for how to report spam in different languages, and the very best place to do a spam report is in the webmaster console (look for the “Tools” drop-down).

  3. You may need to prune me because I don’t know if this is a bug or not.

    site:www.doxcity.com

    http://www.google.com/search?client=safari&rls=en&q=site:www.doxcity.com&ie=UTF-8&oe=UTF-8

    Shows two results when the site has several more pages that are search engine friendly and a google sitemap. It’s been this way for months now. The displayed pages are very old.

    If that’s not a bug, please delete me with extreme prejudice.

  4. Nick, this is the sort of thing that I will delete on future posts (I’m ruthless! Ruthless, I tell ya!).

    Check Yahoo! Site Explorer–they claim no inlinks to your site. I only see inlinks from 1-2 urls, e.g. http://www.oma.org/practiceadvisory/Software/comp_vendor.htm

    The fact is that if you want Google to crawl you deeply (more than the 1-2 urls), you do need to have some links. Submitting a sitemap to Google lets us know those urls exist, but sitemaps are also not a back door; if no one at all in the whole web links to your domain at all, Google won’t crawl you as deeply.

    I may go to bed soon, but I will delete most reports like this; I’m looking for bugs that are self-evident from the query results.

  5. Matt:

    I don’t know if it is a bug, but I think it is something that needs clarification:

    By performing a filetype search alone, one would assume that google would display all search entries for that filetype. Instead, without any explaining, it displays nothing.

    This “filetype:txt” search displays nothing:
    http://www.google.com/search?hl=en&lr=&q=filetype%3Atxt

    Adding some search term such as “filetype:txt readme” obviously works:
    http://www.google.com/search?hl=en&lr=&q=filetype%3Atxt+readme&btnG=Search

    Maybe there is a good reason not to allow just isolated filetype searches (filetype:mp3?) but it seems that users should be presented with some reasoning why the search didn’t work.

    Thanks, Matt. Love the blog.

  6. Thanks matt. That makes sense. The client didn’t want to do any kind of link development so it’s his fault… I thought that the sitemaps would at least get it indexed but I was wrong. thanks again for your response. I should be getting to bed too. Please delete both of these posts since it’s obviously not a bug. I thought it was because of the sitemap. My mistake.

  7. Not sure if this is what you want, but I’ll fire ahead anyway. Spotted this a few days ago while looking for a new desk 🙂

    http://www.google.co.uk/search?hl=en&q=diy+shops&btnG=Search&meta=

    Searching for ‘diy shops’ gives ‘see results for homebase’ … homebase is just one of many DIY shops in the UK, so this is probably erroneous.

  8. I can’t believe there’s no pages on the entire interweb with a word like ‘home’ but not ‘house’:

    [~house -house]
    http://www.google.co.uk/search?hl=en&q=%7Ehouse+-house&btnG=Search&meta=

  9. If you’re interested in non web-search problems, this doesn’t work in Google Maps:

    [57°35′48″N, 13°41′19″W]
    http://local.google.co.uk/maps?f=q&hl=en&q=57%C2%B035%E2%80%B248%E2%80%B3N,+13%C2%B041%E2%80%B219%E2%80%B3W&sll=53.488046,-2.252197&sspn=38.371385,82.441406&ie=UTF8&z=4&om=1

    You have to type:
    [57°35’48″N, 13°41’19″W]
    http://local.google.co.uk/maps?f=q&hl=en&q=57%C2%B035%2748%22N,+13%C2%B041%2719%22W&ie=UTF8&z=12&om=1&iwloc=A

    Seeing as that’s something people try and copy from web pages, that ought to be fixed. It’s only because I realised that it was non-standard characters that I could fix it, but other people probably can’t.

    And finally, where is Rockall on that map? It should be there (or here: http://maps.google.com/maps?ll=57.596667,-13.688611&spn=0.3,0.3) … Msn maps have it: http://maps.msn.com/(drotgl45xu0lnl45nb3lgp55)/map.aspx?&lats1=57.596667&lons1=-13.688611&alts1=42&regn1=2 – it’s a travesty to Britishness 😀

  10. My spam protection asks “what is the sum of 3 + 7?” Do I have to answer with a complete sentence? 😉

    Bug description: “link to” link from info:www.somesite.tld includes “garbage” characters and appears not to include all results in the google database.

    Expected behavior:
    1. google.com/search?q=info%3Awww.somesite.tld
    2. “link to” href = google.com/search?q=link:www.somesite.tld/
    3. following “link to” displays all sites with links containing http://www.somesite.tld

    Actual behavior:
    1. google.com/search?q=info%3Awww.somesite.tld
    2. “link to” href = google.com/search?q=link:lvAO9Fqp09kJ:www.somesite.tld/
    3. following link shows only some of the sites with links containing http://www.somesite.tld and there are “garbage” characters between link: and www.

    To reproduce:
    1. google.com/search?q=info%3Awww.somesite.tld to see the garbage characters in the “link to” link
    2. google.com/search?q=link:www.somesite.tld/ to see the list of links for x site
    3. google.com/search?q=link:%3A%3F.somesite.tld/ to see the complete list of links (which includes links containing www, but not listed in step 2).

    Number 3 will show results containing links with www in the href but not in the results from step 2. You can use this with my site (which I’ll refrain from mentioning… just trying to report a problem).

    Gustave Stresen-Reuter

  11. Hiya Matt,

    Thanks for this wheather report, well i have a blunder for u guys, Its not related to current fluctuation of datacenters. Its some thing related to your add Url page, the security code of add url page is not working correctly. And your are getting millions of automated quries 🙂

    Cheers
    TSG

  12. Hi Matt,

    I’m using two computers under the same server ( and the same external IP)

    So when I query Google on these computers I should access the same Google Data center.

    Both configurations (Google settings included) are the same.

    I get the same results for both computers, but the number of results is different.

    For instance in computer 1 the query [search engine optimization] gets 60.300.000 on computer 2 the same query gets 63.500.000

    Is this normal ?

    Regards

    Manuel

  13. http://www.google.com/search?&q=indian+filetype%3Amp3

    mp3 search operator is not working. There are results for other but they are 2 or 3 results from russian domains.
    try for eg: madonnna filetype:mp3

  14. Ok Matt, check this out. I’m googling google for MySQL. The logical result would be to have the page for mysql.com to appear first, but instead I see the wikipedia before it:

    Query:
    http://64.233.161.147/search?hl=es&q=mysql&btnG=B%C3%BAsqueda

    On this datacenter, it appears as it should:
    http://216.239.37.104/search?hl=en&q=mysql&btnG=Google+Search

    Same thing happens if I’m looking for PHP:
    http://64.233.161.147/search?hl=es&q=php&btnG=B%C3%BAsqueda&lr=
    http://216.239.37.104/search?hl=en&q=php&btnG=Google+Search

    Is this what you are looking for?

  15. Matt,

    http://www.google.com/search?hl=en&lr=&safe=off&q=site%3Atickets4u.com+-site%3Awww.tickets4u.com&btnG=Search

    Returns 34k+ listings that should not be indexed, all supplemental however. All of these sub domain pages are using noindex,nofollow meta tags but seem to be included anyway.

    Problem related to the [site] command, googlebot, or my tag?

  16. Nick, now that I wrote the explanation, it’s good to leave it up if you don’t mind–several people might be thinking “Oh, I’ll sign up for Sitemaps and I never need to get any links at all,” and this example would help.

    A Cuttlet, I pruned your comment because you didn’t give a specific site. If you want to recomment with your site name, I’ll check into it.

    Davak, that’s a good question. It’s expected behavior that filetype: queries in isolation right now don’t return results. But I’m going to ask if they can change that; it doesn’t hurt to allow [filetype:doc] for example.

    Ian, that’s our smart query refinement suggestion. It’s saying that if someone is interesting in do-it-yourself (DIY) shops, then homebase is another helpful query. I don’t think this is a bug, just doing what it should do. For example, if you do the query [abc survivor] we’ll show the suggestion [cbs survivor] because the TV show Survivor is really on CBS instead of ABC, and we think the suggestion could help a searcher. See http://www.mattcutts.com/blog/ui-fun-better-queries/ if you’d like to read more.

    Ian, that query returning 0 results means that we don’t offer any synonyms for house. That might not be wonderful, but I wouldn’t consider it a bug–we might be requiring a hard match of the original word, and subtracting it removes the soft synonyms too. Yeah, check out [~help -help]. If you just do [~help] you’ll see that support is a synonym for help, but adding “-help” to the query points out that we require a hard match for help and subtracting that leaves 0 docs. That’s my supposition, at least. In the past, it was possible to do queries like this to find synonyms; I’ll ask if it’s intended for [~word -word] to return zero results now.

    Ian, I like the Maps suggestions, especially the first one. I’m passing that one on–thanks! 🙂

    Gustave Stresen-Reuter, I almost deleted this one because it didn’t talk about a specific site, but I think I can answer this. There are two things. #1, “link:lvAO9Fqp09kJ:www.somesite.tld/” those characters are expected behavior. It looks like gobbledygook, but it saves Google some lookup time. You also see them for cache: queries, and they’re not bugs and definitely nothing to worry about. #2, “google.com/search?q=link:%3A%3F.somesite.tld/”, once you unescape it, is like doing the query [link::?.somesite.tld/]. See my answer to Hagrin below, but this is not the correct way to do a link: query that matches multiple subdomains. In fact, you can tell from the results highlighting in the results page that it’s doing the query for the word link plus “somesite.tld”. That’s very different from a link query; Google only supports link: queries for specific urls, not for wildcards like ‘?’ in the url.

    Emma, interesting post. I think it’s known/expected that filetype:doc will return urls that end in .doc, even if the file isn’t a Word file. That’s why you should never name your web documents something.exe . 🙂 I’ll ask someone to be sure though in case that’s new behavior.

    The SEO Guru, I just tried google.com/addurl.html, and the captcha and form worked fine for me.

    Manuel SEO, not a bug. Even two computers running from the name IP address (e.g. on a home network and a router doing NAT) can hit different data centers, which can account for the very slight differences in results estimates. Or even if they hit the same data center, one data center might be really busy and send some overflow to a different data center.

    Arun.T, Google doesn’t support mp3 searching with the filetype: command. Any results you find for [madonnna filetype:mp3] are not likely to be actual mp3’s. 🙂 I’ll mention this one though.

    Jojo, I’ll ask about this one. If a .in domain has a German IP address, I believe it should be searchable for “Seiten aus Deutschland” results.

    Chris, I pruned your comment because you didn’t give a specific site.

    Luis Alberto Barandiaran, that’s expected behavior. We’ve seen where if someone does a query in (say) Spanish and gets English results, they aren’t as happy. The home pages for MySQL and PHP still appear on the first page, but if there’s good native-language content we want that to be returned highly too.

    TheMadHat, good question. The answer is to click on the “Cached” link and look in the header. You’ll see that these docs were crawled earlier this year, and in the copy we fetched at that time, there were no “noindex” meta tags. You could use the url removal tool to remove those urls for six months, but I’ve seen enough people shoot themselves in the foot that I personally would wait. The next time that Googlebot fetches those pages, we will see the noindex meta tag and drop those pages.

  17. (I’m just a Webmaster not an SEO guy, so perhaps I’m raising an issue that will be obvious to most. Links to an explanation are welcome!)

    For this query, it’s best to show 50-100 per page.

    link:prefab.com returns about 3 dozen from prefab.com and then 1 dozen from elsewhere: http://www.google.com/search?hl=en&lr=&q=link%3Aprefab.com&btnG=Search

    Now I want to only see results from elsewhere, so I add -site:prefab.com, but that “did not match any documents”:
    lhttp://www.google.com/search?hl=en&lr=&q=link%3Aprefab.com+-site%3Aprefab.com&btnG=Search

    I think it’s “always” been this way, which suggests that I may not understand what is “supposed” to happen for “link:prefab.com -site:prefab.com” — but it sure doesn’t do what those terms lead me to expect. If there’s another way to find “links to prefab.com except from the site prefab.com” please let me know!

    Thanks in advance. (And, thanks to Matt for a great blog!)

  18. I’m really not sure what this counts as, but it comes with the benefit of a little insider information I gained after this search:

    http://www.google.ca/search?hl=en&q=Canon+MF5750+PCL+Driver&meta=

    which led to this in the #4 spot:

    http://72.14.205.104/search?q=cache:r8S9cDGrzTwJ:tech.yahoo.com/ps/canon-imageclass-mf5750/1991722885+Canon+MF5750+PCL+Driver&hl=en&gl=ca&ct=clnk&cd=4

    The reason this is odd is because of this part here:

    These terms only appear in links pointing to this page: pcl driver

    As I discovered afterward (thanks to a subsequent call to Canon Canada), the MF5750 doesn’t even have a PCL driver. Canon actually releases proprietary printer drivers.

    So I can’t see why anyone would link to that page containing the phrase “PCL driver” unless either it’s a bug on the Yahoo! site or Google itself. (Mind you, stupid people do stupid things like look up PCL drivers for printers that don’t have them.)

    I’m reporting it just in case it’s the latter and in case there are other, larger, and much less obscure queries than the crazy stuff I search for and that generate the same type of results. If this is not the case, feel free to blast this post. I won’t take offense. I have enough posts on here anyway. 🙂

  19. A site query for http://www.battlefocused.com returns the https version of my homepage. Is this a bug or is there a reason for giving preference to secure urls?

    http://www.google.com/search?hl=en&q=site%3Awww.battlefocused.com&btnG=Google+Search

    The result seems to be that the site disappears from SERP; dropping from 2nd page to the hundreds. Ouch!!!!

    BTW: Do you really ever sleep? 🙂

    Thanks,

    Robert

    P.S. I’ve recently added .htaccess 301 redirects to http after this started with the new infrastructure update.

  20. Above post continued… Here’s the filter=0 similar pages link which shows the http version second after the https version (at the time of this post).

    http://www.google.com/search?hl=en&q=site%3Awww.battlefocused.com&btnG=Google+Search&filter=0

    Sorry, if this wasn’t what you were looking for!

  21. Hello there,

    I am uncertain how relevant this is to the topic but I am having the same problem with our web site’s inbound links – when initiating link:www.netfreez.com results have been fluctuating in the past week from either displaying 17 backlinks and PR3 for root or 132 backlinks and a PR4 for domain root. Am I to assume that the update is still under progress and that is not yet completed!? Please advise Matt, thank you again.

  22. Scott Lawton, that is expected behavior. On Google, link: queries can’t be mixed and mashed with other modifiers.

    Multi-worded Adam, I didn’t see which link had this anchortext at first glance, so I’ll ask someone about it.

    Robert, sounds like you’re doing the right thing with the 301, but I’d happily ask someone about this.

    Netfreez, this is expected behavior. I talked about this at http://www.mattcutts.com/blog/fall-weather-forecast/ and said “The new infrastructure is live at about 2/3rds of data centers, and I’d expect it to roll out to all data centers within a month or two (again that’s a hope, not a promise). In the mean time, you may see some differences in PageRanks in the Google Toolbar depending on which data center you happen to hit.” So you’re hitting the old vs. new infrastructure for PageRanks in the Google Toolbar and link: queries. That’s what’s causing the difference.

  23. [therapy products] still results in a mid-page “see also” section for Yahoo. That’s been around for 3+ months, so I’m starting to think maybe that’s just the way it is 🙂

  24. Hi Matt,

    Here is a bug that appeared I would say end of August 2006. I submitted a bug report, but….

    The query contains ‘/’ in the inurl pattern
    [site:www.alapage.com inurl:/Fiche/Informatique/]

    It shows 794 000 results, while the real number of these URLs is more likely to be 50 000.
    However, while results on first pages can look OK, on later pages it doesn’t respect the inurl criteria anymore.
    Have a look :
    http://www.google.com/search?q=site:www.alapage.com+inurl:/Fiche/Informatique/&num=100&hl=fr&lr=&safe=off&start=400&sa=N

    If you try it differently (some sort of workaround):
    [site:www.alapage.com inurl:Fiche inurl:Informatique]
    http://www.google.com/search?q=site:www.alapage.com+inurl:Fiche+inurl:Informatique&num=100&hl=fr&lr=&safe=off&start=800&sa=N
    It looks much more correct.

    Thanks for your attention to this,
    Regards

  25. Matt –

    Hey, I was wondering if you could answer a question concerning the difference between:

    link:www.domain.com and
    link: http://www.domain.com (notice the space)

    Now, I would think that a space wouldn’t make a difference at all, but for my domain (check my name for affected domain), the first query returns 123 links and the second query returns 387.

    Is there a technical reason for this and which number should I be more concerned with in terms of Google evaluating links coming into my site?

    Thank you as always,
    hagrin

  26. Erik Dafforn, I’m happy to ask about this one.

    Open Keywords, I believe this is expected behavior. We support e.g. [site:mattcutts.com/blog] but I don’t believe we specifically support queries like [inurl:directory/path]. I’ll pass on the request though.

    Hagrin, the second is looking for the word “link” and a mention of the domain, and has nothing to do with backlinks. If you want to see links to a page, don’t insert the space.

    Okay, it’s past 12 p.m. and the bug discussion meeting today is long over, so I’m going to lock this thread.

    I was happy that the questions were really high-quality, and I was relieved that leaving this up overnight, I still didn’t get a flood of queries that showed bugs. 🙂

  27. The option to “open links in a new window” opens links in a named window (“nw”), not a new window. Anyone that’s tried to use that preference has experienced “lost windows” when suddenly the next Google search result they click ‘eats’ the previous result window. Not nice at all.

  28. It’s not a bug as such, but it’s a feature I’d like to see added.

    When (domain.com) or (domain.ca) is typed into the Google toolbar, the toolbar attempts to take the user to a domain rather than show results. Ideally, I’d like to see a toggle for this since there are reasons some of us type domains into toolbars and not want to go to the sites (in my case, primarily “why is my site banned in Google?” type of requests.)

    It does, however, lead to a rather obscure bug. Type in Proxy.25.ar into the toolbar (a virus name) and you’ll see what big G tries to do. Maybe in cases like this (where no domain matches), results could be shown instead?

  29. Note: I just wanted to add that I understand and accept the rationale behind the behaviour (i.e. that most people search for a domain name, then click the first result to go to the site.) I just want to see a toggle, that’s all. 🙂

  30. I’m pretty sure this doesn’t count as bug, but I’ll mention it anyway:

    When I hover over a Google result URL, let’s say I searched for [yahoo] and the top result is http://www.yahoo.com, the browser’s status bar will read “http://www.yahoo.com” as expected. Now when I right-click the link to copy it to the clipboard, the clipboard will only contain that huge tracker URL Google uses (something like “google.com/url?sa=t&ct=res&cd=1&url=http%3A%2F%2Fwww.yahoo. …” )… not as expected.

    Other than that, one of the constantly weirdest* things on the Google homepage to me is that the behavior of clicking a navigational link after having entered a search query is inconsistent; sometimes it shows the direct result of whatever you entered in the other search engine (this is good, and it’s what happens when you click e.g. News), at other times it only redirects you to the other search engine’s homepage (this is bad, but it’s what happens when you click e.g. Images). Don’t know if this counts as search bug either 🙂

    *Weird as in “just why doesn’t Google fix this? You’d figure the Google homepage is the most important page for Google, where even details like these take on enormous importance…”

  31. I doubt this is a bug, but I have been finding that the site: command has been showing an extremely small amount of pages indexed as of late. This is obviously a global filter change; however it seems not to be closely indicative to the actual amount of pages on the site.

    Happy New Year

  32. Don’t know if this is a bug, but it seems like unwanted behaviour to me:

    When searching for Yahoo, like this: http://www.google.com/search?ie=utf8&oe=utf8&q=yahoo, i get yahoo.com with some sitelinks, the first one being mail.yahoo.com, that however, is also the second search result. The same happens here:
    http://www.google.com/search?ie=utf8&oe=utf8&q=apple, where store.apple.com is among the site links, and is also, atm, the third result. It can even go as far as two results being duplicated, like here: http://www.google.com/search?hl=en&lr=&safe=off&q=w3c&btnG=Search, the last query also shows that it doesn’t just happen for subdomains, but for subdirectories as well.

  33. Bit of a mystery bug and not very useful to fix but buggy/not expected nontheless:

    http://www.google.co.uk/search?q=**************a

    This search (any number of ***’s followed by an alphanumeric character) returns a set of results with randomly bolded phrases. It does so consistently (refresh the SERP) and if you change the number of asterisks you get a different result set.

    Returned document counts are far into the billions and it doesn’t seem to be a wildcard search of sorts.

    This has been bugging me so much, I can’t sleep. Please fix it SuperMatt! 🙂

  34. Here’s another one. I explained it with screenshots in this Digital Point thread: http://forums.digitalpoint.com/showthread.php?t=201015

    The ‘bug’ is that Google is creating a confusing user experience by causing duplicate ‘content’ or results, however you want to call it. IMO if a page is already featured at #1 in an extended/expanded ‘authority’ result, it should not again be shown under #2. At the very least it should be indented like it would if the expanded authority result wasn’t shown like normal.

  35. Here is a good one Matt:

    Google search is serving up pop up images in search results.

    Go to google.com

    allinurl: popup_image pID=

    Click on some of the results and they will take you directly to a pop up image.

    I have noticed this a few times in December while doing some real long tailed google searches. Sometimes they are showing up in the regular search results where they should be showing up in images.

    Its very frustrating because in internet explorer when you click on one of these results, it resizes the window.

  36. This one’s been around for a long time:

    If you search for “link:www.everylastpenny.com” Google returns 26 results.
    If you search for “link: http://www.everylastpenny.com” Google returns 173 results.

    Note the space in the 2nd search. I’ve always wondered about that one…

  37. Read all the comments, Chris S, and you’ll notice Matt’s answer.

  38. In the last few weeks for us, the info: command has been returning the wrong site.

    [info:www.clicksports.co.uk] brings up for results for http://www.stadia-sports.co.uk instead.

    It’s currently working ok using google.co.uk, but I did just reproduce this using 66.249.93.107

    The two sites are hosted on the same server. Stadia is the parent company of Click. So I guess it’s possible that it’s some kind of server problem? The link: command also returns links for Stadia.

  39. Not a bug, but I came across this really old (read: outdated) biography page on Google.com recently:
    http://www.google.com/intl/ca/management.html

  40. Matt or Adam please delete this post once you’ve checked it. Here’s what was happening in that hiccup pre 19th Dec. Sorry I wasn’t more deligent in capturing the DC.

    http://www.search-engine-war.co.uk/odd-google-results.doc

  41. …and I guess you already now about all the geo-location stuff being sub-optimal 😉

  42. Something which I have seen for a while now (at least 1+ years): sites which are indexed with a space, tab or quote before the domain name, eg:
    %20www.domain.com
    %09www.domain.com
    %22www.domain.com

    Google seems to index those URLs – probably from bad html code (who said getting W3C validated pages wasn’t worth the time :-)). Google probably strips the space,tab,quote from the host name before doing the DNS lookup and crawls the site normally, only with the wrong host name.

    The problem with those URLs is that most browsers don’t know that they can strip the prefix before doing the DNS lookup, meaning that they will never reach a URL which is listed in the index like that. Google is showing URLs which are not reachable by clients, but Google continues to keep them (since it doesn’t notice that they’re bad).

    My suggestion: if you are going to fix them before a DNS lookup, why not fix them before they’re indexed and keep the correct URL in the index. Yes, I know with enough good links you’ll never see those URLs, but some sites have precious little link value and get stuck with those listings (and it’s not clean either – crawling something different than is indexed).

    It seems to have gotten better since I started mentioning this, but it’s still not optimal, there are still sites being listed that way.

  43. http://www.google.nl/search?q=webcam+sex&hl=nl&lr=&rls=GGLG,GGLG:2006-21,GGLG:nl&start=20&sa=N

    http://variatee.vara.nl/daarvliegendepanters/gastenboek.jsp

    Sorry to post a adult term but i don’t think there are any BL to this page, so it is just based om onpage text. Bug?

  44. I dont’ know if this is a bug in the software or the wetware, or even a bug at all, but how did Google end up putting ‘romantica’ in the catagory “family friendly”?

  45. Search for “phonitick spewling” and the first result I get is a link to the same query. WEIRD. Something to blog about…

    And why no spelling suggestions?

  46. Slightly bug-like: Why index robots.txt and sitemap XML files? How would you block indexing of a robots.txt or sitemap file?

  47. This has got to be a bug: search for [robots.txt] and get pubcon.com?

  48. Nitpicking: Google calculator should uses the decimal separator specified in the query, instead of the correct one for the language used, but only in some cases e.g.

    “19,27 eur in usd” => 19,27 Euros = 25,555874 U.S. dollars
    but
    “19,270 eur in usd” => 19 270 Euros = 25 555.874 U.S. dollars

    Negative amounts are not displayed consistently either:

    “-19 eur in usd” => (-19) Euros = -25.1978 U.S. dollars

  49. One thing that has always bothered me:
    link:domainname does not include pages from that domain.

    The help pages state that the “query [link:] will list webpages that have links to the specified webpage.”

  50. Hello,
    I’m continuing to see some odd behavior when searching for terms related to some sex/sexuality sites. Specifically, it appears to be happening when you search for terms related to the business/company/site name. Examples of queries that one would expect would return a particular site in the SERPs:
    good vibrations (goodvibes.com)
    my pleasure (mypleasure.com)
    adam eve (adameve.com)

  51. This one is probably not a bug, but definitely a search quality issue:
    http://www.google.com/search?q=apartments&num=30

    16 out of 30 top results are local (websites offering apartments only in specific cities). Shouldn’t they be filtered out, so that users only see those local websites that match user’s location? E.g., if I’m located in LA area, ‘Apartments’ section in the LA Times might be of interest, but what’s up with ‘Orlando Sentinel’ or ‘Miami Herald’?

    Thanks.

  52. Not technically a bug, but misleading: the results for a link:-query should mention that the items shown may not show everything which Google knows about 😉 . Many newbie webmasters get confused by that.

  53. Hi Matt

    The “index page missing bug”, on regional searches.

    Eg. search for site: on google.co.uk and select UK pages, index page goes missing, search the web and it’s there.

    Threads have been going on this for months on WMW, http://www.webmasterworld.com/google/3051237-7-30.htm , sure you’ve seen them.

    Some update on this would be good, it’s been going on a while, problem also comes and goes for sites.

    People from other countries are also reporting the problem, so it’s not just a UK regional issue.

    JB

  54. Hey Matt,

    It’s great you’re doing this! I’m sure it’ll keep you busy for a while 🙂

    My issue is with one of my domains. It happened a few weeks ago, just after the site was released from the new domain penality (sandbox or whatever you’d like to call it).

    The site ranks fine, but a site:www.suzuki-gsx-r.com shows no results. Using Google webmaster tools mentions that the site is not in the Google index. Not sure if anyone else is having the same problem or not, but it did scare me at first 🙂

    Thanks 🙂

    John

  55. Actually I just remembered another bug/issue I’ve noticed, but feel sort of guilty mentioning it as I’m sure people are getting a lot more traffic for it. It’s related to localised traffic (ie. google.com.au).

    if you search for http://www.google.com.au/search?q=southern+highlands

    you’ll notice the first 2 results are from the same domain, but the second result is already included as part of the first result. it’s one of those authority type results with the main link, and then the 4 links beneath it. not sure what you’d call them!

    not sure if it’s a bug, but I’ve noticed it happening on the odd occasion. I’m assuming it’s due to the site not ranking 1st in the global google.com, but when moved up for localised results, it continues to keep the 2nd result even if it’s included in the authority links below the first result.

    Thanks again 🙂

    John

  56. not sure if this is ok only, but doing a search for sprite with safesearch one shows a topless woman as a “sprite”

    http://images.google.co.uk/images?q=sprite&ie=UTF-8&oe=UTF-8&rls=org.mozilla:en-GB:official&client=firefox-a&sa=N&tab=wi

    i wanted the drink not the girl

  57. The search is

    j * hochman

    The first result I get is this page: http://www.amazon.com/gp/pdp/profile/A34AB1FGI1PVD2

    The only visible matching text is “Marla J. Hochman’s”

    Google’s cheat sheet at http://www.google.com/intl/en/help/cheatsheet.html
    says “red * blue — the words red and blue separated by one or more words.” In the example, the keywords are not separated.

  58. It might not a bug, just (my own) bad syntax using the currency conversion operators…

    But converting US Dollars into Australian Dollars…

    [100 USD in AUD] works fine
    http://www.google.com/search?hl=en&lr=&q=100+USD+in+AUD&btnG=Search

    But variations on this don’t work well…

    [$100 USD in AUD]
    http://www.google.com/search?hl=en&lr=&q=%24100+USD+in+AUD&btnG=Search

    [USD100 in AUD]
    http://www.google.com/search?hl=en&lr=&q=USD100+in+AUD&btnG=Search

    [USD$100 in AUD]
    http://www.google.com/search?hl=en&lr=&q=USD%24100+in+AUD&btnG=Search

    Usually Google is smart enough to pick up variations like this.
    But the problem might just be EBKAC (Error Exists Between Keyboard and Chair). 😉

    Brent

  59. Typing a series of different queries into Blogsearch produces a run-time error like this:

    http://www.google.com/sorry/?continue=http://blogsearch.google.com/blogsearch%3Fhl%3Den%26ie%3DUTF-8%26q%3D%2522matt%2Bcutts%2522%26btnG%3DSearch%2BBlogs

    I get that message if I change the sorting sequence to sort by date and then click on the second page.

    Or I can get it if I change the query string (using quotes for Exact Find).

  60. http://www.google.com/search?num=100&hl=en&lr=&safe=off&q=%28%22a%22+OR+-%22a%22%29+***

    Yields 18,310,000,000 results.

    http://www.google.com/search?num=100&hl=en&lr=&safe=off&q=%28%22a%22+OR+-%22a%22%29+***+-dkcbhdhgftred

    Yields 18,140,000,000 results. That means based on those queries there are 170 million pages referencing the phrase “dkcbhdhgftred”. However…

    http://www.google.com/search?hl=en&q=dkcbhdhgftred&btnG=Google+Search

    Yields no results. I know, small numbers, but still, a .928% fluctuation there.

    -Michael

  61. I noticed a bit of a bug while trying to construct an example of the maps finding -> directions problem I’ve been having:
    http://maps.google.co.uk/maps?f=l&hl=en&q=train+station&near=bristol&ie=UTF8&z=15&om=1
    It’s misread a website listing pubs near train stations to actually be train stations, and so there isn’t actually a train station listed in there.
    Same thing goes if you enter the station name, “bristol temple meads”.

    The bug in maps I was going to report appears to have been fixed since I last used it (within the past few weeks), where entering the a location then trying to get directions to it didn’t work. That’s been annoying me for months, so good job whoever finally fixed that 🙂

    Still no Rockall on the map! When will this travesty end? 😀

  62. I’ve also had problems with the calculator/converter. Here’s some more examples:

    This works (without or without ‘s’ on the end):
    £200 in singapore dollar

    But these don’t (some perhaps should not but some should):
    £200 in $sin / £200 in $sing
    £200 in $singapore / £200 in $ singapore
    £200 in singapore$ / £200 in singapore $

    Also, with the Slovenian Tolar being phased out in favour of the Euro from 1st Jan (for 2 weeks iirc), I thought I’d play around with that too:

    [slovenian currency]
    http://www.google.com/search?hl=en&lr=&q=slovenian+currency&btnG=Search
    Gives me the shortcut:
    “Currency: Tolar (SIT)”

  63. Matt,

    I know you probably know about this already but the site: command is certainly screwy. Supplementals are shown before normal pages. The home page doesn’t appear until the last few pages.

    Example:
    http://www.google.com/search?q=site:www.gardenstew.com

    I know it’s not a query that is used by Joe Surfer but surely something is amiss if it displays like so.

    Thanking you! Happy new year.
    // Frank

  64. The info: command sometimes matches up the wrong domains.

    http://www.google.co.uk/search?hl=en&q=info%3Awww.initial.com&meta= doesn’t reveal info about the right domain (but a related domain). I have no other example of this happening.

  65. Site: operator returning a few of thousands.

    I have my preference set to 100 results.

    I believe this is the same as above “- For a brief while last week, site: only returned three results from a host.”

    Details:

    Query = site:cybersmusic.com

    http://www.google.com/search?num=100&hl=en&lr=&newwindow=1&q=site%3Acybersmusic.com&btnG=Search

    Results 1 – 5 of about 126,000 from cybersmusic.com

    In order to show you the most relevant results, we have omitted some entries very similar to the 5 already displayed.
    If you like, you can repeat the search with the omitted results included.

  66. I do not know if it is a bug or not, and I’d rather not share the site involved publicly, but I’ve noticed a page of mine still getting PR through a nofollow link. Or atleast, so says the toolbar.

  67. The definition searc feature does a blind lookup of the keyword without giving importance to the case of the keyword.

    e.g.
    define SIP
    – I get a search result saying “a small drink”. What I was really looking for was the description of Session Initiation Protocol.

    define LAMP; defines a lamp, not the open source web platform.

    “define:” operator does give a list of definitions, but I would really like to see search interpreting the abbreviations properly.

  68. This query:
    213.253.92.115 -inurl:stat

    Returns results with “stat” in url.

    Example:
    http://www.cemi.cg.yu/english/statistika/show_detailed.php?lng=cg

  69. As a C# programmer, I’m constantly searching for queries that contain “.NET” but Google doesn’t seem to notice the very important “.” in front. I usually resort to changing “.NET” to “ASP.NET” or “C#” to get the results I’m looking for.

    http://www.google.com/search?q=.net+commands

  70. It kind of have to do with the Search results but also with the Sitemaps in the web master tools.
    When I set my preferred domain to use www then for some reason non-www pages started to show up, something that never happened before. Now I don’t know what to do. It seems that the numbers of visitors have gone down noticably, but I don’t know if that is the reason.
    Hoping that it haven’t affected my rankings and other stuff.
    Like I said, not sure if it is a bug in the search or not but needs to be looked at.

  71. Hi Matt –

    There’s a strange bug in calculator, when summing two temperatures of different units:

    1f + 1c

    Apparently, the result is 256.927778 Celcius!

    Cheers,

    Andy

  72. Hi!

    Chris S said: This one’s been around for a long time:

    If you search for “link:www.everylastpenny.com” Google returns 26 results.
    If you search for “link: http://www.everylastpenny.com” Google returns 173 results.

    ?some one saw: that the in second example google

    takes link: www. as link%3A+www.

    where comes that + from?

    but it explains why link: www.
    and link+www. brings up the same numbers of links

    Greetings

    Karl Heinz

  73. Have a look at this search:
    http://www.google.com/search?hl=en&q=buy+viagra&btnG=Google+Search

    and let me know if this is what we can expect. Thousands of EDU spam pages. Is this a new Google bomb?

  74. Ok here’s a weird one. Google Webmaster Tools tells me that there are no pages from the site in Google’s index… er, except there are (even though all those except the root page are supplemental). So something there is wrong. I have preferred domain set up if that helps. URL supplied in case it is helpful.

  75. Query:
    (3 celsius + 3 celsius) in celsius

    Result:
    (3 Celsius) + (3 Celsius) = 279.15 degrees Celsius

    3+3=279.15? Not really 🙂

  76. Klaus Johannes Rusch – Actually it’s correct.

    The reason is that 0 degrees Celsius is not absolute zero, that’s zero degrees Kelvin (about -273.15 degrees C).

    So 3 degrees C is actually (273.15+3=276.15) degrees K.

    So twice 3 degrees C is twice 276.15 degrees K which is is 552.3 K.

    Then converting back to degrees C (by subtracting 273.15 degrees) you get 279.15 degrees C.

  77. We’ve got a problem where our doman name is converted into a JIS encoded domain name.

    The query:
    site:wink.com -site:blog.wink.com

    vs.

    http://www.google.com/search?q=site%3A%EF%BC%B7%EF%BD%89%EF%BD%8E%EF%BD%8B.com

    Yeilds a handful of results 8, however the query for the JIS encoded version of our domain name yeilds 287 results. The bug of course is why we’ve got multiple version of our domain name in the system (none of which should be in JIS encoding) and also it would be nice if site:wink.com would pull both domain names (since it’s JIS encoded ASCII).

    There of course is a deeper issue about why we don’t have the ~30,000 pages in the index, but that’s not necessarily a “bug”.

    Thanks.

  78. Searching for ‘diy shops’ gives ’see results for homebase’ … homebase is just one of many DIY shops in the UK, so this is probably erroneous.

css.php