I feel a bit bad about
picking on Google
Doctype’s HTML on launch day. Having
launched products at Google I know there’s nothing more obnoxious than
some know-it-all harping on some tiny problem with the product you’ve
worked for a year to launch. But the irony of the problem was too much
to let pass. It’s fixed now, a plain ol’ HTML link. Yay!
Since I picked on Google’s newest product let me pick next on their
oldest product now; web search. And the ugly URLs it produces.
The first problem is the search page URLs are too big.
If you go to the plain ol’ google.com home page and search for “RFC
1738” in Firefox 3 or MSIE 7 you end up at here:
http://www.google.com/search?
hl=en&
q=rfc+1738&
btnG=Google+Search
The
q parameter is our search term, a lovely thing that belongs there.
But what’s the other stuff?
hl specifies the language Google
itself is in. Google already knows I speak English, why does it have to be
in the URL? And the
btnG is entirely superfluous. Why not
just send me
here?
http://www.google.com/search?q=rfc+1738
So much simpler. Easier to email to other folks, cleaner to bookmark,
cleaner to link to.
It gets worse. Here’s the URL for the first search result:
http://www.google.com/url?
sa=t&
ct=res&
cd=1&
url=http%3A%2F%2Fwww.faqs.org%2Frfcs%2Frfc1738.html&
ei=6KYtSM33GYb6pgTpib3TAQ&
usg=AFQjCNFV9ot1XoJzOO28ndOgjYRgWbbaVw&
sig2=D2Q73werQu-BZTkopfncgA
Follow all that? Three completely mysterious parameters, an escaped
version of the actual URL, and then
three opaque blobs.
What’s
all this garbage for? Why not just have a link to
http://www.faqs.org/rfcs/rfc1738.html
Because of all the URL mucking I can’t really use the URL on the
Google results page. I can’t simply copy the link destination or
bookmark it; I have to click and go to the page itself. And whatever
they’re doing means I don’t even go to the site when I click the link;
I have to make an extra HTTP connection to google to get a redirect
(250ms in the test I just ran).
The destination URLs only get rewritten in this way if I’m logged into
my Google Account; with no google.com cookies you get clean URLs. I
imagine there’s some sort of clickthrough tracking or web search
history maintenance or some other vaguely privacy-invading feature
I’ve long since forgotten about. There’s ways to do clickthrough
tracking without redirecting URLs; why mess with links?
I realize writing this I’m dangerously close to self-parody by
dissecting URL structures. Very few people will care, stuff still
works when you click on it. But Google set a real standard for
parsimony and speed in its web applications. It also has a unique
understanding of the value of a simple HTML link. So it bugs me when
they break stuff like that.
PS: to my friends at Google, no need to send me mail explaining why
all this stuff is necessary. I still vaguely remember the explanations
for most of it. My point is that from the outside, as a user who cares
about simple HTML, all this extra stuff looks like harmful overhead.