Googles social interface
July 23, 2008 6:31 AM   Subscribe

Google is testing a Digg-like social interface to Google Search results, Techcrunch has an early preview video. This is bad news for Jimy Wales's Wikia since this is what they have been trying to build. Perhaps related it looks like Google is buying Digg.
posted by stbalbach (59 comments total) 2 users marked this as a favorite
 
Doesn't Google's algorithm already take popularity (via links and probably also clicks in the search results) into account?
posted by DU at 6:42 AM on July 23, 2008


Too easy to game. Maybe they can't lo$e though.
posted by Brian B. at 6:44 AM on July 23, 2008


This is going to add way too much noise to search results. Google is already my second (or third) choice when it comes to heavy duty searching because of their search filtering. This isn't going to help.
posted by elfgirl at 6:51 AM on July 23, 2008 [1 favorite]


What do you mean by "search filtering"? And what is your first (or second) choice?
posted by DU at 6:53 AM on July 23, 2008


Does every goddamn thing on the internet have to be a social network?
posted by Legomancer at 7:02 AM on July 23, 2008 [9 favorites]


Does every goddamn thing on the internet have to be a social network?

Social networks are marketing aggregators, stuffing a large number of eyeballs into a central location. So, if the intent is to add revenue streams, then, yeah, building a social network isn't a bad way to go.
posted by Thorzdad at 7:09 AM on July 23, 2008


Legomancer, I'm an admin for a group called GetOffMyLawnYouKids, and we'd love to have your comment added to the group!
posted by yhbc at 7:10 AM on July 23, 2008 [20 favorites]


DU: If google thinks I have misspelled a word, I will frequently get a pile of results with the word that it thinks I wanted, without asking. I can't figure out how to ignore that stuff. e.g. if I search for "gruit", I get many results that match "fruit". I don't want to disallow pages with the word "fruit", but I don't want them to be conflated.
posted by mkb at 7:21 AM on July 23, 2008


I'm pretty sure the only thing this will do is increase Google's already-massive collection of spam.
posted by Plutor at 7:21 AM on July 23, 2008


On a related note techcrunch is almost done including every possible external javascript in the world on their webpages.
posted by srboisvert at 7:23 AM on July 23, 2008 [3 favorites]


If google thinks I have misspelled a word, I will frequently get a pile of results with the word that it thinks I wanted, without asking. I can't figure out how to ignore that stuff.

I've figured out a Life Hack to get around this problem. Here's what you do:

1) Look at the top search results only.

It's pretty counter-intuitive, but it turns out that the stuff that matches best ends up on top. So "misspelled" matches come up first, then the stuff it thinks I mean comes up later.

Your "gruit" example doesn't work, because it doesn't detect that as a misspelling. But how about a random search like this one.
posted by DU at 7:32 AM on July 23, 2008


That's not a useful hack; top search results aren't often what I'm actually looking for.
posted by mkb at 7:35 AM on July 23, 2008


"stuffing a large number of eyeballs into a central location."

Doesn't this sort of thing usually come with a death sentence? Or at least a life sentence with eventual shanking, a la Jeffrey Dahmer?
posted by Eideteker at 7:38 AM on July 23, 2008


Not to downplay the skill involved. Stuffing eyeballs anywhere is difficult due to their proclivity to pop under pressure.
posted by Eideteker at 7:40 AM on July 23, 2008


Do you mean you are specifically looking for something not in the top? Or that the thing you want isn't in the top?

If the former, then no amount of community editing is going to help you (although personal editing would). If the latter, then it sounds like their algorithm, which AFAIK already includes "social" information in the form of linking, just needs some tweaking.

I don't want to sound like a Google fanboy, I'm just trying to figure out what problem this feature is solving.
posted by DU at 7:40 AM on July 23, 2008



I don't want to sound like a Google fanboy, I'm just trying to figure out what problem this feature is solving.


Exactly. If it ain't broke, and really, it isn't, don't fix it. One of google's initial strengths was it's incredible simplicity. For the most part, they've held on to it. Don't blow it now.
posted by phrontist at 7:52 AM on July 23, 2008


Well, IMO their search results have really gone downhill in the past few years, especially when it comes to searching for programming information. A lot of times you get these obnoxious websites that are obviously just collections of scraped Usenet postings filled with ads.

On the other hand, Wikipedia has in a lot of ways made Google superfluous, I mean if I'm just looking for 'information' about something then chances are it's in Wikipedia (which should hopefully have lots of source links if I need more detail). Clearly Wikipedia has issues but for the most part it has most of what I'm curious about. But ironically, Wikipedia's own search engine sucks, so Google works well for finding stuff on Wikipedia.
posted by delmoi at 7:56 AM on July 23, 2008 [3 favorites]


If google thinks I have misspelled a word, I will frequently get a pile of results with the word that it thinks I wanted, without asking. I can't figure out how to ignore that stuff. e.g. if I search for "gruit", I get many results that match "fruit". I don't want to disallow pages with the word "fruit", but I don't want them to be conflated.

I find that quoting a single word does wonders for forcing google to returns results that exactly match it, sans the usual stemming/typo substitutions it'll do on an unquoted token.

So search for, not gruit, but "gruit", for example. I wish I had a good proof of concept search for you, but I'm having a hard time recalling any of the specific searches where this has been helpful.
posted by cortex at 7:58 AM on July 23, 2008 [1 favorite]


The Google search results page has looked like this for most of a month, under my account. I didn't realize this was news. I haven't made extensive use of the features other than to remove useless results; you could do that before in any case.

I've yet to see a single edit or comment by any other user on the collaborative interface. I've been assuming everyone else found the feature as stupid and pointless as I did, but I guess it turns out that most people aren't even seeing the feature set. Personally, I'd rather whoever spent their time coming up with this thing book a few hours on culling spam from the index instead.

Anyhow, yes, I've used this stuff. And yes, it's stupid and pointless. I've just been too lazy to cast around for some way to turn off the ugly rubbery buttons it adds to search results.
posted by majick at 7:58 AM on July 23, 2008


Social networks are marketing aggregators, stuffing a large number of eyeballs into a central location. So, if the intent is to add revenue streams, then, yeah, building a social network isn't a bad way to go.
posted by Thorzdad at 10:09 AM on July 23


Actually, it is. Social networks on search results is how lowest-common denominator tastes and browsing havits will become the focus of internet companies. Everyone thinks the long-tail is great when their tastes fall on the long-end of the tail. But for companies, who need wider margins, the money has always, and will always be in the very narrow but tall end of the tail. Ebay, google, and amazon can monetize the long end, certainly, but they'd rather monetize the short end more effectively. For Ebay and Google, whose revenues are based on competitive bidding for resources (ebay gets the fee, google gets the winning bid), there's more money in more people bidding for a smaller quantity of things.

Remember, everything has a marginal cost, even if that cost is relatively low, like disk drive space or bandwidth given the revenues of some of these companies, it is never zero. Because of that, there is always higher total profit in pushing more people towards fewer things.

Furthermore, these net companies have already monetized the long tail. There isn't any more growth there, or at least not the 40% annual growth that these companies need to keep their share prices up. The growth is going to have to come from the short end of the tail that these companies didn't focus on for the first decade of their existences.
posted by Pastabagel at 8:06 AM on July 23, 2008 [3 favorites]


I'm just trying to figure out what problem this feature is solving

My guess that the audience who might provide the contextual information via links on other sites is often different than that who would enter comments about a link directly. The people providing the information via links are often going to be more technically literate and better informed than those who type something in directly - but they will take a little longer to pick up on new information than an audience who can directly comment on the links might do.

The problem is that a link from another site - like a Digg - is something that can be simply counted as an indication of quality. A text comment may be either positive or negative.
posted by rongorongo at 8:09 AM on July 23, 2008 [1 favorite]


"This is bad news for Jimy Wales's Wikia"

I think we all know the real bad news for Wikia is that it's run by Jimmy Wales and his cronies and sycophants.
posted by orthogonality at 8:09 AM on July 23, 2008 [7 favorites]


On the other hand, Wikipedia has in a lot of ways made Google superfluous

I agree completely and think that WP will be Google's biggest competitor down the line. We're back to the old people-created directory vs. automated crawler battle, and the rise in spam/malware/etc. has given a *huge* boost to the people-created directory. Yahoo shouldn't have given up.

I only use (Give Back My) Google as an address-bar search shortcut, and generally when I'm looking for the "I'm Feeling Lucky" first result for a URL I can't remember.

Oh, I also use and highly recommend the Customize Google extension for FF. It lets you customize the search results (e.g. streaming search results with no "Next" pages) in a variety of ways. I can only assume it will let me remove this Digg-crap feature if/when it launches.
posted by mrgrimm at 8:14 AM on July 23, 2008


I guess theoretically it does make sense. Link context is information based on the data producer. This voting feature is based on the data consumer. Basically, Google wants to ask the customer if they are doing a good job of providing results and adjust those results when it isn't the case.

We'll have to see if it ends up causing more problems than it solves, though. Gaming is a big problem, as is complexity (on more than one level).
posted by DU at 8:16 AM on July 23, 2008


Wasn't this widely discussed among Google watchers like ages ago, and even long before that? And doesn't Google's FAQ, while being relatively vague, indicate that this only affects your own searches? So what's the connection with the walruses over at Digg?

(and yeah, as for the "I got fruit in my gruit" thing, cannot you just explicitly exclude the search term you're not interested in? Like, say, "gruit -fruit" or "fiting -fighting" etc.)
posted by effbot at 8:26 AM on July 23, 2008 [1 favorite]


Exactly. If it ain't broke, and really, it isn't, don't fix it. One of google's initial strengths was it's incredible simplicity. For the most part, they've held on to it. Don't blow it now.

Apparent simplicity. Google is all kinds of complex under the covers, but it makes it look easy. That's why it's so good at what it does.
posted by blue_beetle at 8:27 AM on July 23, 2008


Has elfgirl been abducted by the black googlecopters or something? It looks like they silenced her to keep the first choice a secret.
posted by ghost of a past number at 8:28 AM on July 23, 2008 [1 favorite]


Google took a huge usability hit when they started indexing stuff behind paywalls and returning it in the search results. I don't know why they started doing that.

If this new feature lets people vote down that crap, I guess it'll be good, but it would be easier to just crawl pages twice — once as Googlebot and once as a regular browser from a non-Google IP — and throw out all the pages that don't display exactly the same content.

Since that's really my only major complaint with Google, and it could be trivially fixed without any human intervention, this 'voting' thing seems like it will probably do more harm than good.
posted by Kadin2048 at 8:32 AM on July 23, 2008 [1 favorite]


(and yeah, as for the "I got fruit in my gruit" thing, cannot you just explicitly exclude the search term you're not interested in? Like, say, "gruit -fruit" or "fiting -fighting" etc.)

You can, but the problem is that this will prevent you from seeing any pages that contain both gruit and fruit, which is annoying when you want to see all pages that contain "gruit", whether or not they contain "fruit" as well.

So there's these three scenarios with the googling, and I'll use an actual working example that I remember now from yesterday, when I was checking out variants for "microwave":

1. Show me pages that contain something like 'microve':

Search string: microve
Results: ~45,000 hits, including many that contain 'microwave' instead.

2. Show me only pages that contain 'microve' and not 'microwave':

Search string microve -microwave
Results: ~5,000 hits, for pages that contain 'microve' but do not ever contain 'microwave'

3. Show me all pages that definitely contain 'microve':

Search string: "microve"
Results: ~6,300 hits, for pages that in some cases also contain 'microwave'.

Method 1 is how most people search. Method 2 is seemingly common knowledge. Method 3 is not, I don't think, such common knowledge, and while in a way it's logically obvious (nobody said a quoted string has to have more than one token, right?), it sure took me forever to pick up on it myself. But it's handy.
posted by cortex at 8:41 AM on July 23, 2008 [2 favorites]


Google is all kinds of complex under the covers, but it makes it look easy.

For the curious: Behind the Scenes of a Google Query. "Today it’s estimated that Google queries travel across 700-1000* machines, a figure that has nearly doubled since 2006 perhaps due in part to the introduction of Google Universal."

I don't know why they started doing that.

Google does that to "provide a promotion and discovery opportunity for webmasters of sites with restricted content". Outright cloaking may get you banned.
posted by effbot at 8:47 AM on July 23, 2008 [1 favorite]


Method 1 is how most people search. Method 2 is seemingly common knowledge. Method 3 is not, I don't think, such common knowledge,

I would switch 2 and 3. I know about both, but I never think to use the operators. Quoting, however, is pretty common even among the tech-semi-literate.
posted by DU at 8:54 AM on July 23, 2008


orthogonality: "I think we all know the real bad news for Wikia is that it's run by Jimmy Wales and his cronies and sycophants."

Yeah Wales bombed with that Wikipedia thing. Wales is really good at building community. Which is one of the criticisms of Wikia, it's not really a community, it's not what Wales is good at.
posted by stbalbach at 9:04 AM on July 23, 2008


Not to bite the hand that feeds me, but google does seem to give way too much weight to blogs. I mean, for a handful of various search terms, one of my blogs consistently ranks in the top 10 results, but one of my client's sites can't manage to rank in the top 30 for stuff they're actual experts and paid professionals in doing.
posted by Dave Faris at 9:04 AM on July 23, 2008


Agreeing with DU. Unless you have a separate programming background, constructing a "query" with pluses and minuses is not something you think of doing. Putting an exact term in quotes, though, is.
posted by yhbc at 9:06 AM on July 23, 2008


Quoting, however, is pretty common even among the tech-semi-literate.

Oh, I might have been unclear: I think quoting multi-token strings for full-string matches is pretty well known, yeah, probably more broadly than any of the operators. It's specifically the quoting of a single token to prevent the injection of stemming/typo results that I'm characterizing as the Weird Method 3.
posted by cortex at 9:09 AM on July 23, 2008


effbot: While I still think that sucks, that page makes it clear that you're supposed to get the "First Click Free" — that is, you're supposed to be able to click on the link in the Google results and see the information that caused the page to pop up in the results in the first place. Additional clicks from that page can be restricted, but the page actually linked from Google's Search Results isn't supposed to be.

A lot of sites don't even let you do that. There are a ton of "Experts Exchange"-type sites, and academic journal sites, that come up in Google's Results but then won't let you see anything when you click through to the link. They're nothing but results pollution … yet as far as I can tell, Google doesn't do a thing about them.

I'd have a lot less of a problem with Google's policy if they rigorously enforced "First Click Free", but they don't seem to be doing it.
posted by Kadin2048 at 9:11 AM on July 23, 2008 [1 favorite]


Cortex' point is that the form "word" disables the automatic spelling correction (or alternate search, or whatever google might call it), which might be less obvious. I'm not sure about the right order for 2 and 3; I only know that I knew about them both, but completely forgot about 3 when writing my reply.

(I also know that I shouldn't let a comment sit in the textarea for half an hour, and then post it without updating the preview and checking if cortex has already made portions of my comment redundant, but I completely forgot about that too. Mea culpa.)
posted by effbot at 9:12 AM on July 23, 2008


Hmm...used to be you could see the answers at Experts Exchange by using the Google Cache. This no longer seems to work. Drat.
posted by jedicus at 9:14 AM on July 23, 2008 [1 favorite]


(I'd include normal multi-token quoting in method 1, or maybe in special method 1b since "put quotes around it, dammit" seems not be something your average search newbie understands, and that's a whooooole lotta people.)
posted by cortex at 9:21 AM on July 23, 2008


stbalbach writes "Wales is really good at building community."

You got the first letter right, but you misspelled the rest of "cults".
posted by orthogonality at 9:23 AM on July 23, 2008


...the quoting of a single token to prevent the injection of stemming/typo results that I'm characterizing as the Weird Method 3.

I understood what you meant, but I still think I disagree. Quoting is widely understood even in regular non-technical writing to mean that something is literal. Compare:

She said she'd look into it.

She said she'd "look into it".

Since making your query literal is exactly what you want to do, I don't think it's that weird to have quoting a single word be how you do this. More to the point, I think a non-techie would think of/like that.
posted by DU at 9:29 AM on July 23, 2008


Hmm...used to be you could see the answers at Experts Exchange by using the Google Cache. This no longer seems to work. Drat.

Scroll down to the bottom of the page, after their massive grey Experts Exchange Zones. All the solutions are posted below and require no logging in to view. As far as I know, this method works and still does.
posted by Meagan at 9:31 AM on July 23, 2008


There are a ton of "Experts Exchange"-type sites

At least one of the sexchange sites displays a "sign up now for a free trial, and we'll let you read the following grayed-out answers" after the question at the top of the page, followed by tons and tons of crap, and then, at the bottom of the page, displays the answers again. Just keep scrolling to see the page you searched for.

I sure wouldn't mind if they kicked those guys off the searchable portions of the net.

As for academic sites, that's might be related to Google Scholar, rather than the Free Click thing; I'm not that bothered by that personally, since it quite often helps me find the papers I'm looking for. Guess they could use the stuff mentioned in this FPP to filter that out automagically.

(they're easy to skip over manually, since they usually don't have a "cache" link)
posted by effbot at 9:32 AM on July 23, 2008 [1 favorite]


Agree to disagree, then, DU. Like I said above, I think that the multi-token -> single-token quotation analogy makes a kind of logical sense, so I think we agree on that front, but for all that I think it's still pretty opaque in the specific context of metafiltering (heh) google's search results and not something the vast majority of searchers would ever have come to intuitively.
posted by cortex at 9:35 AM on July 23, 2008


In place of quoting a single word, you can also use the + sign, e.g. search string +microve. The two methods give identical results, I think.
posted by sevenyearlurk at 9:40 AM on July 23, 2008


Has elfgirl been abducted by the black googlecopters or something? It looks like they silenced her to keep the first choice a secret.

HAHAHA. No! Of course not! Why would they ever do that? ::looks shifty::

What do you mean by "search filtering"? And what is your first (or second) choice?

(A caveat of sorts to begin with: I'm a business researcher. The types of online searches I do usually go beyond looking for the surface level information that most people look for when they use Google. Google is still my first stop when I'm looking for that restaurant around the corner.)

The way that Google weights their search results means than some of the results that are obscure, yet relevant, often fall below less relevant results that, for what ever reason, have higher weight to Google. I often get better results from Yahoo, despite the fact that they use the Google engine behind their search, to the point that it's becoming my first stop more and more often.

This site give a good visual demonstration of the differences between Google and Yahoo search results for the same term.
posted by elfgirl at 9:50 AM on July 23, 2008 [3 favorites]


In place of quoting a single word, you can also use the + sign, e.g. search string +microve. The two methods give identical results, I think.

That makes a crazy kind of sense. Awesome. Thanks, sevenyearlurk.
posted by cortex at 9:55 AM on July 23, 2008


elfgirl, that's a really cool tool.
posted by cortex at 10:01 AM on July 23, 2008 [1 favorite]


That's an awesome site, elfgirl. And I know exactly the problem it portrays, though I usually don't have that problem with a site as big as MetaFilter. Do different users get different results? Is Google the next Amazon (i.e. useless search function)?
posted by DU at 10:02 AM on July 23, 2008


It seems like there's an issue with how it wrangles common results between the two sites: when I search for e.g. my own full name it returns an entry from my blog as the second hit from Google and the first from Yahoo, which is fine and all but it doesn't display them as connected common entry. The blog entry has a longish url (domain, date, title text urlified), which is listed on the Google side completely in a mouseover but as slightly truncated on the Yahoo side.

So it looks like results with sufficiently long urls won't get proper match-up treatment, exaggerating the apparent disagreement betweent the two engines. How much of an issue, I don't know; it might be just a minor blip.
posted by cortex at 10:14 AM on July 23, 2008


Does every goddamn thing on the internet have to be a social network?

The distinguishing factor between advertising on a standard-issue website and a social networking website is that the social networking site offers precious demographic information to its advertisers (age, gender, location, etc.). Most, as per the EULA, won't sell your personal info, but I think this sort of network-wide data aggregation is pretty standard.
posted by invitapriore at 10:40 AM on July 23, 2008


I went to the extent of setting up a /etc/hosts entry and personal web server just to batter experts-exchange off the fucking face of my web. What a lot of shit. When Google let me exclude domains from results permanently I will ... send them a little kiss.
posted by bonaldi at 12:22 PM on July 23, 2008


Personally I'm just astonished by the idea that anyone over the age of 16 would possibly see any value in Digg.
posted by influx at 12:24 PM on July 23, 2008 [2 favorites]


I went to the extent of setting up a /etc/hosts entry and personal web server just to batter experts-exchange off the fucking face of my web. What a lot of shit. When Google let me exclude domains from results permanently I will ... send them a little kiss.

The customizegoogle firefox add-on will let you do that. Don't know if there's a working version of it for ff3 yet.
posted by juv3nal at 1:09 PM on July 23, 2008 [1 favorite]


Awesome, juv3nal ... now, where is a list of spammy sites to filter out?
posted by Dave Faris at 1:30 PM on July 23, 2008


Yeah, I would not use google without customizegoogle, mainly for the experts exchange reason.
posted by maxwelton at 2:12 PM on July 23, 2008


orthogonality: "stbalbach writes "Wales is really good at building community."

You got the first letter right, but you misspelled the rest of "cults".
"

Some users go through this multi-step process not unlike a love relationship: initial curiosity, full-blown love, honeymoon, long grind, burn out and betryal and finally hatred, then eventually ambiguous.
posted by stbalbach at 8:09 PM on July 23, 2008


err.. ambiguous apathy
posted by stbalbach at 8:11 PM on July 23, 2008


Wales is really good at building community

Also peat and singing Tom Jones songs.
posted by kirkaracha at 8:47 PM on July 23, 2008 [1 favorite]


« Older Urinal Bitchie - Hello   |   Dead Funnies Newer »


This thread has been archived and is closed to new comments