My name is danah boyd and I'm a Principal Researcher at Microsoft Research and the founder/president of Data & Society. Buzzwords in my world include: privacy, context, youth culture, social media, big data. I use this blog to express random thoughts about whatever I'm thinking.

Relevant links:

Archive

quality of Google searches?

Question: has the quality of Google search gone downhill in the last few months or is it just me? Every time i search for an event-based thing, i get crap from 2005 instead of what’s relevant now. I suspect that this is because of the blogger impact on event-type items, but it’s really annoying. It’s also annoying that they’ve stopped correcting my atrocious spelling. I mean, it’s all fine and well that lots of people in the blogosphere can’t spell in exactly the same way that i can’t spell, but the #1 type of search i do everyday is spell check. I throw something god-awful like Cziskentmihalyi into the engine knowing that it’ll return Csikszentmihalyi. This still works quite well for names but it’s stopped working for lots of regular words that i just can’t spell to save my life. How pathetic is it that i’ve started opening up Word for the little red squigglies instead of relying on search? Or maybe both practices are weird…

I’m especially irked that when i search for addresses, they almost never come up. This evening, i searched for “1457 Third Street Promenade.” No dice. I added Santa Monica, CA. No dice. I decided to see if Google Maps would find the address and to my shock, it couldn’t find anything and kept giving me just Third Street generically. I went to Yahoo! Maps (which i prefer in old-skool mode anyhow, but hate typing in addresses to as i’ve noted before) and voila, that worked. I still desperately miss the days when addresses just worked in Google. I can’t believe how many times a day i shove addresses into the searchbar.

Maybe i can convince myself to like the Yahoo! UI now that Google has even further screwed theirs up by cluttering the left-hand side. But i still can’t decide… am i just being old and crotchety about change or has Google’s search actually gotten atrocious?

Grr.. or maybe i shouldn’t switch because Yahoo! wants to give me a support group for crotchedy people rather than tell me that the real spelling is crotchety. At least Google tells me that crotchedy is urban slang, making me feel a bit less crotchety.

Print Friendly

33 comments to quality of Google searches?

  • Yes, agree, I get loads of old stuff too. Strange. Doesn’t seem to be blog content though… Items which were high up the rankings a few weeks ago have disappeared onto the second page even though the content that’s replaced it is v. old…

  • I don’t get stuff cluttering up the left-hand side but I do get Google suggesting spelling mistakes. I can’t get ‘Third Street’ to generate a map but ‘3rd Street’ does. Are you still having it not suggest alternative spellings?

  • Bertil

    If you want spell check, Google provides something in their toolbar which I found better (well: it automatically detects the language and I switch a lot). I’ve noted a few glitches on the international side of things recently, and I’m assuming the spell check for non-names is more difficult with an international user-base.

    Regarding adresses and maps, no issues to me: I get a list of results all over the world in case of an ambiguity, but if the only street of that name is in Italy, I’m send there.

    Note: I’m constantly logged in — I guess it helps, but I would understand some people find it freakish.

  • I really nearly 100% on the Dictionary Widget in Tiger. I’ve noticed that Google’s mapping is pretty atrocious and that most of my general search queries are spitting out spam blogs instead of real content. I’m curious to see whether this will ultimately kill the usefulness of search or if Google can overcome the splog phenomenon. (My money…and probably theirs…is on the latter.)

  • On your Mac, you may find it helpful to hover over the word and tap control+command+d. A dictionary window will pop up. If it’s misspelled, no entries will be found, but clicking More will give you a “Did you mean?” It works for “crotchedy” anyways 🙂

  • Anon

    Firefox’s (2.0+) spell check is my best editor. I also often highlight, then right click, for a quick Webster search. Combined together the approach never fails.

  • I don’t feel so alone now. I have noticed the same thing when I want to look up things like an event or a festival where it gives me results from two or three years ago. I often know the event has an updated site or info, but Google just does not pick it up. Spelling, I have not noticed as much, but I don’t use Google for that as much. Also, you used to be able to look up phone numbers. Not as good for that either as of late. It does make me wonder.

    Best, and keep on blogging.

  • So I’m hearing:
    – if you search for a periodic event (e.g. a conference), you want the most recent one.
    – you’ve found some spellings that haven’t suggested a correction. I would *love* to hear some examples to pass on to the folks here at Google.
    – You want to be able to throw any address into Google and have it detect that pretty well. Thanks for the example, which I’m sure will make it to the right team at Google.

    If you have Google Search History or Web History enabled and can find any example words from the history that we’re not spellchecking correctly, that would be wunnerful.

    Thanks for the feedback; I’ll pass it on.
    Matt

  • P.S. Adding “2007” (or whatever the current year is) should solve the “recent event” issue, but I’ll still mention it to some people that I know are interested in this.

  • Jorge

    I suggest moving on to Yahoo or Live Search. They’ve both become more accurate than Google, especially now that both Yahoo and MS are trying to compete in the search quality market that Google has forsaken.

  • If you hate typing in search engine URLs you should try my search engine LuckyButton which makes this much more useful.

  • Michael Griffiths

    I wouldn’t advise either Live or Yahoo! exclusively.

    I’m using IE7, and will switch between Live, Google, and Yahoo! with the handy search box. I’ve noticed that Google results are generally better but not necessarily on topic; Live results are usually on topic but have more spam and useless results; and Yahoo! is somewhere in the middle.

    I think all of them are rather less then titillating.

  • Mike

    Funny you mention this, some friends of mine were talking about this same thing the other day. It’s like the froze their search results from back in 95 and stopped improving. Yahoo is pretty good nowadays, I installed their toolbar instead.

  • Here’s an example of an out of date search query:

    “harry potter launch date” — way old stuff
    “harry potter launch july” — 3rd one is right, rest is about last year

    If we think of search as trying to parse out my intentions, shouldn’t it be obvious that my intentions are to know *this* year’s launch date?

  • Steve

    I just tried a Google search on “1457 Third Street Promenade” (quotes included with the search string), which yielded 37 hits. The top hit was a directory listing for Broadway Deli in Santa Monica, including phone# and a link to their Web Site. The website has a “directions” link which features written directions and a link to Mapquest. The Mapquest map does not come up for me, but I think that is probably an issue with my browser and/or OS.

  • Steve

    danah wrote:

    “If we think of search as trying to parse out my intentions, shouldn’t it be obvious that my intentions are to know *this* year’s launch date?”

    But I don’t think of search that way at all. I doubt that it would be possible to accurateely parse the intentions of a generalized search user. The audience is too diverse for that.

    I personally am offended when software tries to guess my intentions. Especially since it usually seems to guess wrong. Perhaps that’s because I’m usually a few sd’s off from the mean, but hey, that’s what diversity is about, no?

    I much prefer the model that says that search (or other software function) operates according to a rigorously defined but not overly complex set of rules. The responsibility of the developer is to choose the rule set wisely (as to usability) and impliment it as closely as possible to error-free (and to document it adequately). The responsibility of the user is to be aware of these rules and master them to the extent needed to acheive their goals.

    I am always amazed by the people who don’t know how to use +, – and ” in Google searches, but just type in keywords. I do not accept the strong AI hypothesis, and I think it is entirely inappropriate that people want some gadget to do their thinking for them.

    -Steve

  • Steve

    Using the following search string gives reasonably good results.

    +”Harry Potter” +launch +july +2007

    Out of the first 20 results, 11 give the correct date (July 21 2007) in the results summary. This includes the top 2 hits.

  • zephoria, thanks for the example. I passed your post to a couple different groups at Google with a “How can we do better on this?” remark. It’s a pet peeve of mine to search for a conference and get a conference web site from (say) the year before, so I sympathize.

    Something like Harry Potter is going to be a little harder, since the launches happen at irregular times. But we have been paying more attention to how to combine news and regular web results, e.g. [harry potter launch date] returns a news article in our OneBox at the top of search results. The article I saw was
    http://news.scotsman.com/scotland.cfm?id=713012007
    which mentions the July 21 launch date, but I agree that it would be much better to highlight that answer in snippets or return different results.

  • Bonsai Moss

    Same thing happens when looking for linux info, there’s nothing like borking your machine because you read through a howto post from 1999.

    Google is failing more than winning on relevance issues, to be safe, try your search on several engines, MSN seems to have the freshest document index at the moment.

  • Michael Chui

    1) For dates, add the year as a keyword.

    2) For spelling names, add “wikipedia” as a keyword. The majority of my one-shot searches these days (I’m an info professional, so I fine tune my searches a lot) attach that as a term, and it even gets useful non-Wikipedia results.

    Generally speaking, extra terms typically results in better results, as a rule of thumb.

    And, Matt Cutts, assuming you’re still reading; it’d be really easy to make it better. Just weight more recent results a touch more to offset the extra PageRank older pages have by virtue of being older and having had more time to get linked. Even better would be a way to identify pop culture-based queries against academic or historical-based queries, since in some cases old articles are better than newer ones. In the case of the former, the newer the better; in the latter, it’s neutral.

    But that’s just me. =P

  • Steve: isn’t + mostly a holdover from AltaVista or some other ancient site that used words as “suggestions” rather than absolute search terms? e.g. I get good results for “harry potter launch july 2007” (no quotes)

    From the horse’s mouth:

    Google ignores common words and characters such as where, the, how, and other digits and letters which slow down your search without improving the results. We’ll indicate if a word has been excluded by displaying details on the results page below the search box.

    http://www.google.com/intl/en/help/refinesearch.html

    (Of course it’s possible that you query these words much more often than average….)

  • thank you very much for your informaations..mr. suma…

  • I’ve definitely noticed in my recent Google images searches that my results are a lot less precise than they had been. Now it seems that every image on any page that has the keyword on it shows up. Before it seemed like it only gave results based on file name, alt tags, captions, etc.

  • Moochie Bennen

    try typing in “she invented” and see what it asks you what you meant.

  • Steve

    Scott,

    The + operator was definitely present in Altavista, although I don’t know if they were the originators. For those who might not be familiar, the meaning of the “big three” operators is as folows:

    Quotes: Enclosing a portion of the search string in quotes (“”) mandates that the target document must contain that exact phrase

    Plus: The + operator indicates that the prefaced word or phrase *must* appear in the document. Example:

    +Harry +Potter

    implies that both the term Harry and the term Potter must appear in a returned document. This is the Boolean “AND” condition.

    Search default is the “OR” condition where all documents containing either “Harry” or “Potter “might be returned. Of course, Google then uses its own mysterious “relevance” algorithms which in practice bring the results containing both “Harry” and “Potter” to the top of the list. Probably with a preference for the exact phrase “Harry Potter”. However, I prefer not to rely on a relevance algorithim – especially one which is not open source.

    Minus: The – operator is an exclusion. The returned document must *not* contain the prefixed word or phrase. I fund this useful in refining results after I get a large number of unwanted hits with similar content in common.

    People who just type in keywords without using operators are putting their trust in an unknown relevance algorithim which is constantly being tweaked and retweaked and is driven in large part by commercial concerns – i.e. People with commercial sites try to “optimize” them for higher results ranking and Google constantly tries to tweak their algorithim to outwit Search Engine Optimization strategies that it sees as illegitimate.

    I know those guys have billions of dollars, and I don’t, but I’d rather see plain simple usable search which does not treat the user as a pawn in a commercial chessgame.

    -Steve

  • See your prvious post for the source of the spelling mistakes… As more people come online and make up their own spelling, you can see how they will reweight it. Given 20 years or more of not teaching converged spelling, but encouraging ‘creativity’ instead of explaining the phoneme to glyph mapping that english uses, that is what you end up with.

  • Another failed query: “Southwest airlines san francisco to los angeles” or “Southwest airlines SFO to LAX” — i’m trying to find the various reports on Southwest opening up at SFO again to see if they’ll fly to LAX. Everything is way old.

  • fp

    I’m amused, or as you would say “humored,” by your failure to release from moderation my comment regarding my search for a trotting horse named “Jake the Snake.” The search, the pop cultural facet, the lake of data on harness racing in general — all seemed relevant to your ponderings. One wonders what the subtext might be for screening out a comment like that one.

    Or perhaps I neglected to enter the word in the spam checker and I’m simply paranoid…

  • fp – probably the latter since i have no record of your other comment.

  • Steve

    +”southwest airlines” +sfo

    This search took perhaps 5 minutes to find an appropriate page in the results and navigate to the following paragraph from the Southwest Airlines blog. It was the 4th result, and I had to guess that it might contain info (by implication negative, as it happens) about service to LAX. The closest appears to be San Diego.

    An ABC report on Southwest’s return to SFO was result 2, but didn’t include route info.

    Note that my search was improved by *not* including LAX, since, as it happens, service to LAX is not part of the story – hence not mentioned on the target pages.

    ***************************************************************
    Just as we promised, we�re bringing our LUV Jets back to San Francisco International Airport (SFO)! Effective on Sunday, August 26th, Southwest begins nonstop service between SFO and San Diego, Las Vegas, and Chicago Midway, with direct or connecting service to 46 additional airports from sea to shining sea. We�ll offer eight flights each weekday between SFO and San Diego; seven each weekday between SFO and Las Vegas; and three each day between SFO and Chicago Midway. Southwest�s SFO service will operate from Terminal 1, Concourse B, gates 25 and 31. (Click here for more details.)
    ****************************************************************

  • PhilBradley

    What’s really odd is to continue to use a poor search engine. Exalead for example is far superior to Google when it comes to spelling.

  • some guy

    “If we think of search as trying to parse out my intentions”

    This seems like a disastrous approach. How is Google supposed to divine intentions when they pertain to rapidly developing ephemeral events like the launches of shitty books?

    They have a limited amount of stuff to engage in term/phrase sense disambiguation wherein they suggest frequently occurring senses of it, but that is by no means a central focus of searching and indexing. Taking that to its logical conclusion is a recipe for brittleness and wasted money. Wasting the talents of thousands of brilliant PhDs on wild goose chases like guessing what every little cryptic two/three-word query means is a terrible idea.

  • mika

    Give Ask.com a try, perhaps? They found 1457 Third Street Promenade, Santa Monica, CA right off the bat. Think outside the Google/Yahoo box..

    :0)