I’ve been trying to write a post about this, the discovery of images of women who discuss politics, photoshopped into bikinis, above their Google search results, but I can’t. There’s something that pisses me off about Google showing off all these sexually charged pictures of normally professional women (especially fake pictures) that I can’t articulate in a calm, rational fashion.
I feel like Google needs to dump the entire practice of showing image thumnails above search results until it can filter out any pictures that show women in a demeaning light, but it won’t and it can’t. The ranking algorithm is based on link popularity, and these images tend to be popular with the degenerates that make up so much of the internet population. I don’t distinguish myself from those degenerates, but I’m sickened at the idea that if I have a daughter someday, a disgusting photo of her could appear on her search results.
I don’t like it, and I can’t calmly express why. I’m sure some of you get what I mean. If you have an opinion on this, please give it in the comments, because I’d like to see how others feel.
Danny Sullivan points out that both Google and Windows Live Search feature this photo of a raccoon attempting to subvert the laws of nature with a li’l dog:
So, what are search engines supposed to do? It’s a popular image for obvious reasons (it’s stupid), with probably a ton of places embedding it , so it’s only natural it would wind up making the image thumbnails that appear above search results, and there are a million examples of searches like this one. There really isn’t much to do since it is deserving based on the ranking algorithm, and it isn’t offensive, just silly. The SafeSearch filter would probably catch anything really bad, while the rest of us can enjoy our favorite woodland creature/house pet erotica.
I’m glad to see this change (which I hadn’t noticed before): Google Images will now show more than two images per page from a single site.
Before, Images would show two images from a site, then a link to show more from that site. This seriously decreased the value of Google Images, because in some cases a single site will have 100 relevant images, while the entire rest of the internet will have many irrelevant images. Now, if one site has three or more relevant images, Google will just show you the best ones, no matter where the come from.
Nowhere is this more evident than a search for my name (how do you think I found it!). As you can see in this screenshot, 14 of the 20 results on page 1 come from InsideGoogle, as indicated by the arrows:
This is a great change, one that seriously upps the utility of Images. It should be the default for Images, but available as an option on other Google engines, were sometimes it makes more sense to show more results from more relevant sites. I’d love to have a link to turn this on once in a while for Google web search.
Philipp reported last week that Google is working on image recognition features for Google Images, so that Images will be able to finally recongize what is in photos, not just the names and associated text/metadata, and that two of those features may already be live. You won’t find them anywhere in the Google Images interface, but simply adding a string to the URL will get you access.
If you want to see only images containing the faces of people, add
&imgtype=face to the end of the URL.
If you want to see only images from news articles, add
&imgtype=news to the end of the URL.
To try it out, Philipp coded a search box that lets you choose from the new parameters. Try it out at the end of his post.
There’s also a Greasemonkey script to add it permanently to your Google Images page.
Google researchers are working hard to complete an algorithm that can detect the gender of persons in images. Google says that the most popular images searched for in Google Images are those of people, primarily celebrities, and labeling them all by hand is far from possible. Google is using an Adaptive Boosting meta-algorithm to allow it to identify the sex of persons in a photo from a low resolution grayscale picture of the face.
Google says it has 80% accuracy with 10 pixel comparisons, 90% accuracy with less than 50 pixels, 93% on a 20×20 pixel image. As the article mentions, one possible application could be to scan for adult content in images, too. After all, if you can tell the difference between a man and a woman, you can probably identify a boob, too?
I am so sick of the news on this blog being, on average, a week old. Its my fault. I let these tabs build and build and build, and I don’t have time to write because I’m too busy amassing tabs, and when I finally do write something, it’s a week old. Dammit! I am so not doing this anymore. I hate missing news, but it is beyond stupid to have late and irellevant news because you don’t want to miss anything.
And because of that, here’s everything I’ve got, leading up to just about today:
Google announced the 2007 Summer of Code. Wordpress is part of it, among others.
As part of the 50th anniversary of Gumby, all 200 episodes of Gumby are now on YouTube, absolutely free. Oh boy!
Scott Clark has a Google Doodle for Gumby he thinks Google should have used.
Google acquired video game advertising company AdScape, which everyone knew was coming. They are competing with Microsoft’s acquisition, Massive, which is far more massive and successful. Google will likely use an automated system and have the same success they had with dMarc, which is to say, none at all.
Google also acquired Gapminder, a data visualization firm that makes Trendalyzer. Looks like they are buying new features for Analytics.
Gary has the search engine logos for St. Patrick’s Day, mostly just Google’s. Barry has Yahoo’s and Search Engine Roundtable’s.
Google has added a feature for the Personalized Homepage that lets you customize the top portion of it with some cool themes. The regular Google.com homepage remains the same, but the Personalized one can now have some cool stylings.
Valleywag has a screenshot gallery of the Google homepage over the years.
There’s an easter egg in there. In most of the themes, just visit the page at 3:14 am (get it? Pi time!) and you’ll see something funny happen. Screenshots at Google System.
Blogspot.com has more spam, by far, than any other domain on the internet. I’m shocked!
Google AdSense is doing Pay-Per-Action ads, that pay out when the user clicking the ad actually does something, like buy something or fill out a form. The ads come with a rotating product format, and even embeddable text links, so you can write about a product and link to it as an ad, just like an Amazon affiliate link.
Arrington’s right when he says Google has crossed a line here. We’ll have to see if they’ve crossed the wrong line. Hopefully, unlike the Google referral ads, Google will never make this available to all AdSense publishers, instead holding it for trustworthy publishers.
Some bloggers just plain don’t like it.
The internet is so slow, Google is transferring data by FedExing hard drives!
Philipp has done this page that puts search queries from AOL’s privacy leak of last year with random images from Google Images, resulting in fun and poignant statement. My favorite is when the dog says, “I’m searching for ‘cute glitter myspace’”
The judge has thrown out the Kinderstart lawsuit against Google, saying Google is not liable for PageRank drops. Kinderstart lost so badly, they actually have to pay Google’s legal fees!
Google is classifying some “second class” employees as hourly workers, with compulsory unpaid lunch breaks and other breaks, limits on overtime, and the “threat of a black mark on the review of anyone who fails to punch in properly to the time-tracking window on their desktops.” Yoiks.
Yahoo has released a new version of Yahoo Widgets, the former Konfabulator. New features include a Widget Dock, auto-updating widgets, hidden widgets, 40% improved performance/memory usage, a FLickr widget, and lots of stuff for widget developers.
There’s a new look being tested on AdSense ads. Unlike some of the previous tests, these are pretty cool.
Oh, and holy crap! Lessig responded to an article of mine! I feel honored, and even more so since every point he makes in response to me is dead wrong.
Download Squad posts about Google Image Ripper, which gives you a page of full-sized images grabbed from Google Images results. Just enter a search term and image size (for some reason you can’t just view all sizes), and Image Ripper takes all the search results from Google Images, grabs the actual photo they point to, and displays them on a long single page (pictures along the right). Google Images’ thumbnails often do a very poor job of helping you compare images, so this inelegant solution gives you everything you need to know, just the facts.
By the way, is anyone else eagerly waiting for Zooomr to come back online? They’ve been offline during a major upgrade for three days now, and I just want to get back on the service. Apparently, the reason they’ve been offline is a major new feature: They’re analyzing the colors in every single image, and you’ll be able to search for images by color. Very cool feature.
They’re also adding over 250 other new features, including the ability to sell your photos as stock photos and keep 90% of the sale, unlimited storage, archives and photo sizes, an open API, and other stuff. Frankly, I’ve been souring on Flickr lately, and hoping maybe Zooomr could be more of what I’m looking for. I wasn’t too crazy about it in the one week I’ve used it, but this upgrade gives them a chance to win me over. With the ability to sell photos, and if their pages load faster (and have easy access to embed codes), I might just say goodbye to Flickr.
Someone at Google must be a little sad right about now. This guy decided to take the initiative and update the interface for Google Images a little bit, adding some stuff while retaining the same general look, and he got a very public smackdown from the community (I’m looking at you, mirror). So, after a very negative reception from users, Google has undone all that coder’s hard work, and we are left right back where we started, with the original Google Images interface.
Which is all just a roundabout way of saying that Google has brought back Google Images to the way it used to be, complete with text expanding on all the images. The redesign had removed all the text, save a snippet supposed to inform about the picture, with the rest of the info showing up if you rolled your mouse over, an inefficient method to say the least. In restoring the old version, Google has made one change: They now show only the website the image appears on, not the entire URL of the image.
I also discovered something strange and interesting about Google Images: If you resize your browser, Google reloads (AJAX-style) all the images on the page, reordering them to fit your new browser window. Google aqpparently wants the images to make a perfect grid, even to the point of showing less images per page on larger browser windows!
Take a look:
18 images, three rows of six, running full screen:
20 images, four rows of five:
20 images, five rows of four:
18 images, six rows of three:
21 images, three rows of seven (by setting browser zoom to 85%):
So, Google Images always shows between 18 and 21 images, based on the width of your browser. It doesn’t seem to care about the height. When you resize the page, all the images reload, because it isn’t the browser handling the resizing, it is Google Images. The page doesn’t reload, just the images, some good (and extremely subtle) use of AJAX.
Oddly enough, if you run Google Images large enough to get six images across, you are actually seeing less images on the page than you would with a smaller browser! It might be a good idea, if it is available to you, to run Images in a zoomed mode or smaller browser, just to get more per page. Interesting stuff.
By the way, if you are the Googler who did the redesign, don’t feel bad. Your heart was in the right place, your code pure. Next time, a shower of roses shall await you at your cubicle.
Google has done a terribly unpopular thing, redesigning Google Images so that it presents no new information, no new features, just increases the white space. It could be one of the worst redesigns I’ve ever seen.
The new Google Images is identical to the old one, except all the text that used to appear on the page, save the title of the image, is hidden, and only appears when your mouse is near the image. The redesign does not make for more space on the page, or larger images, or a cooler new look, it just removes info that was always there. If you want to find out what website an image was from, or how big it is, you have to work for it, and there is no way to compare the images against each other.
There’s no real explanation why Google did this. We know Google is obsessed with white space, but this is overkill, actually removing features by painting over them white. This would be the equivelant of Gmail only showing you who sent you an email, and forcing you to mouse over to see what the email was about, or Google Video showing you a thumbnail, with no video title, description, or length information. If Google wants to convince users to stop using Google services, they can start by doing this to all of them.
It certainly seems like Google is ripping off Windows Live Images, and if they are, this is a real crappy way of doing it. Windows Live Image Search shows only thumbnails on an all-white page, but they also let you resize them to show more on the page, the page scrolls into infinity (instead of showing a mere 20, then forcing you to click Next, over and over), and you get a scratchpad which lets you compare images regardless of where they show up.
So, Google has ripped off Microsoft, but did it by removing features it already had and lacking features Microsoft has had for months. Way to show leadership in the industry.
The part that pisses me off is that Google Images (and most of the competition) handles searching for images of similar sizes in a pretty broken way. All of them let you choose between all, large, medium and small (Windows Live adds images equal to your current screen resolution, Yahoo adds “wallpaper”, and Ask adds a buddy icon size), vague ways of distinguishing between them. I have no idea what a “medium” image is, and neither does Google as far as I can tell.
When I’m looking for an interesting image that fits a specific need, Google is always giving me images that are way too small, even when searching under medium (and large is way too big and limited, often). Now, Google doesn’t even show me the image sizes, so I have to guess? Screw that!
In the meantime, I’m going to be paying close attention to Ask X, and see how their image search measures up. I’m getting the feeling Google is determined to fall behind and make mistakes with their search interfaces, and I’d better keep my options open if I need to switch one day.
Google has launched a little game called the Google Image Labeler. In it, you help Google tag all the images in its Image Search database by pairing with a random person. You are both shown a random image and asked to submit labels, or tags, for the image. Once you both use the same tag, it is accepted and you move on to the next image. You can also choose to pass on an image if it isn’t working out.
The game is an interesting idea, to say the least. Google has a huge image index, and images are notoriously hard on machines, so any means of helping catalog them is a good thing. BuyGoogle has some good perspective on the reasoning and the man behind the game. I like the idea of using any sort of fun, even if it is bored fun, to help make search better.
That said, the game suffers from a major flaw: It uses thumbnails, instead of full-size images, during the game. That means that detail is nearly impossible to pull out. Half the images are just too tiny and meaningless to understand, and that results in a hell of a lot of passed images. I’m not going to squint at a tiny thumbnail and try to figure out if the car is a Ford or a Chevy, and being able to tell the details in an image is what would make a search engine great. I realize Google has had many legal issues surrounding Image Search, but they need to find a way to include full-size images if they want to get better tags than “car” for every damn car in the database.
There’s no long tail for this game.
Check out Philipp and John Battelle’s commentary.
Google has acquired Neven Vision, a company that does photo recognition. While Neven Vision’s background is in recognizing faces for biometrics, Google plans to use it to improve Picasa. Imagine if the first version of Picasa with this technology can recognize familiar objects in your photos, and let you tag people so they will be recognized in all your future photos. Could be interesting if the tech could be used in Google Images, as well.
Liz Gannes says:
Neven Vision has offices in Japan, the United Kingdom, Singapore and Germany, as well as 19 families of patents, so this can’t have been chump change. Investors include Anthem Venture Partners and Zone Ventures.
Check out all the other coverage on Techmeme.
Google Earth 4 Released
The fourth version of Google Earth is now out in beta for Windows, Mac and Linux. Super! The interface is even cleaner, with navigation controls integrated into the main windows, elminating the need for that navigation bar at the bottom of the screen. Also included are textured buildings (not just3D, but textured) in the software, via SketchUp.
Craigslist Poster Hot For Googler
Via Valleywag, we find a Craigslist poster who really wants to have a lunch date with a woman of Google. He says he works near Google and sees “lot’s of you hotties walking or riding those sketchy little scooters”. Methinks he’s a liar, and just wants to get into the Google cafeteria, possibly to get a better job. Smart dude, if that is the case.
Ask Banning Some Searches
Definitely don’t like this: Ask.com had been reportedly outright banning searches on some terms, like “sexy girls“, “pedophilia” and “sex with kids“. Searches on those and other similar terms returned no results and a “This query does not comply with Ask.com Terms of Service”. Such a total filtering system is far from ideal for a search engine, and makes research with it difficult.
Thankfully, this was not a new filter, but an old one being slowly phased out. As Gary Price explains, the filter was a holdover from the Ask Jeeves days. Jeeves was considered far too similar to the P. J. Woodhouse character, necessitating a licensing agreement, and it looks like the Woodhouse people wanted a filter that kept their character from being associated with anything “naughty”. With Jeeves gone, the filter is going away, as you can already see on those searches listed above.
Cooler Place To Work: Google or Yahoo?
Valleywag is asking the tough questions: Which company, Google or Yahoo, is the cooler workplace? While Google certainly has the better accomodations, with all the free food and other fun stuff at the Googleplex, there are good arguments that Yahoo is still tops. Two good reasons: “All the millionaires are long gone” and “Googlers are so damn uppity”. I still prefer the Goog.
YouTube Gets NBC Deal
YouTube, needing to get legitimate before a lawsuit crushes them, has signed a deal with NBC to promote much of the peacock’s fall lineup. While YouTube will be airing promos and interviews, and NBC will be encouraging viewers to send videos to YouTube of their office (for a “The Office” contest), the real key to the deal is this: NBC will have closer access to YouTube’s database, and can troll through it and remove almost immediately any videos that infringe on NBC properties.
I can’t wait for the first time NBC removes a video that doesn’t violate their copyrights, but does hurt one of their interests in one way or another. No one thinks they’ll be erring on the side of caution, so some mistakes are guaranteed, and when they do, YouTube will just be sitting with egg on their faces. Enjoy!
Yet Another Other Other Other Other (repeat fourty times) Google Design Experiment
Google Blogoscoped has screenshots of a Google design experiment. This one has a vertical list of services on the left side of the search box on Google.com and at Google Images. I think it is becoming quite clear that the design guys at Google are pushing hardcore for vertical tabs, and will keep tweaking until they can get the cofounders to sign off on it. Expect this change to happen eventually.
Ask A Googler
A poster at the Something Awful forums claims to be a Google employee, and answers some questions from other posters (anything not too confidential). Philipp has a good rundown, and here are some highlights:
- He works on Google Calculator, which does calculations when you run Google searches
- “Google’s mostly C++, Java, and Python, or so I’m told”
- Google keeps a lot of logs of searches, but is really paranoid about them being abused. Good to hear.
- Don’t bother with the Mexican food or hot dog/burger bar in the Google cafeteria, unless you can’t find anything you like.
- Google is concerned with click fraud “like you wouldn’t believe”
- Getting hired by Google can take months. A referral from an employee can push you past a huge queue, and the employee gets a bonus if you get hired.
- Getting satellite maps into Google Maps was a dare, as in “I bet you can’t integrate the Google Earth data into Maps in a week!” Wow.
- “If your main project constantly has looming deadlines, it’s time to talk to your manager or your tech lead and tell them that they’re pushing too hard. The Google management respects this complaint, and knows that you can’t expect a programmer to push as hard as possible for more than a week at a time.”
- Googlers get one-terabyte Gmail accounts
- Lots of people bring their dogs to Google
- Googlers are quite liberal and anti-smoking
- Googlers get to choose between a Powerbook or a Thinkpad
- Currency conversion rates for Google Calculator come from the ad departments, since they need to do constantly accurate currency conversions anyway. Way to leverage resources!
- Previously, if you typed in “Where in the world is Carmen Sandiego” it would say she was in Cairo, but they removed it. Shame.
- Go to any Google Search Appliance, port 8000 (if they don’t block it), click “About“, then the sand dunes. Li’l Easter Egg.
Here’s some more stories, all lumped together in a single place:
Google Blog Search has an altered interface, with a sidebar containing a lot of the drilldown options (see it here). You can click several links to change the duration you are searching within and select RSS feeds of the current search. Considering most users never touch the advanced search page, this is a pretty decent idea, one Google might want to try on more properties. Google News already has a sidebar, so why not add it there?
Google Video also got a slight interface change to go with the new web-based uploading. Much cleaner.
GTalk Profile is a great site, where users of Google Talk submit a little info on themselves, including their location. All the users get plotted out on a Google Map, so you can get an idea of how many there are out there, and, once registered, contact them. Thus far, 1749 users have signed up.
(hat-tip from Jason)
Snap.com relaunched with a new interface that blurs the lines between organic and paid search listings. What is truly interesting is their anti-spam measure; John Battelle explains that they’ve licensed the clickstream data of some ISPs to remove spam, reasoning that spam websites don’t get a lot of visitors. Brilliant idea, with one big problem: As long as those spam sites do well on Google, they’ll get as many, if not more, visitors than a legitimate site.
Some SEOs did a Matt Cutts rap song. Damn, Matt is cooler as a geek than a rapper.
Use Google Image Ripper to get straight to the images in a Google Image search, skiping the thumbnails and interim pages. This violates so many lawsuits that Google will eventually shut it down, but for now, it’s been disabled due to bandwidth issues (thanks, Digg!)
In what could be terrible news for the sad few who rely on Google Images for porn, a court has ruled that some aspects of Image Search infringe on copyrights of porn site Perfect 10. Basically, the judge has ordered both sides to craft a narrow preliminary injunction that will likely prevent Google from showing image thumbnails of images stolen by other websites that are indexed in images, but that does not infringe on Google’s right to link to those images.
While the law is perfectly clear that Google is not responsible when sites in its index infringe on copyrights, I can understand how thumbnails of copyrighted images would be unlawful. Why doesn’t Google do the same thing it does when it gets DMCA requests against pages in search results that contain software that violates copyrights? Why can’t it just remove the images from the index and include a notice on the page?
From the AP:
Want an easy way to view regular Google results side-by-side with censored Google China results? The Google Censorship viewer is the answer to your problems. You get results from Google web search, Google News or Google Images, and every page produces a permalink, Google Maps style. Check it out, or go straight to results for:
Damn, that is one smart filter. Took out all the tanks, Google censorship and mass murder websites.
(via DeWitt Clinton > Findory)
Declan McCullagh & Danny Sullivan vent their thoughts. Quite interesting.
Declan writes in his original story:
China’s government has an extensive Internet filtering process in place that controls which overseas Web sites its citizens can access. (A 2005 study by the Open Net Initiative called it “quite thorough.”) With that filtering as a guide, foreign companies are expected to build their own lists of Web sites to delete from Chinese search listings.
To which Danny ads :
Got it? China will give the search engines some advice on censoring and guidelines, but it remains up to the search engines to do the actual dirty work.
If Google was going to cave in China and create evil Google, at the very least they could have pushed back to say they’d block specific sites given to them. Instead, they seem to be blocking a combination of known sites plus other sites that might have objectionable material, all based on what they decide themselves.
Read Declan’s first story and the update. Also read Danny’s view on SearchEngineWatch
Bill Gates, the billionaire founder of Microsoft, took the rare step of standing up for arch-rival Google today as he argued that state censorship was no reason for technology companies not to do business in China.
The richest man in the world told delegates at the World Economic Forum in Davos that he thought the internet “is contributing to Chinese political engagement” as “access to the outside world is preventing more censorship”.
All three of largest internet companies - Google, Microsoft and Yahoo - have been fiercely criticised by human rights groups for toeing China’s line on restrictions of free speech.
Read more on [TimesOnline] via [techzingo] >[Slashdot]
From the Google Blog:
Google users in China today struggle with a service that, to be blunt, isn’t very good. Google.com appears to be down around 10% of the time. Even when users can reach it, the website is slow, and sometimes produces results that when clicked on, stall out the user’s browser. Our Google News service is never available; Google Images is accessible only half the time. At Google we work hard to create a great experience for our users, and the level of service we’ve been able to provide in China is not something we’re proud of.
This problem could only be resolved by creating a local presence, and this week we did so, by launching Google.cn, our website for the People’s Republic of China. In order to do so, we have agreed to remove certain sensitive information from our search results. We know that many people are upset about this decision, and frankly, we understand their point of view. This wasn’t an easy choice, but in the end, we believe the course of action we’ve chosen will prove to be the right one.[…]
Filtering our search results clearly compromises our mission. Failing to offer Google search at all to a fifth of the world’s population, however, does so far more severely. Whether our critics agree with our decision or not, due to the severe quality problems faced by users trying to access Google.com from within China, this is precisely the choice we believe we faced. […]
No, we’re not going to offer some Google products, such as Gmail or Blogger, on Google.cn until we’re comfortable that we can do so in a manner that respects our users’ interests in the privacy of their personal communications. And yes, Chinese regulations will require us to remove some sensitive information from our search results. When we do so, we’ll disclose this to users, just as we already do in those rare instances where we alter results in order to comply with local laws in France, Germany and the U.S.
Read it all - hat tip: [Niraj]