The browser does download subresources, including images, scripts, and stylesheets. The new page visibility API is intended to help folks keep accurate web stats:
So yes, this will screw up web stats. A new JS API won't help with anything server side, and there's going to be a significant lag before every web app in the world is updated. Think about preloading anything that keeps track of read/unread status as well (e-mail, forum threads, etc).
Actually, it is rendering it, in a hidden tab. As far as the site will know, you visited. Last login times get updated, web stats record hits, stuff gets marked as read, etc.
So scripts execute too. Hm, that is a bit surprising. Well, I trust Chrome so much that I don't block scripts and I go to any site that pops up as a google search without any hesitation so I suppose I don't mind.
I could care little about privacy concern, but I would rather wait a couple of seconds than waste bandwidth downloading pages that I would most likely never look at. Also from web developer point of view, I would like to see how any given site loads under standard use case anyway.
Google is only preloading the top result when it is confident the top result is what you are looking for. If Google does a good enough job then only a small % of preloads will go unviewed.
They're offering a feature that people will like but nothing prevents me from going elsewhere. I don't see how this increases my dependence on them. Relying on them to control all of my email access, storage, etc? Yes, that makes me dependent on them. Pages loading faster? Meh.
And besides, this is a Chrome browser feature. Google would extend Bing the same curtosey, and even if not, Bing could make an extension or do the same with IE9, etc, etc.
They're releasing the source. From the third paragraph of the second page:
"Other Web browsers could also employ Instant Pages, because Google has released the necessary code for all to use. "We are opening up the code because we want other browsers to implement it—it is good for the users and for the Web," said Singhal."
Even if this weren't the case, as drivebyacct2 said, this doesn't do anything to make us dependent on them, especially when compared to gmail or google docs.
Remember, this makes it vitally important that your "delete" links in web apps aren't actually links, but forms. Load up a data grid and boom! All you resources are gone.
The <link rel="prerender"/> will be on Google's search result pages instructing Chrome to prerender the most relevant search result. The control is in Google's hands, not the site owner, by default.
Right, but beyond that it's not going to go crazy prefetching every URL in your web apps as the user browses around triggering delete or other actions, unless you also put that <link> in?
I'd be scared of it automatically downloading inappropriate or even worse illegal stuff from the search results. Try explaining to the boss you did not visit that nsfw site but it merely turned up in your search results.
This is a good point. It's not just the "used in a court of law" or "national security" instances where this is dangerous. Corporations don't have to follow any formal legal process. If you're using a proxy or they're otherwise monitoring outbound traffic and you have this feature turned on, it will look like you are visiting inappropriate sites.
Seems like this would completely destroy any chance of using web server visit history for legal prosecution, since even visiting a site that links to another site could cause you to appear in the second site's logs. Perhaps this is a good thing.
"On November 26, 2008, officials released 700 pages of documents related to the Anthony investigation, which included evidence of Google searches of the terms "neck breaking," "how to make chloroform," and "death" on Casey Anthony's home computer."
The searches themselves, already provided by Google, combined with actually continuing to some of the results, also logged by Google, are already incriminating enough. The only thing you could claim is that ISP traffic logs couldn't and wouldn't make a difference between you and Chrome loading pages, but that didn't even come into play in the case you quote.
As per recent news (amongst other things), links/linking are in the process of being criminalized in some contexts. Do you want to hit a page containing a link and become "guilty" for using/following that link, without ever clicking on it?
It may seem far-fetched at the moment, but I'm reminded of that East Coast teacher who was fired because the classroom computer was infected (apparently, not by her actions) and started popping porn up on the screen.
Common sense quickly demonstrated that she was not, in any personal sense, at fault. Still lost her job; was, IIRC correctly, at least considered for criminal prosecution, and had to spend months and probably lots of lawyer fees attempting to clear her name.
I am mindful to be very cautious about assuming a limit to the stupidity or maliciousness of people who will be making assumptions -- or, regardless of their own understanding of a situation, simply have an agenda to push -- in the future.
If, as is hinted, they restrict this to SERPs that carry one canonical result, such as 'facebook login' or -cough- 'html5 spec', it could be a great thing. If after some interval you do the same search and again click the top result, Chrome might even start redirecting you there straight from the omnibox. But if they apply this technique too broadly it's going to be a big drain on server resources and sites would probably start using visibilityState* to lazy load their stuff.
Wouldn't this be a crazy bandwidth hog? If I typed in multiple search terms, and didn't load any of my results, it would have automatically downloaded everything for no reason.
In Google's own words they only preload the top result if they are confident it is what you are looking for. If they pull it off there will be negligible amount of unused preloaded pages.
Shouldn't the top result inherently represent what Google thinks is what I want? Or is there some secondary "confidence" component to the presumed efficacy of a top result?
I'd presume they could look at the difference between their confidence (relevance?) in the first and second results. If the difference is more than some threshold, bam preload the first.
I currently disable DNS prefetching and "using web services to improve my browsing experience". This is yet another feature that will go on my black list.
First and foremost, I think they are solving a problem nobody has, potentially for "my Chrome is faster than your Firefox" points. I do not mind waiting an extra 300ms for having a modicum of control over what my browser does. I don't know about others, but page loading is not usually something I even notice anymore.
Secondly, fully rendering sites I never visit, including script execution, seems like a seriously bad idea. This turns a mildly annoying while slightly beneficial feature (loading resources in the background) into a monster, especially when combined with abominations like ro.me on autoplay.
Thirdly, I could imagine scenarios in which my Google searches would be leaking this way to sites that have no business seeing my traffic.
I realize that what they are currently talking about is opt-in, but can only hope that in time this feature won't be picked up by the other browsers and become the silent default for everyone.
I do not mind waiting an extra 300ms for having a modicum of control over what my browser does.
Your point is still legitimate, but the 1,100 ms (not 300 ms) in that photo which you refer to is the time to serve the search query results, not time to load the page you then click on, which usually takes a number of seconds more.
Point taken, I was going for dramatic effect there.
Rendering the page, after selecting the result, does take a lot longer (6s for something like the BBC home page here) but I still maintain that it's largely invisible and not something most users would complain about or even notice.
You say you don't mind 300ms in exchange for control over what your browser does. Wearing my technology and privacy conscious security USER hat, I agree completely.
As a webmaster for a site where retention and "stickiness" are critical, I can say without hesitation that those 300ms are potentially HUGE. For the vast majority of the users coming into a site who are unaware of these things, faster page load times make a huge difference in retention and repeat visits. People unconsciously like pages that appear faster. Ask them WHY they like one site over the other, they probably won't say it's the speed, but the stats say otherwise.
This is really bad news for website owners... We'll see increased (sometimes by order of magnitude) number of page views and bandwidth usage, without increase in the number of real visitors.
Same thing happened with DNS when Google enabled DNS pre-fetching in Chrome.
Seems like no major disaster came out of the dns thing. Anyway, this could be a benefit for many webmasters. Google has demonstrated in the past that faster load times mean more users stick with a result. Plus, I'd be really surprised if there were no way to disable it if it is really causing a problem.
I first thought that this was something only implemented in Chrome for Google, but I then saw that this is somehow a HTML feature that they will also add to Google Search.
I don't like if Google uses its own browser to give their sites features, that could not be implemented without - website's features should only be made from source code on the server. In my opinion.
This is core innovation from Google - but I see the problem with incorrect website visitors. It think Chrome when prefetching the pages, should send some information with the request that the request is a prefetch.
You haven't read the rest of the comments here, including the links to the specs for this feature. The entire page is loaded in a hidden tab and all scripts are executed.
isn't it funny that someone has to always comment "X browser had it for a while..."? well, if X browser had it, it didn't care enough to let the populace know about it.
That's fine and dandy until you search for something and the results return something in there randomly and is considered a threat to national security or of interest for you to be watched. Then your browser starts downloading a page you never intend to visit and you get logged as interested in terrorist activity. This is not good news.
I would think the security concerns would be front and foremost. When I hit up google for a list of sites, I want a list of sites. Whether my browser visits them or not is up to me - I don't want it happening automatically.
Now, personally, I can turn this off, I know, and I also understand the speed advantage for end user retention. People might say 300ms doesn't matter to them - maybe not consciously - but it certainly DOES matter when it comes to retention. People like faster sites automatically - it's a subconscious thing - very well proven by good studies. (amazon, etc)
a 300ms savings on page load is a HUGE win from the site owner point of view.
Give the site owners some kind of passive control over this feature (a-la robots.txt for spiders) so those who are really concerned about bandwidth can opt out (though if you are top ranking for a google search term you probably WANT that traffic - that's money down the drain otherwise) and appropriate controls, and it's a good idea.
In fact, it's an inevitable idea. with it's ups and downs. It's going to happen - so let's roll with it.
61 comments
[ 2.4 ms ] story [ 131 ms ] threadThe browser does download subresources, including images, scripts, and stylesheets. The new page visibility API is intended to help folks keep accurate web stats:
http://w3c-test.org/webperf/specs/PageVisibility/
And besides, this is a Chrome browser feature. Google would extend Bing the same curtosey, and even if not, Bing could make an extension or do the same with IE9, etc, etc.
"Other Web browsers could also employ Instant Pages, because Google has released the necessary code for all to use. "We are opening up the code because we want other browsers to implement it—it is good for the users and for the Web," said Singhal."
Even if this weren't the case, as drivebyacct2 said, this doesn't do anything to make us dependent on them, especially when compared to gmail or google docs.
http://en.wikipedia.org/wiki/Death_of_Caylee_Anthony
It doesn't seem too out of left field that investigators would use your browser history as well.
It may seem far-fetched at the moment, but I'm reminded of that East Coast teacher who was fired because the classroom computer was infected (apparently, not by her actions) and started popping porn up on the screen.
Common sense quickly demonstrated that she was not, in any personal sense, at fault. Still lost her job; was, IIRC correctly, at least considered for criminal prosecution, and had to spend months and probably lots of lawyer fees attempting to clear her name.
I am mindful to be very cautious about assuming a limit to the stupidity or maliciousness of people who will be making assumptions -- or, regardless of their own understanding of a situation, simply have an agenda to push -- in the future.
* Thanks for that link, abarth
First and foremost, I think they are solving a problem nobody has, potentially for "my Chrome is faster than your Firefox" points. I do not mind waiting an extra 300ms for having a modicum of control over what my browser does. I don't know about others, but page loading is not usually something I even notice anymore.
Secondly, fully rendering sites I never visit, including script execution, seems like a seriously bad idea. This turns a mildly annoying while slightly beneficial feature (loading resources in the background) into a monster, especially when combined with abominations like ro.me on autoplay.
Thirdly, I could imagine scenarios in which my Google searches would be leaking this way to sites that have no business seeing my traffic.
I realize that what they are currently talking about is opt-in, but can only hope that in time this feature won't be picked up by the other browsers and become the silent default for everyone.
Your point is still legitimate, but the 1,100 ms (not 300 ms) in that photo which you refer to is the time to serve the search query results, not time to load the page you then click on, which usually takes a number of seconds more.
Rendering the page, after selecting the result, does take a lot longer (6s for something like the BBC home page here) but I still maintain that it's largely invisible and not something most users would complain about or even notice.
It's when it gets taken away again that you notice.
As a webmaster for a site where retention and "stickiness" are critical, I can say without hesitation that those 300ms are potentially HUGE. For the vast majority of the users coming into a site who are unaware of these things, faster page load times make a huge difference in retention and repeat visits. People unconsciously like pages that appear faster. Ask them WHY they like one site over the other, they probably won't say it's the speed, but the stats say otherwise.
Same thing happened with DNS when Google enabled DNS pre-fetching in Chrome.
Good example / explanation of the problem:
http://www.pinkbike.com/news/DNS-Prefetching-Implications.ht... (cached: http://webcache.googleusercontent.com/search?q=cache:-3Lip30...)
Google's prefetching is instant search list with the most likely result being automatically be loaded in the background before the user clicks on it.
Web owner nightmare: Having a common irrelevant keyword that means most of the world downloads the content without viewing the page.
I don't like if Google uses its own browser to give their sites features, that could not be implemented without - website's features should only be made from source code on the server. In my opinion.
This is core innovation from Google - but I see the problem with incorrect website visitors. It think Chrome when prefetching the pages, should send some information with the request that the request is a prefetch.
There will probably be a switch, but non-techs usually don't flip switches
Now, personally, I can turn this off, I know, and I also understand the speed advantage for end user retention. People might say 300ms doesn't matter to them - maybe not consciously - but it certainly DOES matter when it comes to retention. People like faster sites automatically - it's a subconscious thing - very well proven by good studies. (amazon, etc) a 300ms savings on page load is a HUGE win from the site owner point of view.
Give the site owners some kind of passive control over this feature (a-la robots.txt for spiders) so those who are really concerned about bandwidth can opt out (though if you are top ranking for a google search term you probably WANT that traffic - that's money down the drain otherwise) and appropriate controls, and it's a good idea.
In fact, it's an inevitable idea. with it's ups and downs. It's going to happen - so let's roll with it.
When the Chrome development team decides something should be a certain way, they are sometimes not at all accommodating of differing viewpoints.