Amazon's Search Engine Goes Live

fjordboy writes "John Battelle has posted a discussion and review of Amazon's new search engine: a9.com. From the article:"What makes this particularly noteworthy is that A9 is built quite literally on top of Google. In short, Amazon has taken the best of Google, and made it, to my mind, a lot better. Sound familiar? Yup, it's what Google did to Yahoo, Yahoo to Netscape...you get the picture." "

Comments

Re:A9 Search Plugin for Firefox

[Patik]

Alternatively, make a bookmark for http://a9.com/%s , right-click on the bookmark to edit its properties, and give a keyword like "a" or "a9". Then just type "a searchterm" in the address bar to search.

Also note...

[ackthpt]

They redirect and try to trap you from backing out. How refreshing. One of the web page practices I most despise.

Re:Tinfoil Hat Not Required

[orthogonal]

You don't have to be among the tin-foil hat crowd to have a low regard for this "feature". There are just some searches that you *don't* want to remember.

You make a very good point.

And note that the Amazon page carefully says that you can "hide" an entry -- not that you can delete it.

But please remember that Google already logs your IP address and search terms; so presumably thus means that now both Google and Amazon will be keeping tabs on you.

a9 has a big problem

[cr@ckwhore]

Ok, so A9 is supposed to be "better" than google? Well, on my first search on a9 just a moment ago, I searched on "amazon sucks" (completely irrelevant to this post) ... and then pressed the "back" button on the search results page. Well, guess what?!?! A9 has a back-button-trap making the "back" button basically useless on their site.

So, they're supposed to be better than google? What about google's clean, simple, no BS web design approach? That's google's value!! Don't you get it Amazon?!?!? Well of course not! Take a direct look at Amazon.com and you'll get an idea of Amazon's design principles.

Re:Ads

[SphericalCrusher]

Google built on top of Yahoo? What? They both use two different ways to search the internet.

Yahoo with its meta tags and Google with its linkback system makes them two different search engines.

Re:Tinfoil Hat Not Required

[Minna Kirai]

As if you couldn't find history in Mozilla...

Mozilla history is stored on your local harddrive. You can easily delete it completely, or just disable the recording.

A visible history in the search engine will be something else- not only do you have no straightforward reliable way to delete the records, but the records could follow you around from home to job and back.

Re:Tinfoil Hat Not Required

[orthogonal]

Yeah, the fact that they don't provide direct links to the pages also throws my paranoia into overdrive

I was going to give you a Proxomitron script to convert these nosey links into real links; in doing do, I took a look at the page source for Amazon's search.

I found a few very interesting things:

var newloc = "/-/search/loadHist" + search;
openInHiddenIframe(newloc, document.getElementById("histContent"));

It look as if one's private search history is sent, in the clear (without SSL encryption) to a hidden frame. Good luck keeping it private if someone else administers the proxy server you use.

Consider that you log in to Amazon's search: will logging into Amazon search from work mean that the IT guys at work get to see the searches ("gay tentacle anime") you made at home?

Other goodies from the page source: function loadDeferredImages(). I don't know what this is, but is a deferred image anything like a pop-under? It uses NoSetTimeOut-->NoNoSetTimeOut-->NoNoSetTimeOut-->N oNoSetTimeOut-->NoNoSetTimeOut-->NoNoSetTimeOut--> NoNoSetTimeOut-->NoNoSetTimeOut to do the deferred load.

Now, maybe this is all benign, and it's just bad programming (apparently they're using JSPs) that inefficiently does work on the client using javascript. But I'd rather be skeptical now than find my search history being used to market to me.

Oh, the Proxomitron script to convert these links to straightforward links; note that it exempts Google and wikipedia.

Name = "Un-Prefix URLs (RK modified) and leave original too"
Active = TRUE
URL = "(^www.google.com|groups.google.com|*.wikipedia.or g)"
Bounds = "<a *>"
Limit = 256
Match = "<a (*href=)\0("|)\1(*(/|\?)*)\2(('|)http(%3A|:)(%2F|/ )+)\3([^&;=>"*]+)\4\5("|)>"
Replace = "<a \0\1\2\3\4\5\1><font size=1 color=red>[orig]</font></a>\r\n"
&nb s p; "<a $UESC(\0\1\3\4\1)>"

Re:Smells like consumer profiling engine...

[dalelane]

hmmm.... but have you seen generic.A9.com?

On generic.a9.com, we will not recognize your A9.com or Amazon.com cookie. Information we gather on generic.a9.com will not be used in our data analysis (other than to detect abuse) and will not be used to personalize the services we offer you.

surely they deserve bonus points for that....

(plus the colour scheme aint so bad on generic.a9.com - reason enough to prefer it to regular a9! :) )

Re:People who searched for "warez" also read...

[bobbis.u]

It doesn't display any results, but clearly it does find them. Look at the bottom and you can still jump to all the blank pages that would contain results (warez "returns" 11 pages).

Nice bug... erm, I mean "feature".

Re:People who searched for "warez" also read...

[LostCluster]

Nah, that's simply what Google returns if you search for "warez" with SafeSearch on.