Has anyone else noticed Google making a lot of autocorrection mistakes recently? I don't mind the traditional "did you mean" links, but I've been getting redirected to pages for different queries, with a "did you mean [what you actually typed]" link.
It's especially a problem with programming because a lot of words that are "wrong" are actually acronyms or strange library names.
I think that google is losing sight of what made them popular to begin with.
[within reason] Every workshop on the planet has a hammer in it. This is because hammers are very simple, they never break, they never behave unexpectedly, they're highly useful, and they do exactly what they're supposed to do as perfectly as anybody can imagine. Hammers are the perfect tools; something that should have been idolized in Brave New World.
Google was a hammer for a long time. It wasn't even a question that google would be the default search engine for everybody who knew anything about anything. Google did search really well, behaved exactly as advertised, rarely broke.
It was perfect. It was a hammer.
It feels like google (overall, I don't just mean search) is losing their grasp on this.
Stuff like charts API. Maybe it's just me, but the new version of this is annoying. It's some javascripty nonsense that feels complicated. It's a laser-guided, nuclear-powered gryoscopically stabilized, diamond-tipped screwdriver that can also make coffee and fold your laundry. You could build a submarine with it.
But it's also got an instruction manual that is a foot thick.
And now I can't +term, they took away the links to "cached" versions of the pages, things autocomplete (incorrectly), it assumes I meant something that I didn't, it wants to integrate into a social network, etc. etc.
Google...you're starting to feel a little microsofty. Are you okay?
The cached links are still there. They just got moved to the preview page. I assume not enough people used them for the amount of clutter they created in the SERPs.
thanks for mentioning the cache links, I thought they were gone and used the "Cache:URL" query. I don't like the preview pane, never paid any attention to it (actually, I think it's pretty stupid :-)...
I understand how Google indexes the web. Let me control my own queries. Give us geeks the ability to enable/disable freshness, even if it means appending &fresh=0 to the URL
It is a bit like small dogs shouting at the caravan, but if I had to express a concern about Google, it would be more about their new design, the replacement of default scrollbar, the autoloading of next images, etc.
There's a fine line between being helpful and taking control from the user. Google used to be safely on the helpful side of the line. Now more and more SERPs are wasting my time by making me refine my search with qualifiers to ensure I actually get what I want.
Google recently corrected "ruby dreampie" (I wanted a Ruby equivalent of the excellent Python REPL) to "ruby creampie", with extremely NSFW results. It should avoid correcting a search term if the corrected version is potentially offensive.
Most punctuation isn't indexed. A few terms like C++ and C# are special-cased, and a hyphen usually works, but mostly you just can't search for anything punctuation-related for programming. I was recently trying to learn the difference between the <%= and <%# tags in ASP.NET and Google couldn't even understand the query.
Not that it's an easy problem. How are the spiders and indexers supposed to distinguish "true?" as a programming lexeme versus just an interrogative sentence that ends with "true"?
Sometimes we forget that Google and the other search engines aren't a byte-for-byte search across the entire web. The pages are all tokenized and indexed, with most punctuation dropped since that is indeed the more common usage case. You don't want a search for a word to miss out on pages where the word was glued to quotes or a comma or something.
Ah, this is one of those situations where you'd love to remember an example or more, but everything you try actually works - then sometime a few days later it happens!
I came across this one yesterday when I was looking for more information about something mentioned on HN:
Search: carmack +zfail gpl
Date: 2011-11-02
Result: Bad autocorrect; Plus operator broke
"""
Showing results for carmack fail gpl
Search instead for carmack zfail gpl
"""
Because the plus operator is broken my final search ended up being `carmack "z-fail" OR "zfail" gpl` because different sources included or excluded the dash. Note that I think it's good that quotes are still exact and they shouldn't do the synonym search, but this is something the plus operator would have worked on, if I remember correctly.
every page in the top 10 organic results were coming from pages that were either added or updated within 2011 except for the youtube results. those 2 videos were added in 2009.
i also noticed that your youtube videoes were higher in the results than articles from gizmodo, arstechnica, engadget, Time magazine and whole slew of other articles that were released 2 days ago.
I tried this, and began to doubt my stat -- I saw no instances of the problem. But then I tried to repro the problem and I see that history doesn't record it.
E.g. search for [barack obam] (typo intentional). Then when it autocorrects, click the "I meant what I typed" link. Now look at your history: it doesn't mark that in any way.
I can't imagine Google devs are happy with this. Cmon Google, let us geeks in on the internal search that your devs use. Host it at geek.google.com or whatever. Probably won't happen since geeks don't click on ads much, and Google is cutting down on dev friendly side projects like code.google.com these days.
Yeah, not even the quotes help in some cases, and I expect Google to show me results that contain the exact word I'm searching for, no matter if it thinks I misspelled it (it's like the algorithm says "let me correct that for you, stupid" :-)...
What I dislike the most from recent updates is the preview feature. I find it completely gets in my way, and it's very easy to unintentionally make it preview something, when I just want to scan the results. Does anyone else find that feature useful?
I never understood the usefulness of the preview feature. The only time I find it marginally interesting is when a website is currently down or had recently been taken offline. Though, I find the "cache:" query more useful in those cases anyway.
The preview window does pop-up unexpectedly for me as well. I guess my mouse pointer wanders over to the right side of the search result every once in a while, triggering the preview.
I think the preview thumbnail is a very poor substitute for the 'cached' links that have disappeared. Sometimes the cache link was the best thing for an underperforming website or one that was down. It was simple, everyone understood it, now it's gone. Boo.
I don't at all, and find it incredibly distracting to have movement on the entire right half o the page when I'm just trying to scan the search result text.
The following CSS should disable preview and put the Cached and Similar links back into the correct spot. I use UserCSS in Opera, not sure what to use in other browsers but I know there are solutions for custom CSS available:
.vspib { display: none !important; } /* This is the preview specific line */
.vshid {
display: inline !IMPORTANT;
padding-left: .7em !important;
}
.vshid > a {
color: #1122cc !important;
text-decoration: none !important;
}
.vshid > a:hover { text-decoration: underline !important; }
.vshid > a:visited { color: #6611cc; }
This next bit of CSS should puts the URL back below the snippet text. It works for most result types but there are a few where it's buggy, for example any results that show how many people "+1'd" a URL:
Yeah, it's just... stupid. It would've actually been more useful if it was smaller (like Windows 7 taskbar previews), not take a whole third of the page. Thankfully, it's not automatically showing up anymore (at least for me), so it's cool...
This should be good. Many queries get clogged up with the long tail of pages from 2003 that are often useless to answer the question at hand.
I hope Google will let us turn it off easily ("-fresh" keyword? advanced search?) if we truly want older pages.
I also hope this won't lead to content farms auto-republishing their pages every six hours with the latest timestamp included to look "fresh". Or rather, I know it will lead to that, and I hope Google is prepared for it.
I also hope that the effect that "freshness" has on the total ranking of a result is not high. There are huge swaths of the internet and subjects in which "freshness" would not improve quality.
Changes of this blanket nature are a result of some form of the "Valley Effect" or new is always better. IMO.
The example used, the Olympics, is an outlyer comparison in which this specific change is overly relevant.
I hope this works. Searches for error messages and the like are often cluttered with results from 2009 or earlier, describing troubleshooting steps that are out of date or unnecessary. At this point I add the current year to the end of any given troubleshooting search – and even that isn't always enough.
I run into this problem all the time. I usually hit the "More search tools" link and then limit my search to the past year or so (depending on the issue).
Also including specific version numbers helps as well.. but maybe this update will make those techniques irrelevant.
Lately I've noticed that search results are much worse than they used to be. I used to be able to find the things I've been looking for on the first page, now it seems like most of the links are not very relevant to my query. Any one else experience a degradation in the quality of results?
Yes, they are worse. A lot of times, Bing actually shows something more relevant, so it can't be all the spammers. Maybe it has something to do with the search history Google keeps, because sometimes I get different results if I'm logged out of my Google account...
Yes, i noticed better results outside of personalized search... It is a hassle to logout thought. I actually use bing for 90% of searches... google is better at tech searches like mysql questions, but outside of that there results have gotten more irrelevant and crowded with useless content...
I hope they thought everything through, because there are a lot of cases where old content is just as good or even better - there are a lot of "timeless" articles and video out there.
However, when it comes to software/tech/anything fast moving, I'd really like to see the latest results first. I'm tired of searching for "how to set psa-proftpd to standalone in plesk" (just a quick example off the top of my head) and get 5 year old results that have long since stopped being relevant...
The real question: does it matter who "breaks" a story first? Will the folks who got the story first get to the top of the "fresh" pile even though their story may be a day older? Or does the most recent content always win?
I'm actually a little conflicted about this. For instance:
If I search for [olympics], I probably want information about next summer’s upcoming Olympics, not the 1900 Summer Olympics
I'm guessing that Google has the data to back that up, but it seems presumptuous. In any case, when you search for "olympics" it returns results about the Olympics as an institution, not any particular year. As it should be. In fact, the first specific year mentioned is 2016. Maybe Google turned the dial a little too far.
As for the "occupy oakland" and "nbc lockout" searches... isn't that's what News search is for?
> I'm guessing that Google has the data to back that up, but it seems presumptuous.
It's interesting that you find the idea of a freshness signal presumptuous, but the hundreds or thousands of existing signals seem to be completely ok. The whole point of search is to try to guess from hilariously little signal (a couple of words) exactly which parts of a huge corpus (tens of billions of pages) you want to see. Making arbitrary decisions on what you meant is the main functionality of the site, not some annoying extra feature.
They were also experimenting with automatic scrolling in Image Search - that left me dumbfounded when I was trying to click on an image and it was scrolling down instead :-). It seems to be gone now, though...
color me unimpressed that they're now able to sort their livecrawl with a 'date' flag. And this seems to simply be a case of firing that filter automatically for 35% of their searches.
It's especially a problem with programming because a lot of words that are "wrong" are actually acronyms or strange library names.