Google Announces Page PreFetching (again) - beta

[engine]

Google just announced Instant Pages which works within Chrome (Beta) to load sites almost instantly. Looks like it pre-loads the top query using pre-rendering.


Related:
Banning Google FireFox PreFetching: [webmasterworld.com...]

[edited by: Brett_Tabke at 2:11 pm (utc) on Aug 3, 2011]

[MarvinH]

the website and the user each pay the bandwidth, so Goog are costing both of them money with this "predictive" service..Amit wouldn't smile so much if he was the one who'd be paying for this.

... and it's not good for the environment either ...

:-)

[netmeg]

Why bother going to your website?

Hey you know what, people still do.

Since I own or have control over plenty of #1 spots right now (and no BIG brands either, but just little ones), I choose to see this as a good thing; at least until I see how badly it screws up my various non-Google analytics and stats programs. Which it probably will, at least initially.

[incrediBILL]

Prefetch shoved down your throat whether you like it or not, analytics accuracy out the window along with your bandwidth and the functionality of your page since it starts playing behind the scenes.

OK, so the first page loads fast, what if the second doesn't? Fooled the surfer into thinking your site is screaming fast when it isn't.

Wonder how many little sites using hosts that limit bandwidth will end up displaying "Bandwidth Exceeded" like Google gives a rats ass if your site gets seen or not.

The only real issue I have with concept of preload is the idiot PhDs that originally shoved the function into browsers just randomly load crap instead of giving any control like "rel=preload" so we can actually make the process make sense for our sites.

Here's the real rub - Chrome OS is a wireless broadband/wifi-based OS, so preloading pages automatically will put more strain on already overloaded wifi sites like airports and other high volume wireless locations. If those Chrome laptops catch on they'll be prefetching the available bandwidth into oblivion.

[lucy24]

Idle query: Do they perform the pre-loading activity before they physically display the SERP page? Sure, it's a matter of milliseconds, but it's a nice extra subliminal: This site loads up even faster than google did! Especially, of course, if the user has taken the time to look at the whole top 10 before deciding that, Yeah, I guess no. 1 is the one I want. By then the site been raring to go for at least half a minute. (How long do people spend on the default 10-result page? I bet someone knows.)

[heisje]

Now that the "mad scientists" at Google have successfully resolved all search challenges, through such "innovations" like Panda, they are busy with the "fancy" stuff.

Heaven help us all . . . .

.

[koan]

they are busy with the "fancy" stuff.

This is a classical false argument. Just because one important thing is not complete, it doesn't mean that putting everyone on that project will help, and it doesn't mean various groups cannot work on various projects at the same time, and it doesn't mean other projects may not help solve other problems that are just as important for other people. The first important problem may not be solved for decades... we can't put everything else on hold in the meanwhile.

It's like saying "Why research for migraine drugs when we haven't cured cancer yet?"

[tedster]

Thank you, koan. This is an example of what I sometimes call "thinking of a company as if it were a person." A person can really do only one thing at a time, so how they allocate their present time is a very big deal. Corporations do not operate under the same kind of practical restrictions... they have others ;)

[Sgt_Kickaxe]

Where would Google be without our content?

Anyway, back on topic, how should we plan to optimize SEO to, you know, actually GET visitors from this new change?

[DirigoDev]

I’d like to see a calculation on the global cost for wasted bandwidth. What is this going to cost in terms of CPU cycles, cooling, power, etc?

If 1 out of 5 searches is predicted incorrectly, the cost is going to be gigantic. Or is it? Google has 1 billion searches a day. 6 billion preloads not used per month. 320kb/page (a Google number). 1,920TB of bandwidth wasted per month (someone check my math on this). Not so huge. This assumes a preload of a single page per query. Still, not so green and certainly bad for the planet. Does Google own any dark fiber? Any extension of the preloading would be a slick way to chew up lots of existing bandwidth.

[tedster]

Does Google own any dark fiber? Oh my, yes. They've been collecting it for years.

Google buying up Dark-Fibre [webmasterworld.com] - 2005

[dstiles]

Anyone know what IP ranges / UA google (plan to) use for this excercise? And does/will it include a proper X-Moz/preload header tag?

[tangor]

Throwing this into the mix... if Google has to say anything about it, then we might wish to be afraid...

Google has downplayed concerns that refinements to its search technology could leave surfers more exposed to search engine manipulation attacks.

Google Inside Search aims to speed up web searches by pre-loading content from remote sites. The so-called Instant Pages technology only works with Google Chrome.

[theregister.co.uk...]

Then we have this... does this suggest that Google will be adding yet more code for webmasters to wrestle to prevent these new "improvements"? (As if the average Joe or Sue Webmaster will do that... or even know what an API is for...)

Velocity Google has urged website developers to use Chrome's experimental Page Visibility API to reduce their sites' activities when they're not actually being viewed by browser users.

Now included with the developer version of Chrome – and due for arrival in the beta version next week – the PageVisability API allows websites to determine when they're actually being viewed by users – and when they're just sitting in the background. The API can tell you when a site is sitting inside a background tab, but also when a site has been pre-rendered by Chrome's new Instant Pages tool.

[theregister.co.uk...]

What I am reading from other sources it that Instant Pages only works with Chrome... and if that's the case, if one is really worried... it wouldn't take much to .htaccess Chrome ... if one is that concerned.

I'm not thrilled that G is continuing to futz with page code and server configs by bringing out all these "newfangled" things. At some point, if it ain't broke, don't break it!

[lucy24]

Possibly not even the right thread to ask, but this stuff makes my log-wrangling routines go haywire:

IP: {human}
Request: {all files except the page itself, including four images that were moved and 301'd 12 days ago}
Referer: /webcache.googleusercontent.com/search
?q=cache:{buncha letters & numbers):{name of my page}+{search string}
UA: {human}

Is this a cache or a search response or what? Why are they fetching brand-new copies of the peripheral files but not the actual page? (I checked. Their last pickup of the page itself was 4 days ago-- including the relocated images via up-to-date links, so why were they even looking for the old ones?)

[steerpikegg]

I think the article here [thedeceiver.net...] has a valid point and that it could be a way to mask their poor serps.

[tedster]

I think that article is a pile of juvenile crud. Can you imagine a world class search team saying: "Hey, our search results are really bad. Instead of improving, let's pour resources into building technology that lowers the time spent getting to those bad pages."

Sometimes people take Google bashing way too far.

[steerpikegg]

Hi Ted, I get your point about the Google bashing, but it did get me thinking about possible reasons behind the instant pages idea because the speed of pages loading hasn't ever been a factor of concern at least for us over here in Britain. Perhaps it's more relevent in other parts of the world.

[Robert Charlton]

...has a valid point and that it could be a way to mask their poor serps.

Addressing this point only... increasing speed has been a concern of Google's for many years. I don't know how anybody who's watched Google could miss that.

From Google's Technology Overview [google.com]...

Speed is a major search priority, which is why in general we don’t turn on new features if they will slow our services down. Instead, search engineers are always working not just on new features, but ways to make search even faster. In addition to smart coding, on the back end we’ve developed distributed computing systems around that globe that ensure you get fast response times. With technologies like autocomplete and Google Instant, we help you find the search terms and results you’re looking for before you’re even finished typing.

[true_INFP]

Chrome's experimental Page Visibility API

If our server accepts the pre-load request, will there be any way to determine whether the user actually clicked through the search-result link to the target page?

If not, we would have to deny all such requests (to prevent the distortion of our traffic stats).

We would also appreciate if any such pre-load APIs were cross-browser and cross-platform (a simple RFC might do the trick?).

[tedster]

Speed depends on connectivity. I ride the Acela train frequently between Boston and NYC. I don't use their public wi-fi, but rather use my own "mi-fi" card. During those trips, speed is an issue. If anyone is on the edge of decent mobile reception, speed is an issue. If anyone is on a satellite connection, speed is an issue.

And with the obvious explosion of mobile connectivity these days, an emphasis on speed makes plenty of sense to me.

to prevent the distortion of our traffic stats

That's a concern for me, too. I'm still looking for more clarity about analytics questions. At the same time, the traffic itself is even more important to me than it's measurement. I deposit dollars in the bank, not statistics.

[indyank]

Google is not just in U.S and there are still several countries where speed is an issue. In the country where some of the google engineers work, India, speed is a major issue.

I would associate this more with chrome than with the Google search engine as it is browser dependent and chrome has always focused on speed.

But what will irritate many is their total disregard for webmasters who don't want to opt into this.Google's treatment of webmasters is pathetic.They do provide an opt out for users but doesn't seem to provide one for the webmasters.This has been their attitude in several of their new changes or features.

Google is not just dependent on their search users but also on webmasters. But they seem to be totally ignoring one big community.

[Leosghost]

If anyone is on the edge of decent mobile reception, speed is an issue. If anyone is on a satellite connection

They'd better be on unlimited data plans..because #1 results with large amounts of data ( like slide shows or Flash pages ) are going to start burning a hole in their bank balance, before they know it..and more importantly..even if they don't click through..:(

Almost every mobile device ( phone, pad, tab,slab, laptop ) is on data caps with nasty charges for going over..

Chrome users on mobile are going to get some nasty surprise bills.

[tedster]

...they seem to be totally ignoring one big community

Different Google teams often get tunnel vision around various technical "enhancements." Remember the beginnings of AJAX search results? They even trashed Google Analytics!

I am confident that the issues around this change will get sorted - both bandwidth and challenges to the major Analytics vendors, too. Google uses a kind of rapid iteration model all the time. Roll out early and fix many of the bugs later. This kind of "agile" development is taking over the world, and it is hard for an old codger like me to adapt to sometimes.

[indyank]

Chrome users on mobile are going to get some nasty surprise bills.

does chrome work on mobile phones? I thought this feature is only for the desktop version.

[indyank]

This kind of "agile" development is taking over the world, and it is hard for an old codger like me to adapt to sometimes.

let it be the modern agile or the traditional waterfall development methodology. Can't they provide an opt out for webmasters when they can provide one for the end users?

[Leosghost]

Unless ( in the USA at least) Google start using that dark fibre and open free wi-fi and a GSM type service ( take on the Telcos and the ISPs ) and offer free internet and mobile services all over ..supported by ad money..

Which might explain why they bought all the dark fibre anyway..they could cut out the middle men..in the densely populated areas at least.

I remember 100 % ad supported internet here ( there are still services which offer 50% ad supported DSL to the home for domestic users )..and I have still one ( no contract ) mobile phone which gives me 50% of my calls free, if I accept to listen to a 20 second ad before the call is put through..( costs me about $5.oo per year to keep that "service" option ) ..other mobiles here are mostly unlimited and paid monthly or bundled with unlimited data via ISPs DSL services.

[indyank]

Leosghost, have you used chrome on a mobile phone? where did you get it?

[Leosghost]

I thought this feature is only for the desktop version.

Amit said in the presentation, if it works OK on desktop / fixed machines.. they'll roll it out to mobile very soon.

[arikgub]

One additional impact to many others mentioned here, is that Google is effectively "training" people to click on the 1st result. So those who aren't #1 will see a bit less traffic

[tedster]

WebProNews published an article about Instant Pages - with some official comment from a Google engineer.

Analytics and advertising solutions will have to be updated to take account of prerendering via the page visibility API. In most cases the end site owner shouldn’t have to make any modifications to his page; the 3rd party will simply make a minor change to the javascript that is pulled into publishers’ pages. You should check with your analytics or advertising providers to check if their scripts are prerendering-aware.

Google on How Instant Pages Affect Analytics [webpronews.com]

CPM advertising services better be all over this or the burn-through rate for advertiser money could get scary!

[tangor]

Yet another example of G changing the landscape and existing infrastructure of web to suit their ends... then again, if I were the bull in the pasture I might make my own rules, too. At present G has not only the best access to the sandpile, but think they have the pail and shovel, too. And they do if we let them.

Note that all these changes merely tighten the G leash on webmasters (desperate). After all, they created the "free" money (also a pit) webmasters lust after. Each time there's a new change, it costs webmasters more bandwidth or access and the G bottom line is enhanced. Don't get me wrong, as a capitalist I give them thumbs up. Yet, I'm liking the "other alternative" out there better each day. If only there was a realistic third party where we might party...